IPS Working Group M. Rajagopal, R. Bhagwat, R. A. Helland, INTERNET-DRAFT LightSand Comm. E. Rodriguez, Lucent Tech. (Expires December, 2001) C. Carlson, QLogic Category: standards-track D. Fraser, Compaq D. Peterson, Cisco L. Lamers, SAN Valley V. Chau, G. Hecht, Gadzoox Networks S. Wilson, B. Snively, R. Weber, Brocade Comm. M. O'Donnell, A. Rijhsinghani, McDATA S. Rupanagunta, Aarohi Comm. V. Rangan, Rhapsody Networks J. Nelson, K. Hirata, Vixel M. Merhar, Pirus Networks N. Wanamaker, Akara Fibre Channel Over TCP/IP (FCIP) Status of this Memo This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC 2026 [1]. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as Reference material or to cite them other than as "work in progress". The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/lid-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html Abstract Fibre Channel Over TCP/IP (FCIP) describes mechanisms that allow the interconnection of islands of Fibre Channel storage area networks over IP-based networks to form a unified storage area network in a single Fibre Channel fabric. FCIP relies on IP-based network services to provide the connectivity between the storage area network islands over local area networks, metropolitan area networks, or wide area networks. Rajagopal, et al. Standards Track [Page 1] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 Conventions used in this document The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [2]. Table Of Contents 1. Purpose, Motivation and Objectives . . . . . . . . . . . . . . . 3 2. Relationship to Fibre Channel Standards . . . . . . . . . . . . 4 2.1 Relevant Fibre Channel Standards . . . . . . . . . . . . . . . 4 2.2 This Specification and Fibre Channel Standards . . . . . . . . 4 3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . 5 4. Open Issues . . . . . . . . . . . . . . . . . . . . . . . . . . 6 5. Protocol Summary . . . . . . . . . . . . . . . . . . . . . . . . 7 6. The FCIP Model . . . . . . . . . . . . . . . . . . . . . . . . . 9 6.1 FCIP Protocol Model . . . . . . . . . . . . . . . . . . . . . . 9 6.2 FCIP Link . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 6.3 FC Entity . . . . . . . . . . . . . . . . . . . . . . . . . . 10 6.4 FCIP Entity . . . . . . . . . . . . . . . . . . . . . . . . . 11 6.5 FCIP Link Endpoint (FCIP_LEP) . . . . . . . . . . . . . . . . 12 6.6 FCIP Data Engine (FCIP_DE) . . . . . . . . . . . . . . . . . . 13 6.6.1 FCIP Encapsulation of FC Frames . . . . . . . . . . . . . . 15 6.6.2 FCIP Data Engine Error Detection and Recover . . . . . . . . 16 6.6.2.1 TCP Assistance With Error Detection and Recovery . . . . . 16 6.6.2.2 Errors in FCIP Headers and Discarding FCIP Frames . . . . 16 6.6.2.3 IP Network Transit Time Validation . . . . . . . . . . . . 17 6.6.2.4 Synchronization Failures . . . . . . . . . . . . . . . . . 17 7. TCP Connection Management . . . . . . . . . . . . . . . . . . . 18 7.1 TCP Connection Establishment . . . . . . . . . . . . . . . . . 18 7.1.1 Creating a New TCP Connection . . . . . . . . . . . . . . . 18 7.1.2 Processing TCP Connect Requests . . . . . . . . . . . . . . 19 7.2 TCP Connection Parameters . . . . . . . . . . . . . . . . . . 19 7.2.1 TCP Selective Acknowledgement Option . . . . . . . . . . . . 20 7.2.2 TCP Window Scale Option . . . . . . . . . . . . . . . . . . 20 7.2.3 IP DSCP Option . . . . . . . . . . . . . . . . . . . . . . . 20 7.2.4 Protection against sequence number wrap . . . . . . . . . . 20 7.2.5 TCP No Delay Option . . . . . . . . . . . . . . . . . . . . 20 7.2.6 TCP Acknowledgement Timeout . . . . . . . . . . . . . . . . 20 7.3 TCP Connection Considerations . . . . . . . . . . . . . . . . 20 7.4 Flow Control Mapping between TCP and FC . . . . . . . . . . . 21 8. Security . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 8.1 Considerations . . . . . . . . . . . . . . . . . . . . . . . . 22 8.2 IP Network Security Requirements . . . . . . . . . . . . . . . 22 8.3 Integrated Security . . . . . . . . . . . . . . . . . . . . . 23 8.4 External Security Gateway . . . . . . . . . . . . . . . . . . 24 8.5 Security Information Exchanged Between FC and FCIP Entities . 24 9. Performance . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Rajagopal, et al. Standards Track [Page 2] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 9.1 Considerations . . . . . . . . . . . . . . . . . . . . . . . . 24 9.2 QoS Support . . . . . . . . . . . . . . . . . . . . . . . . . 25 9.3 QoS Information Exchanged Between FC and FCIP Entities . . . . 26 10. Dynamic Discovery of Participating FCIP Entities . . . . . . . 26 10.1 Requirements . . . . . . . . . . . . . . . . . . . . . . . . 26 10.2 Discovery Information Exchanged Between FC and FCIP Entities 26 11. References . . . . . . . . . . . . . . . . . . . . . . . . . . 27 12. Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . 28 13. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 28 14. Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 29 15. Full Copyright Statement . . . . . . . . . . . . . . . . . . . 31 Annex A Example of synchronization recovery algorithm . . . . . . . . . 31 B Relationship between FCIP and IP over FC (IPFC) . . . . . . . . 36 C FC Frame Format . . . . . . . . . . . . . . . . . . . . . . . . 36 D FCIP Requirements on an FC Entity . . . . . . . . . . . . . . . 38 E FC-BB-2 Inputs . . . . . . . . . . . . . . . . . . . . . . . . 39 1. Purpose, Motivation and Objectives Fibre Channel (FC) is a gigabit speed networking technology primarily used to implement Storage Area Networks (SANs). See section 2 for information about how Fibre Channel is standardized and the relationship of this specification to Fibre Channel standards. This specification describes mechanisms that allow the interconnection of islands of Fibre Channel SANs over IP Networks to form a unified SAN in a single Fibre Channel fabric. The motivation behind defining these interconnection mechanisms is a desire to connect physically remote FC sites allowing remote disk access, tape backup, and live mirroring. Fibre Channel standards have chosen nominal distances between switch elements that are less than the distances available in an IP Network. Since Fibre Channel and IP Networking technologies are compatible, it is logical to turn IP Networking for extending the allowable distances between Fibre Channel switch elements. The fundamental assumption made in this specification is that the Fibre Channel traffic is carried over the IP Network in such a manner that the Fibre Channel Fabric and all Fibre Channel devices on the Fabric are unaware of the presence of the IP Network. This means that the FC datagrams MUST be delivered in such time as to comply with existing Fibre Channel specifications. The FC traffic MAY span LANs, MANs and WANs, so long as this fundamental assumption is adhered to. Rajagopal, et al. Standards Track [Page 3] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 The objectives of this document are to: 1) specify the encapsulation and mapping of Fibre Channel (FC) frames employing FC Frame Encapsulation [23]. 2) apply the mechanism described in 1) to an FC Fabric using an IP network as an interconnect for two or more islands in an FC Fabric. 3) address any FC concerns arising from tunneling FC traffic over an IP-based network, including security, data integrity (loss), congestion, and performance. This will be accomplished by utilizing the existing IETF-specified suite of protocols. 4) be compatible with the referenced FC standards. While new work may be undertaken in T11 [8] to optimize and enhance FC Fabrics, this specification requires conformance only to the referenced FC standards. 5) be compatible with all applicable IETF standards so that the IP Network used to extend an FC Fabric can be used concurrently for other reasonable purposes. 2. Relationship to Fibre Channel Standards 2.1 Relevant Fibre Channel Standards FC is standardized under American National Standard for Information Systems of the National Committee for Information Technology Standards (ANSI-NCITS) in its T11 technical committee. T11 has specified a number of documents describing FC protocols, operations, and services. T11 documents of interest to readers of this specification include: - FC-BB - Fibre Channel Backbone [3] - FC-BB-2 - Fibre Channel Backbone -2 [4] - FC-SW-2 - Fibre Channel Switch Fabric -2 [5] - FC-FS - Fibre Channel Framing and Signaling [6] - FC-GS-3 - Fibre Channel Generic Services -3 (FC-GS-3) [7] Additional information regarding T11 activities is available on the committee's web site [8]. 2.2 This Specification and Fibre Channel Standards Building a high performance device that successfully extends a FC Fabric over an IP Network requires tight integration of FC and TCP/ IP technologies. Since these two technologies are standardized by Rajagopal, et al. Standards Track [Page 4] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 multiple organizations, specifying all the requirements in one document and getting high quality review of that document has proven to be impossible. Therefore, this specification addresses only the requirements necessary to properly utilize an IP Network as a conduit for an FC Fabric. The result is a specification for an FCIP Entity (see section 6.4). A product that tunnels an FC Fabric through an IP Network must combine the FCIP Entity with an FC Entity (see section 6.3) using an implementation specific interface. Although the requirements placed on an FC Entity by this specification are listed in annex D, the list here is not exhaustive. More information about FC Entities can be found in the Fibre Channel standards and an example of an FC Entity can be found in FC-BB-2 [4]. No attempt is being made to define the interface between an FCIP Entity and an FC Entity at this time because doing so risks compromising the performance and efficacy of the resulting products. Current experience in this area is simply insufficient to guide definition of the interface appropriately. The objectives and motivations of this specification are not impacted by the decision not to standardize the interface between FCIP Entities and FC Entities because fully functional and compliant products can be built provided they contain both