Internet Draft L. Yang Expiration: May 2003 Intel Labs File: draft-yang-forces-model-01.txt J. Halpern Working Group: ForCES R. Gopal Nokia R. Dantu Univ. of Texas Nov 2002 ForCES Forwarding Element Functional Model draft-yang-forces-model-01.txt Status of this Memo This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC2026. 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/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. Abstract This document defines a functional model for forwarding elements (FEs) used in the Forwarding and Control Plane Separation (ForCES) protocol. This model is used to describe the capabilities and state of ForCES forwarding elements within the context of the ForCES protocol, so that ForCES control elements (CEs) can control the FEs accordingly. The model is to specify what logical functions are present in the FEs, what capabilities these functions support, and in what order these functions are or can be performed. The forwarding element model defined herein is intended to satisfy the requirements specified in the ForCES requirements draft [FORCES- REQ]. Using this model, predefined or vendor specific logical Internet Draft ForCES FE Functional Model Nov 2002 functions can be expressed and configured. However, the definition of these individual functions are not described and defined in this document. Table of Contents Abstract...........................................................1 1. Definitions.....................................................2 2. Motivation and Requirements of FE model.........................3 3. Capability Model versus State Model.............................3 4. FE Model........................................................6 4.1. FE Blocks..................................................7 4.2. FE Block Library...........................................7 4.2.1. QoS Functions.........................................8 4.2.2. Generic Filtering Functions..........................10 4.2.3. Vendor Specific Functions............................10 4.2.4. Port Functions.......................................10 4.2.5. Forwarding Functions.................................11 4.2.6. High-Touch Functions.................................12 4.2.7. Security Functions...................................12 4.2.8. Off-loaded Functions.................................12 4.3. FE Stage and Directed Graph of FE.........................13 4.3.1. Basic Concepts.......................................13 4.3.2. Topological versus Encoded State Approaches..........13 4.3.3. Cascading FE Blocks..................................16 5. Data Modeling and Representation...............................16 6. Security Considerations........................................17 7. Intellectual Property Right....................................17 8. IANA consideration.............................................18 9. Normative References...........................................18 10. Informative References........................................18 11. Acknowledgments...............................................18 12. Authors' Addresses............................................19 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]. 1. Definitions A set of terminology associated with the ForCES requirements is defined in [FORCES-REQ] and is not copied here. The following list of terminology is relevant to the FE model defined in this document. Datapath -- A conceptual path taken by packets within the forwarding plane, inside an FE. There might exist more than one datapath within an FE. Forwarding Element (FE) Block -- An abstraction of the basic packet processing logical functions in the datapath. It is the building Yang, et. al. Expires May 2003 [Page 2] Internet Draft ForCES FE Functional Model Nov 2002 block of FE functionality. This concept abstracts away implementation details from the parameters of interest for configuration, control and management by CE. Forwarding Element (FE) Stage -- Representation of an FE block instance in a FE's datapath. As a packet flows through an FE along a datapath, it flows through one or multiple distinct stages, with each stage implementing an instance of a certain logical function block. There may be multiple instances of the same functional block in a FE's datapath. 