an FCIP Entity and an FC Entity. The only products that cannot be built are those that contain only one or the other and there is no urgent need for such products at this time. 3. Terminology Terms needed to clarify the concepts presented in FCIP are defined here. FC End Node - A FC device that uses the connection services provided by the FC Fabric. FC Entity - The Fibre Channel specific element that combines with an FCIP Entity to form an interface between an FC Fabric and an IP Network (see section 6.3). FC Fabric - An entity that interconnects various Nx_Ports (see [6]) attached to it, and is capable of routing frames using only the destination ID information in a frame header (see annex C). Rajagopal, et al. Standards Track [Page 5] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 FC Frame - The basic unit of Fibre Channel data transfer (see annex C). FC Receiver Portal - The access point through which an FC Frame enters an FCIP Data Engine from the FC Entity. FC Transmitter Portal - The access point through which a reconstituted FC frame leaves an FCIP Data Engine to the FC Entity. FCIP Data Engine (FCIP_DE) - The component of an FCIP Entity that handles FC Frame encapsulation, de-encapsulation, and transmission through a single TCP connection (see section 6.6). FCIP Entity - The principle FCIP interface point to the IP Network (see section 6.4). FCIP Link - One or more TCP connections that connect one FCIP_LEP to another (see section 6.2). FCIP Link Endpoint (FCIP_LEP) - The component of an FCIP Entity that handles FC Frame transmission through a single FCIP Link (see section 6.5). Encapsulated Frame Receiver Portal - The TCP access point through which an FCIP encapsulated frame is received from the IP Network by an FCIP Data Engine. Encapsulated Frame Transmitter Portal - The TCP access point through which an FCIP encapsulated frame is transmitted to the IP Network by an FCIP Data Engine. 4. Open Issues This draft is a work in progress and this section identifies areas where the work is known to be incomplete and discusses the current status of these efforts. This section will be removed before this draft is considered for standardization. - FCIP Entity Discovery - The basic principles of FCIP Entity discovery are agreed and represented in section 5. Work on the details of dynamic FCIP Entity discovery are incomplete (see section 10). Work on FCIP Entity Discovery may change the way an FCIP Entity is identified from the currently specified IP Address usage. - Security - In general, FCIP will follow or subset the security mechanisms agreed for iSCSI. The basic principles of FCIP security requirements are agreed and described in section 5. Rajagopal, et al. Standards Track [Page 6] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 Section 8 contains the latest information on the details of FCIP security. It must be noted that the association between IP Addresses and FCIP Entities is open to changes based on yet to be finalized decisions about security. The point at which a TCP connection is authorized to carry data is still being debated. - Timeout Coordination with Fibre Channel - In this revision, the only timeout Fiber Channel timeout consideration enforced by the FCIP Entity is R_A_TOV. All other timeout issues are the responsibility of the FC Entity. Section 7.2.6 contains a discussion of the TCP Acknowledge Timeout that needs to be reviewed to determine if it is still needed. - Performance - The discussion of performance considerations in section 9 and particularly the quality of service discussion in section 9.2 are known to require additional work. A particular concern is that quality of service not be limited to using diffserv. 5. Protocol Summary The FCIP protocol is summarized as follows: 1) The primary function of an FCIP Entity is forwarding FC frames, employing FC Frame Encapsulation described in [23]. 2) Viewed from the IP Network perspective, all FCIP Entities are peers and communicate using TCP/IP. Each FCIP Entity is a TCP endpoint in the IP-based network. 3) Viewed from the FC Fabric perspective, each pair of FCIP Entities, in combination with their associated FC Entities, serves as a frame transmission component of the FC Fabric. The FC End Nodes are unaware of the existence of the FCIP Link. 4) FC Primitive Signals, Primitive Sequences, and Class 1 FC Frames are not transmitted across an FCIP Link because they cannot be encoded using FC Frame Encapsulation [23]. 5) The path (route) taken by an encapsulated FC Frame follows the normal routing procedures of the IP Network. 6) An FCIP Entity SHALL have exactly one IP Address. 7) An FCIP Entity may contain multiple FCIP Link Endpoints, but each FCIP Link Endpoint (FCIP_LEP) communicates with exactly one other FCIP_LEP, possibly with multiple FCIP Data Engines. (FCIP_DEs) and multiple TCP connections. Rajagopal, et al. Standards Track [Page 7] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 8) When multiple FCIP_LEPs or multiple FCIP_DEs are in use, selection of which FCIP_DE to use for encapsulating and transmitting a given FC Frame outside the scope of this document. FCIP Entities do not actively participate in FC Frame routing. 9) The FCIP Control & Services function MAY use TCP/IP quality of service features (see section 9.2) to support Fibre Channel capabilities. 10) Each FCIP Entity is statically or dynamically configured with a list of IP addresses and port numbers corresponding to participating FCIP Entities. If dynamic discovery of participating FCIP Entities is supported, the function SHALL be performed using the Service Location Protocol (SLPv2) [21]. It is outside the scope of this specification to describe any static configuration method for participating FCIP Entity discovery. Refer to section 10 for a detailed description of dynamic discovery of participating FCIP Entities using SLPv2. 11) FCIP Entities do not actively participate in the discovery of FC source and destination identifiers. Discovery of FC addresses (accessible via the FCIP Entity) is provided by techniques and protocols within the FC architecture as described in FC-FS [6], FC-SW-2 [5], and FC-GS-3 [7]. 12) To support IP Network security, FCIP Entities MUST: a) implement cryptographically protected authentication and cryptographic data integrity keyed to the authentication process, or b) be capable of operating with external IP security mechanisms that provide cryptographically protected authentication and cryptographic data integrity keyed to the authentication process. FCIP entities MAY implement data privacy security features. Security features and requirements are detailed in section 8. 13) FCIP relies on TCP to recover from re-ordering in the IP network. 14) FCIP relies on both TCP and FC error recovery mechanisms to detect and recover from data loss and corruption within the IP Network. Rajagopal, et al. Standards Track [Page 8] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 6. The FCIP Model 6.1 FCIP Protocol Model The relationship between FCIP and other protocols is illustrated in figure 1. +------------------------+ FCIP Link +------------------------+ | FCIP |===========| FCIP | +--------+------+--------+ +--------+------+--------+ | FC-2 | | TCP | | TCP | | FC-2 | +--------+ +--------+ +--------+ +--------+ | FC-1 | | IP | | IP | | FC-1 | +--------+ +--------+ +--------+ +--------+ | FC-0 | | LINK | | LINK | | FC-0 | +--------+ +--------+ +--------+ +--------+ | | PHY | | PHY | | | +--------+ +--------+ | | | | | | | | | V +--------------------+ V to Fibre to Fibre Channel Channel Environment Environment Fig. 1 FCIP Protocol Stack Model Note that the objective of the FCIP Protocol is creation and maintenance of one or more FCIP Links. 6.2 FCIP Link The FCIP Link is the basic unit of service provided by the FCIP Protocol to a FC Fabric. As shown in figure 2, an FCIP Link connects two portions of an FC Fabric using an IP Network as a transport to form a single FC Fabric. /\/\/\/\/\/\ /\/\/\/\/\/\ /\/\/\/\/\/\ \ FC / FCIP \ IP / Link \ FC / / Fabric \=========/ Network \=========/ Fabric \ \/\/\/\/\/\/ \/\/\/\/\/\/ \/\/\/\/\/\/ Fig. 2 FCIP Link Model At the points where the ends of the FCIP Link meet portions of the FC Fabric, an FCIP Entity (see section 6.4) combines with an FC Entity as described in section 6.3 to serve as the interface between FC and IP. Rajagopal, et al. Standards Track [Page 9] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 An FCIP Link SHALL contain at least one TCP connection and MAY contain more than one TCP connection. The endpoints of a single TCP connection are FCIP Data Engines (see section 6.6). The endpoints of a single FCIP Link are FCIP Link Endpoints (see section 6.5). 6.3 FC Entity A product that tunnels an FC Fabric through an IP Network must combine an FC Entity with an FCIP Entity (see section 6.4) to form a complete interface between the FC Fabric and IP Network as shown in figure 3. +----------+ /\/\/\/\/\/\ +----------+ | FCIP | FCIP \ IP / Link | FCIP | | Entity |=========/ Network \=========| Entity | +----------+ \/\/\/\/\/\/ +----------+ | FC | | FC | | Entity | | Entity | +----------+ +----------+ | | /\/\/\/\/\/\ /\/\/\/\/\/\ \ FC / \ FC / / Fabric \ / Fabric \ \/\/\/\/\/\/ \/\/\/\/\/\/ Fig. 3 FC Entity and FCIP Entity Model The interface between the FC and FCIP Entities is implementation specific. The minimum requirements placed on an FC Entity by this specification are listed in annex D. More information about FC Entities can be found in the Fibre Channel standards and an example of an FC Entity can be found in FC-BB-2 [4]. Rajagopal, et al. Standards Track [Page 10] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 6.4 FCIP Entity The model for an FCIP Entity is shown in figure 4. ....................................................... : FCIP Entity : : : : +-----------+ : : | FCIP | : : | Control & |------------------------------------+ : : | Services | | : : | Module | | : : +-----------+ | : : | +--------------------+ | : : | +-------+--------------------+|----+ | : : | |+-----+--------------------+|----+| | : : | ||+----| FCIP Link Endpoint |----+|| | : : | ||| +--------------------+ ||| | : :.............................................|||.....: | ||| ||| | | ||| ||| o<--+ | ||| unique TCP ||| | | | ||| connections-->||| | | | ||| ||| | | +----------+ /\/\/\/\/\/\ | | FC | \ IP / | | Entity | / Network \ | +----------+ \/\/\/\/\/\/ | | | /\/\/\/\/\/\ +------------------+ \ FC / +->IP Address & / Fabric \ +->Well Known Port \/\/\/\/\/\/ Fig. 4 FCIP Entity Model The FCIP Entity is the connection interface point for the IP Network and is the owner of the IP Address and Well Known Port used to form TCP connections. An FC Fabric to IP Network interface product SHALL contain one FCIP Entity for each IP Address assigned to the product. An FCIP Entity contains an FCIP Control & Services Module to provide the FC Entity with an interface to key IP Network features. The interfaces to the IP Network features is implementation specific, however, to maintain interoperability, the TCP/IP mechanisms used are specified in this document as follows: - TCP Connections - see section 7 Rajagopal, et al. Standards Track [Page 11] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 - Security - see section 8 - Performance - see section 9 - Discovery - see section 10 The FCIP Link Endpoints in an FCIP Entity provide the FC Frame transmission features of FCIP. 6.5 FCIP Link Endpoint (FCIP_LEP) Each time a TCP connection is formed to an IP Address for which no TCP connection already exists, the FCIP Entity SHALL create a new FCIP Link Endpoint containing one FCIP Data Engine. An FCIP_LEP is a transparent data translation point between an FC Entity and an IP Network. A pair of FCIP_LEPs communicating over one or more TCP connections create an FCIP Link to join two islands of a FC Fabric, producing a single FC Fabric. The IP Network over which the two FCIP_LEPs communicate is not aware of the FC payloads that it is carrying. Likewise, the FC End Nodes connected to the FC Fabric are unaware of the TCP/IP based transport employed in the structure of the FC Fabric. As shown in figure 5, the FCIP Link Endpoint contains one FCIP Data Engine for each TCP connection in the FCIP Link. ................................................ : FCIP Link Endpoint : : +------------------+ : : +-------+------------------+|----+ : : |+-----+------------------+|----+| : : ||+----| FCIP Data Engine |----+|| : : ||| +------------------+ ||| : :..............................................: ||| ||| +----------+ /\/\/\/\/\/\ | FC | \ IP / | Entity | / Network \ +----------+ \/\/\/\/\/\/ | /\/\/\/\/\/\ \ FC / / Fabric \ \/\/\/\/\/\/ Fig. 5 FCIP Link Endpoint Model Rajagopal, et al. Standards Track [Page 12] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 An FCIP_LEP uses normal TCP based flow control mechanisms for managing its internal resources and matching them with the advertised TCP Receiver Window Size. An FCIP_LEP MAY communicate with its FC Entity counterpart to coordinate flow control. 6.6 FCIP Data Engine (FCIP_DE) The model for one of the multiple FCIP_DEs that may be present in an FCIP_LEP is shown in figure 6. +--------------------------------+ | | |-+ +------------------+ +-| C |p| | Encapsulation | |p| N F h -->|1|--->| Engine |--->|2|--> e i a |-+ +------------------+ +-| t b n | | I w r n |-+ +------------------+ +-| P o e e |p| | De-Encapsulation | |p| r l <--|4|<---| Engine |<---|3|<-- k |-+ +------------------+ +-| | | +--------------------------------+ Fig. 6 FCIP Data Engine Model Data enters and leaves the FCIP_DE through four portals (p1 - p4). The portals do not process or examine the data that passes through them. They are only the named access points where the FCIP_DE interfaces with external world. The names of the portals are as follows: p1) FC Receiver Portal - The interface through which an FC Frame enters an FCIP_DE from the FC Entity. p2) Encapsulated Frame Transmitter Portal - The TCP interface through which an FCIP encapsulated frame is transmitted to the IP Network by an FCIP_DE. p3) Encapsulated Frame Receiver Portal - The TCP interface through which an FCIP encapsulated frame is received from the IP Network by an FCIP_DE. p4) FC Transmitter Portal - The interface through which a reconstituted FC frame exits an FCIP_DE to the FC Entity. The work of the FCIP_DE is done by the Encapsulation and De- Encapsulation Engines. The Engines have two functions: Rajagopal, et al. Standards Track [Page 13] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 1) Encapsulating and de-encapsulating FC Frames using the encapsulation format described in FC Frame Encapsulation [23] and in section 6.6.1 of this document, and 2) Detecting some data transmission errors and performing minimal error recovery as described in section 6.6.2. Data flows through the FCIP_DE as follows: 1) An FC Frame arrives at the FC Receiver Portal and is passed to the Encapsulation Engine. The FC Frame is assumed to have been processed by the FC Entity according to the applicable FC rules and is not validated by the FCIP_DE. 2) In the Encapsulation Engine the encapsulation format described in FC Frame Encapsulation [23] and in section 6.6.1 of this document SHALL be applied to prepare the FC Frame for transmission over the IP Network. 3) The entire encapsulated frame SHALL be passed to the Encapsulated Frame Transmitter Portal where it SHALL be inserted in the TCP byte stream. 4) Transmission of the encapsulated frame over the IP Network follows all the TCP rules of operation. This includes but is not limited to the in-order delivery of bytes in the stream, as specified by TCP [9]. 5) The encapsulated FC Frame arrives at the partner FCIP Entity where it enters the FCIP_DE through the Encapsulated Frame Receiver Portal and is passed to the De-Encapsulation Engine for processing. 6) The De-Encapsulation Engine SHALL validate the incoming TCP byte stream as described in section 6.6.2 and SHALL de-encapsulate the FC Frame according to the encapsulation format described in FC Frame Encapsulation [23] and in section 6.6.1 of this document. 7) In the absence of errors, the de-encapsulated frame SHALL be passed to the FC Transmitter Portal for delivery to the FC Entity. Every FC Frame that arrives at the FC Receiver Portal SHALL be transmitted on the IP Network as described in steps 1 through 4 above. Data bytes arriving at the Encapsulated Frame Receiver Portal SHOULD be transmitted to the FC Transmitter Portal as described in steps 5 through 7, but this MAY NOT always be the case. Rajagopal, et al. Standards Track [Page 14] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 6.6.1 FCIP Encapsulation of FC Frames The FCIP encapsulation of FC frames employs FC Frame Encapsulation [23]. The features from FC Frame Encapsulation that are unique to individual protocols SHALL be applied as follows for the FCIP encapsulation of FC frames. The Protocol# field SHALL contain 1 in accordance with the IANA Considerations annex of FC Frame Encapsulation [23]. The Protocol Specific field SHALL have the format shown in figure 7. Note: the word numbers in figure 7 are relative to the complete FC frame encapsulation header, not to the Protocol Specific field. W|------------------------------Bit------------------------------| o| | r|3 3 2 2 2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 | d|1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0| +---------------------------------------------------------------+ 1| replication of encapsulation word 0 | +-------------------------------+-------------------------------+ 2| reserved | -reserved | +-------------------------------+-------------------------------+ Fig. 7 FCIP Usage of FC Frame Encapsulation Protocol Specific field Word 1 of the Protocol Specific field SHALL contain an exact copy of word 0 in FC Frame Encapsulation [23]. Word 2 of the Protocol Specific field is reserved for future enhancements to the FCIP protocol. The reserved field (bits 31-16 in word 2): SHALL contain 0. The -reserved field (bits 15-0 in word 2): SHALL contain 65535 (or 0xFFFF). The CRCV (CRC Valid) Flag SHALL be set to 0. The CRC field SHALL be set to 0. Rajagopal, et al. Standards Track [Page 15] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 6.6.2 FCIP Data Engine Error Detection and Recover 6.6.2.1 TCP Assistance With Error Detection and Recovery The FCIP_LEP assumes that, if TCP determines that there are TCP checksum errors, TCP applies the appropriate TCP retransmission and error recovery procedures. So the FCIP_DE gets an ordered delivery of FCIP frames with the TCP detected errors being transparent to the FCIP_DE. 6.6.2.2 Errors in FCIP Headers and Discarding FCIP Frames Bytes delivered through the Encapsulated Frame Receiver Portal that are not correctly delimited as defined by the FC Frame Encapsulation [23] SHOULD NOT be forwarded on to the FC Entity. Synchronization of the FCIP_DE to the FCIP Frames in the data stream entering the Encapsulated Frame Receiver Portal is maintained using the FC Frame Encapsulation header's frame length field to determine where in the data stream the next FC Encapsulation header is located. Synchronization SHALL be verified by checking the validity and positioning of any combination of the following FC Frame Encapsulation information: a) Protocol # field and its ones complement; b) Version field and its ones complement; c) Replication of encapsulation word 0 in word 1; d) Reserved field and its ones complement; e) Flags field and its ones complement; f) Length field and its ones complement; g) Time stamp [integer] and time stamp [fraction] fields; h) CRC field is equal to zero; i) SOF fields and ones complement fields; j) Format and values of FC header; k) CRC of FC frame; l) EOF fields and ones complement fields; and/or m) FC Encapsulation header information of next encapsulated frame. For FCIP Frames with header errors, the FCIP_DE SHALL discard the frame. Such errors should be considered carefully, since some may be synchronization errors. Errors in encapsulated FCIP Frames detected by the FCIP_DE that affect synchronization with the Encapsulated Frame Receiver Portal byte stream SHALL be handled as defined by section 6.6.2.4. An error in an encapsulated FCIP Frame that effects the synchronization may require the FCIP Entity to notify the FC Entity that the previously delivered FC Frame was invalid. Rajagopal, et al. Standards Track [Page 16] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 Whenever an FCIP_DE discards bytes delivered through the Encapsulated Frame Receiver Portal, it SHALL cause the FC Entity to be notified and provided with a suitable description of the reason bytes were discarded. 