2. Motivation and Requirements of FE model The ForCES architecture allows Forwarding Elements (FEs) of varying functionality to participate in a ForCES network element (NE). The implication of this varying functionality is that CEs can make only minimal assumptions about the functionality provided by its FEs. Before CEs can configure and control the forwarding behavior of FEs, CEs need to query and discover the capabilities and states of their FEs. [FORCES-REQ] mandates that this capabilities and states information be expressed in the form of an FE model, and this model will be used as the basis for CEs to control and manipulate FEs' behavior via ForCES protocol. [FORCES-REQ] describes all the requirements placed on the FE model in detail. We provide a brief summary here to highlight some of the design issues we face. . The FE model MUST express what logical functions can be applied to packets as they pass through an FE. . The FE model MUST be capable of supporting/allowing variations in the way logical functions are implemented on an FE. . The model MUST be capable of describing the order in which these logical functions are applied in a FE. . The FE model SHOULD be extendable and should have provision to express new or vendor specific logical functions. . The FE model SHOULD be able to support minimal set of logical functions that are already identified, such as port functions, forwarding functions, QoS functions, filtering functions, high- touch functions, security functions, vendor-specific functions and off-loaded functions. 3. Capability Model versus State Model Since the motivation of an FE model is to allow the CEs later to control and configure the FEs' behavior via ForCES protocol, it becomes essential to examine and understand what kind of control and configuration the CEs might do to the FEs. It is also equally essential to understand how configurable or programmable FEs are today and will be in the near future. To understand the issue better, it is helpful to make a distinction between two different kinds of FE models û the FE state model and FE capability model. Yang, et. al. Expires May 2003 [Page 3] Internet Draft ForCES FE Functional Model Nov 2002 The FE state model describes the current state of the FE, that is, the instantaneous values or operational behavior of the FE. The FE state model presents the snapshot view of the FE to the CE. On the other hand, the FE capability model describes the configurable capabilities of an FE in terms of variations of functions supported or limitations contained. Conceptually FE capability model presents the many possible states allowed on an FE. The information on the capabilities of the FE helps the CE to make more intelligent decision on the configuration it wants to send to the FE. So the configuration is the desirable state that the FE should be in. Figure 1 shows the concepts of FE state, capabilities and configuration in the context of CE-FE communication via ForCES protocol. +---------+ +---------+ | | FE state: what it is now. | | | |<------------------------------------| | | | | | | CE | FE capabilities: what it can be. | FE | | |<------------------------------------| | | | | | | | FE configuration: what it should be.| | | |------------------------------------>| | +---------+ +---------+ Figure 1. Illustration of FE state, capabilities and configuration in the context of CE-FE communication via ForCES. For example, using the FE state model, an FE may be described to its CE as the following: - on a given port the packets are classified using a given classification filter; - the given classifier results in packets being metered in a certain way, and then marked in a certain way; - the packets coming from specific markers are delivered into a shared queue for handling, while other packets are delivered to a different queue; - a specific scheduler with specific behavior and parameters will service these collected queues. On the other hand, the capability model may describe the FE at the coarsest level such as: - this FE can handle IPv4 and IPv6 forwarding; - this FE can perform classification on the following fields: source IP address, destination IP address, source port number, destination port number, etc; - this FE can perform metering; - this FE can handle up to N queues; - this FE can add and remove encapsulating headers of types including IPSec, GRE, L2TP. Where it gets more complicated is for the capability model to cope with the detailed limits, issues such as how many classifiers the FE Yang, et. al. Expires May 2003 [Page 4] Internet Draft ForCES FE Functional Model Nov 2002 can handle, how many queues, and how many buffer pools the FE can support, how many meters the FE can provide. There is also the issue as how flexibly these various functions can be interconnected within the FE, in another word, how programmable the FE really can be and how the FE capability model can reflect that. While one could try to build an object model for representing capabilities in full, other efforts have found this to be a significant undertaking. A middle of the road approach is to define coarse-grained capabilities and simple capacity measures. Then, if the CE attempts to instruct the FE to set up some specific behavior it is not capable of, the FE will return an error indicating the problem. It is clear that in the context of ForCES, a state model is definitely necessary. The question is how much of the capability model is needed in addition to the state model. A simple state model without any capability flavor will severely limit ForCESÆs ability to take advantage of the flexibility offered by programmable FEs. On the other