6.6.2.3 IP Network Transit Time Validation The FCIP_DE SHALL use the valid time stamp information in the FC Frame Encapsulation [23] header to determine if received FCIP Frames have been delayed by more than R_A_TOV in the IP Network. If an FCIP Frame has been delayed by more than R_A_TOV in the IP network, the FCIP_DE SHALL discard the FCIP Frame as described in section 6.6.2.2. The discarding of delayed FCIP frames SHALL continue until a FCIP Frame is processed whose life in the IP Network is smaller than R_A_TOV. Note that unlike a physical Fibre Channel link, an FCIP Link MAY involve IP routing dynamics that produce reliable, ordered delivery at the TCP layer, with the result that some FC Fabric operating constraints may be violated. The FCIP_DE is responsible for detecting violations R_A_TOV FC Fabric constraint and discarding affected frames. 6.6.2.4 Synchronization Failures If an FCIP_DE determines that it cannot find the next FCIP header in the byte stream entering through the Encapsulated Frame Receiver Portal, the FCIP_DE SHALL either: a) close the TCP connection [9] [11]; b) recover synchronization by searching the bytes delivered by the Encapsulated Frame Receiver Portal for a valid FCIP Frame header having the correct properties, and discarding bytes delivered by the Encapsulated Frame Receiver Portal until a valid FCIP Frame header is found; or c) attempt to recover synchronization as described in b) and if synchronization cannot be recovered close the TCP connection as described in a). If the FCIP_DE attempts to recover synchronization, the resynchronization algorithm used SHALL meet the following requirements: a) discard or identify with an EOFa (see FC-FS [6] and FC Frame Encapsulation [23])those FC frames and fragments of frames identified before synchronization has again been completely Rajagopal, et al. Standards Track [Page 17] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 verified. The number of FC frames not forwarded may vary based on the algorithm used; b) return to sending valid FC frames only after synchronization has been verified; and c) close the TCP/IP connection if the algorithm ends without verifying successful synchronization. The probability of failing to synchronize successfully and the time necessary to determine whether or not synchronization was successful may vary with the algorithm used. An example algorithm meeting these requirements can be found in annex A. 7. TCP Connection Management 7.1 TCP Connection Establishment 7.1.1 Creating a New TCP Connection The FC Entity SHALL request creation of a new TCP Connection by transmitting at least the following information to the FCIP Entity: - IP Address - R_A_TOV for the FCIP_Link - TCP Connection Parameters (see section 7.2) - Security Parameters (see section 8) - Quality of Service Parameters (see section 9) In response to a request from the FC Entity the FCIP Entity shall generate a TCP connect request [9] to the FCIP Well-Known Port at the specified IP Address. If the TCP connect request is rejected, the FCIP Entity SHALL so inform the FC Entity. If the TCP connect request is accepted, and the IP Address is one to which no other TCP connections exist, the FCIP Entity SHALL: 1) Create a new FCIP_LEP for the new FCIP Link, 2) Create a new FCIP_DE within the newly created FCIP_LEP to service the new TCP connection, and 3) Inform the FC Entity of the new FCIP_LEP and FCIP_DE. Rajagopal, et al. Standards Track [Page 18] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 If the TCP connect request is accepted, and the IP Address is one for which a TCP connection already exists, the FCIP Entity SHALL: 1) Create a new FCIP_DE within the existing FCIP_LEP to service the new TCP connection, and 2) Inform the FC Entity of the FCIP_LEP and new FCIP_DE. 7.1.2 Processing TCP Connect Requests The FCIP Entity SHALL listen for new TCP connection requests [9] on the FCIP Well-Known Port. An FCIP Entity MAY also accept and establish TCP connections to a TCP port number other than the FCIP Well-Known Port, as configured by the network administrator. Upon receipt of a TCP connect request, the FCIP Entity SHALL determine if a TCP connection already exists for the IP Address making the TCP connect request. The FCIP Entity SHALL notify the FC Entity of the TCP connect request, transmitting at least the following information: - IP Address - R_A_TOV for the FCIP_Link (zero for a new FCIP_LEP) - Information about the FCIP_LEP, new or existing - Information about the FCIP_DE for the new TCP connection - TCP Connection Parameters (see section 7.2) - Security Parameters (see section 8) - Quality of Service Parameters (see section 9) In response to the information provided by the FCIP Entity, the FC Entity MUST either accept or reject the TCP connect request. If the FC Entity rejects the TCP connect request, the FCIP Entity SHALL terminate the TCP connect request [9]. If the FC Entity accepts the TCP connect request, the FCIP Entity SHALL: 1) Accept the TCP connect request, 2) Finalize creation of the new FCIP_DE for the new TCP connection, and 3) If the new TCP connection is to an IP Address for which no other TCP connection exists, finalize the creation of the FCIP_LEP. 7.2 TCP Connection Parameters In order to provide efficient management of FCIP_LEP resources as well as FCIP Link resources, coordination of certain TCP connection parameters between the FC Entity and FCIP Entity is RECOMMENDED. Rajagopal, et al. Standards Track [Page 19] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 7.2.1 TCP Selective Acknowledgement Option The Selective Acknowledgement option RFC 2883 [22] allows the receiver to acknowledge multiple lost packets in a single ACK, enabling faster recovery. If authorized by the FC Entity, an FCIP Entity MAY negotiate use of TCP SACK and use it for faster recovery from lost packets and holes in TCP sequence number space. 7.2.2 TCP Window Scale Option This option allows TCP window sizes larger than 16-bit limits to be advertised by the receiver. It is necessary to allow data in long fat networks to fill the available pipe. This also implies buffering on the TCP sender that matches the (bandwidth*RTT) product of the TCP connection. An FCIP_LEP SHALL use locally available mechanisms to set a window size that matches the available local buffer resources and the desired throughput. 7.2.3 IP DSCP Option The RECOMMENDED IP DSCP field setting is 101110 corresponding to the EF service. 7.2.4 Protection against sequence number wrap It is RECOMMENDED that FCIP Entities implement protection against sequence number wrap. It is quite possible that within a single connection, TCP sequence numbers wrap within a timeout window. 7.2.5 TCP No Delay Option FCIP Entities SHALL disable the Nagle TCP No Delay option. This option is designed for usage in a telnet environment. 7.2.6 TCP Acknowledgement Timeout TCP has a TCP acknowledgement timeout. This is a variable timeout. 7.3 TCP Connection Considerations An FCIP_LEP SHALL implement established TCP mechanisms as defined in RFC 2581 [18] for congestion control on its connections. Rajagopal, et al. Standards Track [Page 20] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 It is RECOMMENDED that FCIP_LEPs use the TCP mechanisms for Long Fat Networks (LFNs) (i.e. an IP network with large (bandwidth*delay) product), as defined in RFC 1072 [10]. In idle mode, a TCP connection "keep alive" option of TCP is normally used to keep a connection alive. However, this timeout is fairly large and may prevent early detection of loss of connectivity. In order to facilitate faster detection of loss of connectivity, FC Entities SHOULD implement some form of Fibre Channel connection failure detection. 7.4 Flow Control Mapping between TCP and FC The FCIP Entity and FC Entity are connected the IP Network and FC Fabric, respectively, and they need to follow the flow control mechanisms of both TCP and FC, which work independent of each other. This section provides guidelines as to how the FCIP Entity can map TCP flow control to status notifications to the FC Entity. There are two scenarios when the flow control management becomes crucial: 1) When there is line speed mismatch between the FC and IP interfaces. Even though it is RECOMMENDED that both the FC and IP interfaces to the FC Entity and FCIP Entity, respectively, be of comparable speeds, it is possible to carry FC traffic over an IP Network that has a different line speed and bit error rate. 2) When the FC Fabric or IP Network encounters congestion. Even when both the FC Fabric or IP network are of comparable speeds, during the course of operation the FC Fabric or the IP Network could encounter congestion due to transient conditions. The FCIP Entity and FC Entity need to work cooperatively to use the available flow control mechanisms in the TCP and FC protocols to handle these situations. This specification does not specify any particular mechanism to handle the flow control but leaves this to implementation's choice. If the Encapsulation Frame Transmitter Portal is unable to transmit encapsulated FCIP Frames at the experienced data rate, the FCIP Entity MUST request that the FC Entity reduce the rate at which new FC Frames arrive at the FC Receiver Portal. Rajagopal, et al. Standards Track [Page 21] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 If the FC Receiver Portal is unable to accept de-encapsulated FC Frames at the experienced rate, the FC Entity MAY request the FCIP Entity to reduce the rate at which new FC Frames are delivered. The FCIP_DE MAY use TCP windowing techniques to control the packet arrival rate from the IP Network. This MAY involve advertising zero- window on TCP connection(s) occasionally so that the TCP connection(s) are flow controlled while the FC Fabric is encountering congestion. 