hand, an all too powerful capability model is difficult to develop and may impose unnecessary overhead for most of the FEs that only offer static functionalities. In order to strike a good balance, it is necessary to examine the kinds of control and configuration that the CEs may do to the FEs. The first kind of control and configuration is the simplest of all. It assumes that the logical functions that an FE supports are already given and the interconnection of these functions remains static in its lifetime. Therefore, the CE can only control FEÆs behavior by manipulating the parameters for each individual function, but it cannot change either the datapath or the functions along each datapath. We call this "static FE" control and configuration. For example, Figure 4 and 6 each show an FE configuration example by representing the processing steps in a directed graph interconnecting all the functional stages that packets can possibly traverse. If such a configuration remains static during FE's lifetime, then all CE can control is the parameters associated with each stage in the graph, for example, the routing table in the LPM forwarder in Figure 4, or the token bucket parameters associated with meter1 in Figure 6. However, the CE cannot reconfigure the graph topology dynamically, such as adding another meter or queue onto the FE in Figure 6 on the fly. For this kind of static control and configuration purpose, the useful FE model should describe how the graph is connected and what are the ôdials and knobsö (i.e., the parameters or attributes) each function allows CE to manipulate. It should also include the statistics and events that FEs can collect and report to CEs. Even for such a ôstatic FEö, some capability model at the individual functions level may be desirable to convey the flexibility of each function. However, a lot of other information may not be necessary, like the packet formats supported between meter1 and counter1 in Figure 6 as Yang, et. al. Expires May 2003 [Page 5] Internet Draft ForCES FE Functional Model Nov 2002 an example, because such information is only useful when the graph can be re-configured dynamically on the fly. The second kind of control and configuration builds on top of the first kind. Using Figure 6 as an example, instead of presenting the static FE graph to the CE, the FE can convey its capabilities to the CE by telling "this FE can support one classifier with up to N filters. This FE can also support up to M meters, X queues, etc." We call this dynamic FE control and configuration. For such dynamic control and configuration, a more powerful and flexible FE capability model is required. For example, it becomes necessary to model not only the capability of the building blocks like classifiers, filters, meters etc., but also the linkage flexibility and constraints between the blocks, so that CE can have the intelligence to build a dynamic FE graph that makes sense. The third level of control and configuration is even more powerful and future looking. In addition to dynamic configuration, CEs might even be allowed to download a given functionality onto FEs at run time. This is similar to the active network concept and so we call it active FE control and configuration. Like active network, this is still considered a research area and is not being considered here. The FE model proposed in this document intends to fully support the static FE control and configuration at the minimum. It is also our intention to allow dynamic FE control and configuration to a certain degree when it makes sense. This FE model currently makes no attempt to address issues beyond the first two kinds of control and configuration scenarios. 4. FE Model This section proposes a ForCES FE model to satisfy all the requirements in [FORCES-REQ] for FE control and configuration. The approach taken is to model the FE datapath(s) and its packet treatment behavior via a directional graph where each node in the graph is an instance of a well-defined logical function block. The FE model defines a generic FE block akin to an abstract base class in object-oriented terminology. The generic FE block contains basic information like block type and textual description of the block function. Based on this generic FE block, a set of well-known FE logical functions are defined with additional state and capability information pertinent to each specific function. A name space is used to associate a unique name or ID with each type of FE block. New logical functions can also be added later to accommodate future innovation in the forwarding plane, as long as the new functions are modeled as an FE block. With such a set of basic building blocks defined, any FE can be modeled by a directional graph where each node is an instance of an FE block, representing a processing stage in the packet datapath. Each node contains information like block name or ID (indicating the block type), stage Yang, et. al. Expires May 2003 [Page 6] Internet