8. Security 8.1 Considerations Using a wide-area, general purpose network such as an IP Network in a position normally occupied by physical cabling introduces some security problems not normally encountered in Fibre Channel Fabrics. FC transport media are typically protected physically from outside access; IP Networks typically invite outside access. The general effect is that the security of the entire FC Fabric is only as good as the security of the entire IP Network through which it tunnels. The following broad classes of attacks are possible: 1) Unauthorized Fibre Channel elements can gain access to resources through normal Fibre Channel processes. 2) Unauthorized agents can monitor and manipulate Fibre Channel traffic flowing over physical media used by the IP Network and under control of the agent. To a large extent, these security risks are typical of the risks facing any other application using an IP Network. They are mentioned here only because Fibre Channel storage networks are not normally suspicious of the media. Fibre Channel Fabric administrators will need to be aware of these additional security risks. 8.2 IP Network Security Requirements Security protocols and procedures used in other IP applications MAY be used for FCIP. FCIP Entities MUST ensure secure operation of FCIP Links by implementing one of the following two methods: 1) by using ESP [13] from the IPSec Security Protocol Suite with NULL encryption [14] for cryptographic data integrity and integrity of authentication. Authentication is performed using SRP [RFC2945]. This method is discussed in section 8.3; or Rajagopal, et al. Standards Track [Page 22] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 2) by appropriate configuration of an external entity that implements IP security using mechanisms such as IPSec and Virtual Private Networks. This method is discussed in section 8.4. The mechanism for configuring whether a particular deployment uses 1) or 2) is outside the scope of this document. Note: Two overviews of the IPSec Security Protocol Suite are available in RFC 2401 [12] and RFC 2411 [15]. 8.3 Integrated Security When both FCIP Entity and IP Security implementations are integrated into a single device, IPSec ESP (in transport mode) MUST be implemented. Upon receiving a TCP connection request, the receiving FCIP Entity SHALL identify the FCIP Link per the IP address pair of the connection. It SHALL then verify that the FCIP Link has been previously authenticated. If not, the FCIP Entity SHALL authenticate a new peer using a separate TCP connection. This TCP connection is used for negotiation of SRP related parameters. If authentication fails, the original TCP connection that initiated the authentication exchange is terminated and the FC Entity is not informed that a TCP connect request was received. If the authentication is successful, a new FCIP_LEP is created, with the authenticated FCIP Link as described in section 7.1.2. The FCIP Entity remembers the IP pair and the key material for authentication, so that any future TCP connections for that IP address pair bypasses this authentication step. The key material is then used as part of the ESP Security Parameters. Rajagopal, et al. Standards Track [Page 23] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 8.4 External Security Gateway Figure 8 illustrates the use of an externally supplied security gateway for securing the FCIP Link. +--------+ Insecure +--------+ Secure +--------+ Insecure +-------+ | FCIP | Network | IPSec | Network | IPSec | Network |FCIP | | Entity |----------| Device |---------| Device |----------|Entity | +--------+ +--------+ +--------+ +-------+ Fig. 8 External Security Gateway Model In this deployment, only certain parts of the FCIP Link are exposed to security threats and so only these specific parts of the FCIP Link need to be secured. The part of the network between the two security gateways is secured using devices implementing IPSec. The IPSec Device or any other equivalent gateway is required to operate in tunnel mode, so that the IP addresses of the two FCIP Entities are visible through the security devices that are implemented. 8.5 Security Information Exchanged Between FC and FCIP Entities TBD 9. Performance 9.1 Considerations The FCIP_DE does not interpret the contents of an FC Frame (except for attaching the correct byte-encoded SOF and EOF) nor does it do any FC payload processing. This allows any FC traffic to be tunneled across at high throughput rates. If fragmentation at the data link and IP layers is avoided by the use of path MTU discovery, throughput performance is enhanced. The Flow Control Protocol (discussed in section 7.4) provides the ability to stream gigabit FC data when using a large window size. It is RECOMMENDED that FCIP Entities use the TCP mechanisms for Long Fat Networks (LFNs) when they are used in IP networks with a large (bandwidth*delay) product. These mechanisms include TCP window scale option, Selective Acknowledgement, among others. See section 7.2. In order to achieve better TCP aggregate throughput in the face of packet losses, a pair of peer FC Entities MAY use multiple TCP Rajagopal, et al. Standards Track [Page 24] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 connections, and use appropriate policies for mapping FC Frames to these connections. Section 7.1 describes the steps an FCIP Entity takes in support of multiple TCP connection usage. All other aspects of using multiple TCP connections are outside the scope of this document. The reason for using multiple TCP connections is the TCP's slow- start algorithm, which reduces TCP's window whenever it detects congestion in the network. If, on the other hand, the traffic is distributed across multiple connections, all the connections will not be affected at the same time, resulting in a better aggregate throughput. Note that even though multiple connections provide better aggregate throughput (when packet losses occur on IP Networks), their use is not a requirement. A pair of FC Entities MAY choose to use single TCP connection to tunnel the FC traffic. 9.2 QoS Support The Differentiated Services Architecture (diffserv) provides a "Class of Service" to a flow aggregate [16], [17]. At so-called diffserv boundaries, IP packets are classified and marked. Within the diffserv domain, resources - bandwidth and buffers - are allocated for each classification. Packets with the same classification use the resources allocated for the classification. IP packets with the same destination and class marking exit a diffserv capable router in the same order they arrived. Packets with the same destination but different class markings exit according to priorities assigned to the different class markings. The Diffserv has renamed the Ipv4 TOS field as Differentiated Services Code Point (DSCP). The DSCP indicates the particular behavior a packet is to receive at each router. How a packet gets marked is based on a policy administered and configured into the network. [20] and [19] provide various encodings of the DSCP field to achieve a specific behavior from the routers. There may be several ways to administer the policies and the policy definition is up to the network provider. That is one network provider MAY choose to mark all packets going from one source IP address to a specific destination as "high priority", while another might mark just a specific traffic type (e.g., HTTP) as "high priority". Thus packets carry the desired class information and each diffserv-capable router treats the packet according to the information in its DSCP field. This is referred to as Per Hop Rajagopal, et al. Standards Track [Page 25] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 Behavior (PHB). Currently, the IETF standards define essentially 3 types of services: Expedited Forwarding (EF) [20], Assured Forwarding (AF) [19], or Default Forwarding (DF) [16], [17] - that corresponds to its DSCP. [17] specifies the AF service AF PHB provides a way to prioritize best-effort traffic. Currently, 4 AF classes and 3 drop precedence levels are specified providing 12 different levels of forwarding assurances. The DSCP value specifies a drop-order in the event that a packet experiences congestion at a subsequent diffserv router. [20] specifies the DSCP code point equal to 101110 EF service which is also sometimes refereed to as "Premium" service. When supported, this class behavior has the lowest levels of latency, packet loss, and delay variation. This service behavior most closely matches the Fibre Channel characteristics. This is therefore the RECOMMENDED DSCP setting in the IP DSCP field. What resources are not used for EF and AF are left for the DF services which is really a best-effort service. Note that if a packet is being forwarded over an underlying network without diffserv support, then the packet would simply receive best- effort service regardless of its DSCP field setting. 9.3 QoS Information Exchanged Between FC and FCIP Entities TBD 10. Dynamic Discovery of Participating FCIP Entities 10.1 Requirements If dynamic discovery of participating FCIP Entities is supported the function SHALL be performed using the Service Location Protocol (SLPv2) [21]. Additional details TBD. 10.2 Discovery Information Exchanged Between FC and FCIP Entities TBD Rajagopal, et al. Standards Track [Page 26] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 11. References [1] Bradner, S., "The Internet Standards Process -- Revision 3", BCP 9, RFC 2026, October 1996. [2] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [3] Fibre Channel Backbone (FC-BB), T11 Project 1238-D, Rev 4.8, March 5, 2001 (www.t11.org). [4] Fibre Channel Backbone -2 (FC-BB-2), T11 Project 1466-D, (www.t11.org). [5] Fibre Channel Switch Fabric -2 (FC-SW-2), T11 Project 1305-D, Rev. 5.2, May 23, 2001 (www.t11.org). [6] Fibre Channel Framing and Signaling (FC-FS), T11 Project 1331-D, Rev 1.2, February 16, 2001 (www.t11.org). [7] Fibre Channel Generic Services -3, ANSI NCITS.348-200x, November 28, 2000. [8] http://www.t11.org [9] "Transmission Control Protocol", RFC 793, Sept. 1981. [10] Jacobson & Braden, "TCP Extensions for Long-Delay Paths", RFC 1072, October 1988. [11] Braden, "Requirements for Internet Hosts -- Communication Layers", RFC 1122, October 1989 [12] Kent, S. and Atkinson, R., "Security Architecture for the Internet Protocol", RFC 2401, Nov. 1998. [13] Kent, S. and Atkinson, R., "IP Encapsulating Security Payload (ESP)", RFC 2406, Nov. 1998. [14] Glenn, R., Kent, S., "The NULL Encryption Algorithm and Its Use With IPsec", RFC 2410, Nov. 1998 [15] Thayer, R., Glenn, R., and Doraswamy, N., "IP Security Document Roadmap", RFC 2411, Nov. 1998. [16] Nichols, K., Blake, S., Baker, F. and D. Black, "Definition of the Differentiated Services Field (DS Field) in the IPv4 and Ipv6 Headers", RFC 2474, December 1998. Rajagopal, et al. Standards Track [Page 27] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 [17] Blake, S., Black, D., Carlson, M., Davies, E., Wang, Z., Weiss, W., "An Architecture for Differentiated Services", RFC 2475, Dec. 1998. [18] Allman, et. al., "TCP Congestion Control", RFC 2581, April 1999. [19] Heinanen, J., Baker, F., Weiss, W., Wroclawski, J., "An Assured Forwarding PHB", RFC 2597, June 1999. [20] Jacobson, V., Nichols, K., Poduri, K., "An Expedited Forwarding PHB Group", RFC 2598, June 1999. [21] E.Guttman, C. Perkins, J. Veizades, M. Day. Service Location Protocol, version 2, RFC 2608, July, 1999. [22] Floyd, et al, "SACK Extension", RFC 2883, July 2000. [23] Weber, Rajagopal, Travostino, Chau, O'Donnell, Monia Merhar, "FC Frame Encapsulation", draft-ietf-ips-fcencapsulation-__.txt (RFC reference and date to be added during standards action). The following reference concerns SLP, see [21]. It is not referenced in this revision of this draft but may be referenced in future revisions. [24] E.Guttman, C. Perkins, J. Kempf. Service Templates and Service: Schemes, RFC 2609, July 1999. 12. Bibliography The following references may prove informative to readers unfamiliar with Fibre Channel. Kembel, R., "The Fibre Channel Consultant: A Comprehensive Introduction", Northwest Learning Associates, 1998 13. Acknowledgments Funding for the RFC Editor function is currently provided by the Internet Society. Rajagopal, et al. Standards Track [Page 28] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 14. Authors' Addresses Murali Rajagopal Raj Bhagwat LightSand Communications, Inc. LightSand Communications, Inc. 24411 Ridge Route Dr. 24411 Ridge Route Dr. Suite 135 Suite 135 Laguna Hills, CA 92653 Laguna Hills, CA 92653 USA USA Phone: +1 949 837 1733 x101 Phone: +1 949 837 1733 x104 Email: muralir@lightsand.com Email: rajb@lightsand.com R. Andy Helland Elizabeth G. Rodriguez LightSand Communications, Inc. Lucent Technologies 375 Los Coches Street 1202 Richardson Drive, Suite 104 Milpitas, CA 95035 Richardson, TX 75080 USA USA Phone: +1 408 404 3119 Phone: +1 972 231 0672 Fax: +1 408 941 2166 Fax: +1 972 644 5857 Email: andyh@lightsand.com Email: egrodriguez@lucent.com Sriram Rupanagunta Neil Wanamaker Aarohi Communications Akara 3200 Montelena Drive 10624 Icarus Court San Jose, CA 95135 Austin, TX 78726 USA USA Phone: +1 408 966 8309 Phone: +1 512 257 7633 Email: sriramr@aarohi-inc.com Fax: +1 512 257 7877 Email: nwanamaker@akara.com Steve Wilson Bob Snively Brocade Comm. Systems, Inc. Brocade Comm. Systems, Inc. 1745 Technology Drive 1745 Technology Drive San Jose, CA. 95110 San Jose, CA 95110 USA USA Phone: +1 408 487 8128 Phone: +1 408 487 8135 Fax: +1 408 487 8101 Email: rsnively@brocade.com email: swilson@brocade.com Ralph Weber ENDL Texas, representing Brocade Suite 102 PMB 178 18484 Preston Road Dallas, TX 75252 USA Phone: +1 214 912 1373 Email: roweber@acm.org Rajagopal, et al. Standards Track [Page 29] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 David Peterson Donald R. Fraser Cisco Systems - SRBU Compaq Computer Corporation 6450 Wedgwood Road 301 Rockrimmon Blvd., Bldg. 5 Maple Grove, MN 55311 Colorado Springs, CO 80919 USA USA Phone: +1 763 398 1007 Phone: +1 719 548 3272 Cell: +1 612 802 3299 Email: don.fraser@compaq.com Email: dap@cisco.com Vi Chau Gaby Hecht Gadzoox Networks, Inc. Gadzoox Network, Inc. 16241 Laguna Canyon Road 16241 Laguna Canyon Road Suite 100 Suite 100 Irvine, CA 92618 Irvine, CA 92618 USA USA Phone: +1 949 789 4639 Phone: +1 949 789 4642 Fax: +1 949 453 1271 Fax: +1 949 453 1271 Email: vchau@gadzoox.com Email: ghecht@Gadzoox.com Ken Hirata Jim Nelson Vixel Corporation Vixel Corporation 15245 Alton Parkway, Suite 100 15245 Alton Parkway, Suite 100 Irvine, CA 92618 Irvine, CA 92618 USA USA Phone: +1 949 788 6368 Phone: +1 949 450 6159 Fax: +1 949 753 9500 Fax: +1 949 753 9500 Email: ken.hirata@vixel.com Email: Jim.Nelson@vixel.com Michael E. O'Donnell Anil Rijhsinghani McDATA Corporation McDATA Corporation 310 Interlocken Parkway 5 Brickyard lane Broomfield, Co. 80021 Westboro, MA 01581 USA USA Phone: +1 303 460 4142 Phone: +1 508 870 6593 Fax: +1 303 465 4996 Email: Email: modonnell@mcdata.com anil.rijhsinghani@mcdata.com Milan J. Merhar Craig W. Carlson 43 Nagog Park QLogic Corporation Pirus Networks 6321 Bury Drive Acton, MA 01720 Eden Prairie, MN 55346 USA USA Phone: +1 978 206 9124 Phone: +1 952 932 4064 Email: Milan@pirus.com Email: craig.carlson@qlogic.com Rajagopal, et al. Standards Track [Page 30] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 Venkat Rangan Larwrence J. Lamers Rhapsody Networks Inc. SAN Valley 3450 W. Warren Ave. 4611 Park Norton Place Fremont, CA 94538 San Jose, CA 95136-2523 USA USA Phone: +1 510 743 3018 Phone: +1 408 626 1285 Fax: +1 510 687 0136 Email: ljlamers@ieee.org Email: venkat@rhapsodynetworks.com 15. Full Copyright Statement Copyright (C) The Internet Society (2001). All Rights Reserved. This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this paragraph are included on all such copies and derivative works. However, this document itself may not be modified in any way, such as by removing the copyright notice or references to the Internet Society or other Internet organizations, except as needed for the purpose of developing Internet standards in which case the procedures for copyrights defined in the Internet Standards process must be followed, or as required to translate it into languages other than English. The limited permissions granted above are perpetual and will not be revoked by the Internet Society or its successors or assigns. This document and the information contained herein is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. ANNEX A - Example of synchronization recovery algorithm Synchronization may be recovered as specified in section 6.6.2.4. An example of an algorithm for searching the bytes delivered to the Encapsulated Frame Receiver Portal for a valid FCIP Frame header is provided in this annex. This resynchronization uses the principle that a valid FCIP data stream must contain at least one valid header every 2148 bytes (the maximum length of an encapsulated frame). Although other data patterns containing apparently valid headers may be contained in the Rajagopal, et al. Standards Track [Page 31] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 stream, the FC CRC or frame validity of the data patterns contained in the data stream will always be either interrupted by or resynchronized with the valid FCIP Frame encapsulation headers. Consider the case shown in figure 9. A series of short FCIP Frames, perhaps from a trace, are embedded in larger FCIP Frames, say as a result of a trace file being transferred from one disk to another. The headers for the short frames are denoted SFH and the long frame headers are marked as LFH. +-+--+-+----+-+----+-+----+-+-+-+---+-+--- |L| |S| |S| |S| |S| |L| |S| |F| |F| |F| |F| |F| |F| |F|... |H| |H| |H| |H| |H| |H| |H| +-+--+-+----+-+----+-+----+-+-+-+---+-+--- | | |<---------2148 bytes-------->| Fig. 9 Example of resynchronization data stream A resynchronization attempt that starts just to the right of an LFH will find several SFH frames before discovering that they do not represent the transmitted stream of frames. Within 2148 bytes plus or minus, however, the resynchronization attempt will encounter an SFH whose length does not match up with the next SFH because the LFH will fall in the middle of the short frame pushing the next header farther out in the byte stream. Note that the resynchronization algorithm cannot forward any prospective FC Frames to the FC Transmitter Portal because until synchronization is completely established there is no certainty that anything that looked like an FCIP Frame really was one. For example, an SFH might fortuitously contain a length that points exactly to the beginning of an LFH. The LFH would identify the correct beginning of a transmitted frame, but that in no way guarantees that the SFH was also a correct FCIP Frame header. There exist some data streams that cannot be resynchronized by this algorithm. If such a data stream is encountered, the algorithm causes the TCP connection to be closed. The resynchronization assumes that security and authentication procedures outside the FCIP Entity are protecting the valid data stream from being replaced by an intruding data stream containing valid FCIP data. Rajagopal, et al. Standards Track [Page 32] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 The following steps are one example of how an FCIP_DE might resynchronize with the data stream entering the Encapsulated Frame Receiver Portal. 1) Search for candidate and strong headers: The data stream entering the Encapsulated Frame Receiver Portal is searched for 12 bytes in a row containing the required values for: a) Protocol field, b) Version field, c) ones complement of the Protocol field, d) ones complement of the Version field, e) replication of encapsulation word 0 in word 1, and f) Reserved field and its ones complement. If such a 12-byte grouping is found, the FCIP_DE assumes that it has identified bytes 0-2 of a candidate FCIP encapsulation header. All bytes up to and including the candidate header byte are discarded. If no candidate header has been found after searching a specified number of bytes greater than some multiple of 2148 (the maximum length of an encapsulated frame), resynchronization has failed and the TCP/IP connection is closed. Word 3 of the candidate header contains the Frame Length and Flags fields and their ones complements. If the fields are consistent with their ones complements, the candidate header is considered a strong candidate header. The Frame Length field is used to determine where in byte stream the next strong candidate header should be and processing continues at step 2). 