Draft ForCES FE Functional Model Nov 2002 ID (local to FE), number of downstream blocks and a list of the stage IDs of those downstream blocks. The rest of this section is devoted to describe the informal data model of FE. The description here is intended to be abstract and conceptual, and examples are used for illustration purpose only. Separate document(s) will serve as specifications by using a formal data modeling language and those specifications should be consistent with the conceptual model described here. 4.1. FE Blocks The generic FE block is the basic building block of the FE model, like an abstract base class in object-oriented terminology. Actual FE logical functions like classifiers, IPv4 forwarders and meters are examples of real FE blocks derived from the generic FE block concept. A well-defined block has a well-defined packet processing behavior and a well-defined set of state and capabilities that CE can potentially configure or control via ForCES. A namespace is needed to specify different types of blocks. The namespace assigns either a unique ID or label to each distinct block type. Such a namespace must be extensible so that new functions can be easily added later. Therefore, the following defines a generic FE Block: - block ID or label which uniquely identifies the block type; - textual description of block function. 4.2. FE Block Library We expect a small set of well-understood FE functional blocks to be defined initially. Such a set of blocks can be viewed as a FE block library. The minimum set of FE functions required in [FORCES-REQ] must be part of this library. It is expected that new FE blocks would be defined and added into this library over time. The actual model for each functional block may differ and contains information pertinent to the semantics of the function itself. However, some general guideline is still useful. For example, typically it is important to specify information such as: - how many inputs it takes and what kinds of packets and meta data it takes for each input; - how many outputs it produces and what kind of packets and meta data it emits for each output; - the packet processing (such as modification) behavior; - what information is programmed into it (e.g., LPM list, next hop list, WRED parameters, etc.) and what parameters among them are configurable; - what statistics it keeps (e.g., drop count, CRC error count, etc.); - what events it can throw (e.g., table miss, port down, etc.). Yang, et. al. Expires May 2003 [Page 7] Internet Draft ForCES FE Functional Model Nov 2002 This document only intends to describe the conceptual FE model and illustrate it with some examples. However, it is not the intention of this document to define any specific block or the library itself. Separate document(s) would be written to do that. The minimum set of FE functions required in [FORCES-REQ] is listed and discussed briefly in the following subsections. The IETF DiffServ (Differentiated Services) and RAP (Resource Allocation Protocols) working groups have done some relevant work in modeling the provisioning policy data for QoS functions and filtering functions. Therefore, we will start our discussion from these related models. 4.2.1. QoS Functions The IETF community has already done some work in modeling the QoS functions in the datapath. The IETF DiffServ working group has defined an informal data model [RFC3290] for QoS-related functions like classification, metering, marking, actions of marking, dropping, counting and multiplexing, queueing, etc. The latest work on DiffServ PIB (Policy Information Base) [DS-PIB] defines a set of provisioning classes to provide policy control of resources implementing the Diferentiated Services Architecture. DiffServ PIB also has an element of capability flavor in it that can potentially enable more dynamic and intelligent configuration of individual functions and the interconnection of the functions. The IETF Policy Framework working group is also defining an informational model [QDDIM] to describe the QoS mechanisms inherent in different network devices, including hosts. This model is intended to be used with the QoS Policy Information Model [QPIM] to model how policies can be defined to manage and configure the QoS mechanisms present in the datapath of devices. Unclassified classified traffic traffic +------------+ | |--> match Filter1 --> OutputA ------->| classifier |--> match Filter2 --> OutputB | |--> no match --> OutputC +------------+ Figure 2. An Example Classifier Using DiffServ Model We use the classifier defined in [RFC3290] as an example to illustrate the DiffServ model. "Classifiers are 1:N (fan-out) devices: they take a single traffic stream as input and generate N logically separate traffic streams as output. Classifiers are parameterized by filters and output streams. Packets from the input stream are sorted into various output streams by filters which match the contents of the packet or possibly match other attributes