2) Use multiple strong candidate headers to locate a verified candidate header: The Frame Length in one strong candidate header is used to skip incoming bytes until the expected location of the next strong candidate header is reached. Then the tests described in step 1) are applied to see if another strong candidate header has successfully been located. All bytes skipped and all bytes in all strong candidate headers processed are discarded. Strong candidate headers continue to be verified in this way for at least 4296 bytes (twice the maximum length of an encapsulated Rajagopal, et al. Standards Track [Page 33] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 frame). If at anytime a verification test fails, processing restarts at step 1 and a retry counter is incremented. If the retry counter exceeds 3 retries, resynchronization has failed and the TCP connection is closed. After strong candidate headers haves been verified for at least 4296 bytes, the next header identified is a verified candidate header and processing continues at step 3). Note: If a strong candidate header was part of the data content of an FC frame, the encapsulated frame defined by that or a subsequent strong candidate header will eventually cross an actual header in the byte stream. As a result it will either identify the actual header as a strong candidate header or it will lose synchronization again because of the extra 28 bytes in the length, returning to step 1 as described above. 3) Use multiple strong candidate headers to locate a verified candidate header: Incoming bytes are skipped and discarded until the next verified candidate header is reached. Each verified candidate header is tested against the full collection of tests listed in section 6.6.2.2 as would normally be the case. Verified candidate headers continue to be located and tested in this way for a minimum of 4296 bytes (twice the maximum length of an encapsulated frame). If all are verified candidate headers encountered are valid, the last verified candidate header is a valid header. At this point the FCIP_DE stops discarding bytes and begins normal FCIP de-encapsulation begins, including for the first time since synchronization was lost, delivery of FC frames through the FC Transmitter Portal according to normal FCIP rules. If any verified candidate headers are invalid but meet all the requirements of a strong candidate header, increment the retry counter and return to step 2). If any verified candidate headers are invalid and fail to meet the tests for a strong candidate header, increment the retry counter and return to step 1. If the retry counter exceeds 4 retries, resynchronization has failed and the TCP/IP connection is closed. A flowchart for this algorithm can be found in figure 10. Rajagopal, et al. Standards Track [Page 34] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 Synchronization is lost | _____________v_______________ | | | Search for candidate header | +----------->| | | | Found Not Found | | | (Strong candidate) | | |_____________________________| | | | | | + --------->Close TCP/IP | _______v_____________________ Connection | | | | | Enough strong candidate | | +---->| headers identified? | | | | | | | | No Yes | | | | (Verified candidate) | | | |_____________________________| |___________________| | ^ | | | | | | | _______________________v_____ | | | | | | | Enough verified candidate | | | | headers validated? | | | | | | | | No Yes | | | | (Resynchronized) | | | |_____________________________| | | | | | | ______v__________ | Resume | | | | + ---> Normal | | | Synchronization | De-encapsulation | | | Lost? | | | | | | | | No Yes | | | |_________________| | | | | | |________| | |___________________________| Fig. 10 Flow diagram of simple synchronization example Rajagopal, et al. Standards Track [Page 35] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 ANNEX B - Relationship between FCIP and IP over FC (IPFC) IPFC (RFC 2625) describes the encapsulation of IP packets in FC frames. It is intended to facilitate IP communication over an FC network. FCIP describes the encapsulation of FC frames in TCP segments which in turn are encapsulated inside IP packets for transporting over an IP network. It gives no consideration to the type of FC frame that is being encapsulated. Therefore, the FC frame may actually contain an IP packet as described in the IP over FC specification (RFC 2625). In such a case, the data packet would have: - Data Link Header - IP Header - TCP Header - FCIP Header - FC Header - IP Header Note: The two IP headers would not be identical to each other. One would have information pertaining to the final destination while the other would have information pertaining to the FCIP Entity. The two documents focus on different objectives. As mentioned above, implementation of FCIP will lead to IP encapsulation within IP. While perhaps inefficient, this should not lead to issues with IP communication. One caveat: if a Fibre Channel device is encapsulating IP packets in an FC frame (e.g. an IPFC device), and that device is communicating with a device running IP over a non-FC medium, a second IPFC device may need to act as a gateway between the two networks. This scenario is not specifically addressed by FCIP. There is nothing in either of the specifications to prevent a single device from implementing both FCIP and IP-over-FC (IPFC), but this is implementation specific, and is beyond the scope of this document. ANNEX C - FC Frame Format All FC frames have a standard format much like LAN's 802.x protocols. However, the exact size of each frame varies depending on the size of the variable fields. The size of the variable field ranges from 0 to 2112-bytes as shown in the FC Frame Format in figure 11 resulting in the minimum size FC Frame of 36 bytes and the maximum size FC frame of 2148 bytes. Valid Fibre Channel frame lengths are always a multiple of four bytes. Rajagopal, et al. Standards Track [Page 36] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 +------+--------+-----------+----//-------+------+------+ | SOF |Frame |Optional | Frame | CRC | EOF | | (4B) |Header |Header | Payload | (4B) | (4B) | | |(24B) |<----------------------->| | | | | | Data Field = (0-2112B) | | | +------+--------+-----------+----//-------+------+------+ Fig. 11 FC Frame Format SOF and EOF Delimiters On an FC link, Start-of-Frame (SOF) and End-Of-Frame (EOF) are called Ordered Sets and are sent as special words constructed from the 8B/10B comma character (K28.5) followed by three additional 8B/ 10B data characters making them uniquely identifiable in the data stream. On an FC link the SOF delimiter serves to identify the beginning of a frame and prepares the receiver for frame reception. The SOF contains information about the frame's Class of Service, position within a sequence, and in some cases, connection status. The EOF delimiter identifies the end of the frame and the final frame of a sequence. In addition, it serves to force the running disparity to negative. The EOF is used to end the connection in connection-oriented classes of service. It is therefore important to preserve the information conveyed by the delimiters across the IP-based network, so that the receiving FCIP Entity can correctly reconstruct the FC frame in its original SOF and EOF format before forwarding it to its ultimate FC destination on the FC link. When an FC frame is encapsulated and sent over a byte-oriented interface, the SOF and EOF delimiters are represented as sequences of four consecutive bytes, which carry the equivalent Class of Service and frame termination information as the FC ordered sets. The representation of SOF and EOF in an encapsulation FC frame is described in FC Frame Encapsulation [23]. Frame Header The FC Frame Header is transparent to the FCIP Entity. The FC Frame Header is 24 bytes long and has several fields that are associated with the identification and control of the payload. Current FC Standards allow up to 3 Optional Header fields [6]: - Network_Header (16-bytes) Rajagopal, et al. Standards Track [Page 37] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 - Association_Header (32-bytes) - Device_Header (up to 64-bytes). Frame Payload The FC Frame Payload is transparent to the FCIP Entity. An FC application level payload is called an Information Unit at the FC-4 Level. This is mapped into the Frame Payload of the FC Frame. A large Information Unit is segmented using a structure consisting of FC Sequences. Typically, a Sequence consists of more than one FC frame. FCIP does not maintain any state information regarding the relationship of frames within a FC Sequence. CRC The FC CRC is 4 bytes long and uses the same 32-bit polynomial used in FDDI and is specified in ANSI X3.139 Fiber Distributed Data Interface. This CRC value is calculated over the entire FC header and the FC payload; it does not include the SOF and EOF delimiters. Note: When FC frames are encapsulated into FCIP frames, the FC frame CRC is untouched by the FCIP Entity. ANNEX D - FCIP Requirements on an FC Entity The capabilities that FCIP requires of an FC Entity include: 1) The FC Entity must deliver FC frames to the correct FCIP Data Engine (in the correct FCIP Link Endpoint) and forward FC Frames from FCIP Data Engine(s) to the FC Fabric. 2) The only delivery ordering guarantee provided by FCIP is correctly ordered delivery of FC Frames between a pair of FCIP Data Engines. FCIP expects the FC Entity to implement all other FC Frame delivery ordering requirements. 3) The FC Entity must support the FCIP Entity in the processing of incoming connect requests by deciding to accept a connect request. 4) The FC Entity may generate connect and terminate requests. 5) The FC Entity may instruct the FCIP Entity regarding TCP connection parameter settings and the R_A_TOV to be applied to an FCIP Link. 6) The FC Entity may recover from connection failures. Rajagopal, et al. Standards Track [Page 38] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 7) The FC Entity must recover from events that the FCIP Entity cannot handle, such as: a) loss of synchronization with FCIP Frame headers from the Encapsulated Frame Receiver Portal requiring resetting the TCP connection, b) additional examples, TBD 8) The FC Entity must work cooperatively with the FCIP Entity to manage flow control problems in either the IP Network or FC Fabric. 