associated with the packet." To further define filters: "A filter consists of a set of conditions on the component values of a packet's classification key (the header values, contents, and Yang, et. al. Expires May 2003 [Page 8] Internet Draft ForCES FE Functional Model Nov 2002 attributes relevant for classification)." Figure 2 illustrates an example classifier. Based on this conceptual model, [DS-PIB] specifies a classifier of 1:N by N classifier elements. Each classifier element specifies the following: - element ID which identifies the particular output out of N; - classifier instance ID which identifies the classifier instance (all the N classifier elements belong to the same classifier have the same classifier instance ID); - precedence which is an unsigned integer value to represent the relative order in which classifier elements are applied (the classifier element with the highest precedence will be matched first); - next datapath element which provides a pointer to the next function along this branch out of N fan-out; - filter ID which points to the filter used for this branch (Note that filter is defined independent of the classifier and used here as a parameter to the classifier). It is clear from the example above that DiffServ model uses a topological approach to capture the multiple datapath a packet can potentially take. Graphically, a classifier of 1:N has N output branches leading to the next N datapath elements. This has significant implication when we consider the interconnected graph of the functions on FE (see Section 4.3). The alternative is to use an encoded state approach where each packet gets some state information associated with it that indicates the datapath it takes next. For example, using the encoded state approach, a classifier of 1:N may be represented by just one output branch, if all N of the next datapath elements are of the same block function, say, shaper. +----------------+ | Meter-A | | | ----->| In -|-----PM-1---> | | | Out -|-----PM-2---> +----------------+ Figure 3: Meter Followed by Two Preamble Markers The QDDIM model uses the alternative encoded state approach so that information about the treatment that a packet received on an ingress interface is allowed to be communicated along with the packet to the egress interface (see [QDDIM] Section 3.8.3). QDDIM model represents this information transfer in terms of a packet preamble. Figure 3 shows the same example used in [QDDIM] (section 3.8.3) in which meter results are captured in a packet preamble. ôPreamberMarker PM- Yang, et. al. Expires May 2003 [Page 9] Internet Draft ForCES FE Functional Model Nov 2002 1 adds to the packet preamble an indication that the packet exited Meter A as conforming traffic. Similarly, PreambleMarker PM-2 adds to the preambles of packets that come through it indications that they exited Meter A as nonconforming traffic. A PreambleMarker appends its information to whatever is already present in a packet preamble, as opposed to overwriting what is already there.ö ôTo foster interoperability, the basic format of the information captured by a PreambleMarker is specified.ö ôOnce a meter result has been stored in a packet preamble, it is available for any subsequent Classifier to use.ö Section 4.3 has more discussion on the difference between the topological approach (as used by DiffServ model) and the encoded state approach (as used by QDDIM). [DS-PIB] also defines a capability model for classifiers by specifying a bit set to indicate the ability to classify based on IP source address, IP destination address, IP protocol numbers, IP DSCP field, layer 4 port number for UDP and TCP, and Ipv6 flow ID. The capability is thus made known by simply setting the bits accordingly. Similar technique is also used to indicate capabilities of other functions like meters, droppers, etc. While the DiffServ and QDDIM models are not designed with the primary goal of direct machine implementation, we can still use them as our starting point. 4.2.2. Generic Filtering Functions The framework PIB ([FRMWK-PIB]) from the IETF RAP (Resource Allocation Protocol) working group defines four groups of PRCs (Provisioning Classes) that are expected to be common to all clients that provision policy using COPS-PR ([RFC3084]). One of the four PRC groups is classifier group, which contains the Base Filter Class and the other extended filters including the IP Filter, the IEEE 802 Filter and the Internal Label Filter. Even if SPPI ([RFC3159]) is not the final chosen data model for our FE model, it may still be valuable to use the work done here as a starting point for the generic filter functions modeling. 4.2.3. Vendor Specific Functions New and currently unknown FE functionality can be derived (i.e., extended) based on the generic FE Block. The name space used to identify the FE block type must be extensible such that new logical functions can be defined and added later to accommodate future innovation in forwarding plane, as long as the new functions are modeled as an FE block. 