9) The FC Entity may test for failed TCP connections. 10) TBD support for dynamic discovery 11) TBD support for security 12) TBD support for connection QoS features 13) TBD support for monitoring Note that the Fibre Channel standards MUST be consulted for a complete understanding of the requirements placed on an FC Entity. ANNEX E - FC-BB-2 Inputs This annex contains text from previous FCIP drafts that, because of the new model structure, probably belongs in FC-BB-2 [4]. As soon as the correctness of this annex is agreed, its contents will be transferred to a T11 document do be used in the development of FC-BB- 2. No attempt has been made to rewrite this text for inclusion in an T11 standards, so it should be considered a guide to T11 content no a specification. All section references in this annex come from draft-ietf-ips- fcovertcpip-02.txt. When only parts of a section are included here, "{partial}" is appended to the section title. 4.3 FCIP's Interaction with FC and TCP {partial} Since FC Primitive Signals and Primitive Sequences are not exchanged between FCIP devices, there may be times when an FC frame is lost within the IP network. When this event occurs it is the responsibility of the communicating FC devices to detect and correct the errors based on the features defined in FC-FS [6]. The FCIP devices MAY choose not to generate Fibre Channel's F_BSY or F_RJT Rajagopal, et al. Standards Track [Page 39] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 frames or otherwise participate in FC frame recovery. Note that the order of the FC Frames sent by the Encapsulated Frame Transmitter MAY not be the same as the order sent by the source FC End Node. This is due to the fact that some types of FC login allow FC Frames to be re-ordered in the FC Fabric before reaching the FC Receiver Port. 5.3 TCP Connection Management In order to realize a Virtual ISL between two FC end-points, an FCIP Device establishes TCP connection(s) with its peer FCIP Device. In order to achieve better TCP aggregate throughput properties in the face of packet losses, a pair of peer FCIP devices MAY use multiple TCP connections between them, and use appropriate policies for mapping FC frames to these connections. It may also be useful to assign a pool of connections for transmission of high priority and control messages (e.g., Class F messages) on connections so they do not encounter "head of line" blocking behind Class 2 or Class 3 traffic. The use of multiple connections and policies for distributing frames on these connections are described in section 5.5. A Virtual ISL and the two FCIP Device endpoints that are involved are operational only after the first TCP connection is established. The sequence of operations performed in order to establish a Virtual ISL is as follows. 1) The FCIP device initializes its local resources to enable it to listen to TCP connection requests. 2) The FCIP device discovers the FCIP device endpoints that it can establish a virtual ISL. The result of the discovery SHALL be, at the minimum, the IP address and the TCP port of the peer endpoint. The discovery process MAY rely on administrative configuration or on services such as SLP or iSNS (TBD). (Needs to have its own section eventually). 3) The FCIP device endpoint SHALL exchange security context and authenticate itself to the peer endpoint. The use of security context is explained in section 8. 4) After connection establishment, FCIP devices use the FCIP frame encapsulation defined in FC Frame Encapsulation [23] and in section 6.6.1 of this document. 5) At this point the FCIP device endpoint SHALL exchange Fibre Channel port initialization frames (SW_ILS) to enable and identify port operation. Port state machine and initialization Rajagopal, et al. Standards Track [Page 40] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 are described in FC-SW-2 [5]. 6) An FCIP device operates in E-port or B-Port mode. When operating in E-Port mode, normal FC-SW2 FSPF messages are exchanged and the switch port becomes operational. 7) For computing the link cost of the ISL, the following formula SHALL be used: . In certain deployments, a single FCIP device endpoint MAY establish virtual ISLs with multiple FCIP device endpoints. In this situation, the FCIP device endpoint SHALL manage TCP operational parameters independently for each ISL. Also, the FCIP Device Endpoint SHALL perform the E_Port or B_Port initialization independently, for each connection. 5.4.1 Determining loss of connectivity {partial} In order to facilitate faster detection of loss of connectivity, FC Switches MAY process the Hello (HLO) SW_ILS request through a pair of FCIP devices. The relationship between the HLO SW_ILS and the paired FCIP devices is TBD. Upon detecting a loss of connectivity, an FCIP Device SHALL establish a new connection, or SHALL use an existing TCP connection to the same FCIP Device endpoint. An FCIP Device SHALL NOT retransmit an FCIP frame on the new connection. This is to ensure exactly-once delivery semantics to the Fibre Channel endpoint. 5.5 Multiple Connection Management A pair of FCIP device endpoints MAY establish a certain number of TCP connections between them. Since a Virtual ISL potentially maps a fairly large number of FC flows (where a flow is a pair of Fibre Channel S_ID, D_ID addresses), it may not be practical to establish a separate TCP connection for each Fibre Channel flow. In order to address this, an implementation MAY choose to manage a pool of TCP connections for a single Virtual ISL and map Fibre Channel flows to TCP connections of that ISL. However, while assigning Fibre Channel flows to TCP connections, an implementation SHALL follow the following rules: 1) Once a Fibre Channel flow is assigned to a TCP connection within the virtual ISL, it SHALL send all Fibre Channel frames of that flow on that connection. Rajagopal, et al. Standards Track [Page 41] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 2) When an FCIP endpoint processes any response traffic from a particular target, the Endpoint SHALL send the response on the same connection on which the request was sent. 3) Any class 2 ACK frames SHALL be sent on the same connection in which the original frame was sent. These rules are in place to honor any in-order delivery guarantees that may have been made between the two end points of the Fibre Channel flow. 5.6 Multi Virtual ISL Management It is quite likely that a single switch MAY provide multiple Virtual ISLs, all providing alternate connectivity paths between two switches. In this situation, a switch SHALL select any of the available ISLs for mapping a FCIP flow. In doing so, a switch MUST follow a flow allegiance model, where a pair of Fibre Channel [S_ID, D_ID] end points are always mapped to the same Virtual ISL. Furthermore, switches SHALL implement a connection allegiance policy, which ensures that the responses to particular [S_ID, D_ID] pair is always sent back on the same Virtual ISL. 8.3 Corruption {partial} Data corruption is detected at two different levels: TCP checksum and Fibre Channel frame encapsulation errors. Data corruption detected at the TCP level SHALL be recovered via TCP data recovery mechanisms. The recovery for Fibre Channel frame errors is described below. The TCP and Fibre Channel frame recovery operations are performed independently. Fibre Channel frame errors and the expected resolution of those errors are described below: a) Incoming frames on the FC Receiver Port SHALL be verified for correct header, proper format, valid length and valid CRC. Frames having incorrect headers or CRC SHALL be discarded or processed in accordance with the rules for the particular type of FC Port. b) All frames transmitted by the Encapsulated Frame Transmitter shall be valid FC Encapsulations of valid FC frames with correct TCP check sums on the correct TCP/IP connection. e) The FC frames contained in incoming encapsulated frames on the FC Receiver Port SHALL be verified for a valid header, proper content, proper SOF and EOF values, and valid length. FC frames Rajagopal, et al. Standards Track [Page 42] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 that are not valid according to those checks SHALL be managed according to the following rules. 1) The frame may be discarded before being transmitted by the FC Transmitter Port. 2) The frame may be transmitted in whole or in part by the FC Transmitter Port and ended with an EOF indicating that the content of the frame is invalid. f) Any encapsulated frame received by the Encapsulated Frame Receiver that has an invalid Fibre Channel CRC shall be managed according to the following rules. 1) The frame may be transmitted unchanged by the FC Transmitter Port. The frame will be discarded by the receiving FC Port because of invalid CRC. 2) The frame may be discarded before being transmitted by the FC Transmitter Port. 3) The frame may be transmitted in whole or in part by the FC Transmitter Port and ended with an EOF indicating that the content of the frame is invalid. The FC encapsulation header Frame Length field MUST correctly specify the transmitted frame length. 8.4 Timeouts {partial} Fibre Channel has two important timeouts to consider in FCIP. These are: E_D_TOV, and R_A_TOV. E_D_TOV determines the life of an individual Fibre Channel frame in any particular fabric element. The effects of E_D_TOV on the fabric as a whole are typically cumulative since each fabric element contains it's own E_D_TOV timers for any frame received. R_A_TOV determines the life of an individual Fibre Channel frame in the fabric as a whole. For a fabric, R_A_TOV implies that no particular frame will remain in (and thus be emitted from) the fabric after the timer expires. 10.1 Flow control on FC network When the Fibre channel traffic is encapsulated over TCP connection(s), the FCIP device needs to ensure that the TCP connections can handle the frame arrival rate from FC Fabric. This MAY require FCIP device to use Buffer-to-Buffer flow control (see FC- Rajagopal, et al. Standards Track [Page 43] Internet-Draft Fibre Channel Over TCP/IP (FCIP) June, 2001 FS [6]) on its Fibre Channel port(s) to control the frame arrival rate. Rajagopal, et al. Standards Track [Page 44]