4.2.4. Port Functions Yang, et. al. Expires May 2003 [Page 10] Internet Draft ForCES FE Functional Model Nov 2002 Every FE contains a certain number of interfaces (ports), including both the inter-NE interfaces and intra-NE interfaces. The inter-NE interfaces are the external interfaces for the NE to receive/forward packets from/to the external world. The intra-NE interfaces are used for FE-FE or FE-CE communications. Certain types of physical ports have sub-interfaces (frame relay DLCIs, ATM VCs, Ethernet VLans, etc.) as virtual or logical interfaces. Some implementations treat tunnels (e.g., GRE, L2TP, IPSec, MPLS, etc.) as interfaces, while others do not. [FORCES-REQ] treats tunneling as high-touch functions and so FE model does not model tunneling as part of the port functions. Instead, tunneling is covered in Section 4.2.6. Port function expresses: - the number of ports on the FE; - the sub-interfaces if any; - the static attributes of each port (e.g., port type, direction, link speed); - the configurable attributes of each port (e.g., IP address, administrative status); - the statistics collected on each port (e.g., number of packets received); - the current status (up or down). 4.2.5. Forwarding Functions Support for IPv4 and IPv6 unicast and multicast forwarding functions must be provided by the model. Typically, the control plane maintains the Routing Information Base (RIB), which contains all the routes discovered by all the routing protocols with all kinds of attributes relevant to the routes. The forwarding plane uses a different database, the Forwarding Information Base (FIB), which contains only the active subset of those routes (only the best routes chosen for forwarding) with attributes that are only relevant for forwarding. A component in the control plane, termed Route Table Manager (RTM), is responsible to manage the RIB in the CE and maintain the FIB used by the FEs. Therefore, the most important aspect in modeling the forwarding functions is the data model for the FIB. The model also needs to support the possibility of multiple paths. At the very minimum, each route in the FIB needs to contain the following layer-3 information: - the prefix of the destination IP address; - the length of the prefix; - the number of equal-cost multi-path; - the next hop IP address and the egress interface for each path. Yang, et. al. Expires May 2003 [Page 11] Internet Draft ForCES FE Functional Model Nov 2002 Another aspect of the forwarding functions is the method to resolve a next hop destination IP address into the associated media address. There are many ways to resolve Layer 3 to Layer 2 address mapping depending upon link layer. For example, in case of Ethernet links, the Address Resolution Protocol (ARP, defined in RFC 826) is used for IPv4 address resolution. Assuming a separate table is maintained in the FEs for address resolution, the following information is necessary for each address resolution entry: - the next hop IP address; - the media address. Different implementation may have different ways to maintain the FIB and the resolution table. For example, a FIB may consist of two separate tables, one to match the prefix to the next hop and the other to match the next hop to the egress interface. Another implementation may use one table instead. Our model of the forwarding functions should allow such flexibility. 4.2.6. High-Touch Functions High-touch functions are those that take action on the contents or headers of a packet based on content other than what is found in the IP header. Examples of such functions include NAT, ALG, firewall, tunneling and L7 content recognition. The ForCES working group first needs to agree upon a small set of common high-touch functions with well-defined behavior to be included in the initial FE block library. 4.2.7. Security Functions The FE model must be able to describe the types of encryption and/or decryption functions that an FE supports and the associated attributes for such functions. 4.2.8. Off-loaded Functions In addition to the packet processing functions that are typical to find on the FEs, some logical functions may also be executed asynchronously by some FEs, according to a certain finite-state machine, triggered not only by packet events, but by timer events as well. Examples of such functions include finite-state machine execution required by TCP termination or OSPF Hello processing off- loaded from the CE. The FE model must be capable of expressing these asynchronous functions, so that the CE may take advantage of such off-loaded functions on the FEs. The ForCES working group first needs to agree upon a small set of such off-loaded functions with well-understood behavior and interactions with the control plane. Yang, et. al. Expires May 2003 [Page 12] Internet Draft ForCES FE Functional Model Nov 2002 4.3. FE Stage and Directed Graph of FE With a set of basic FE functions defined in the block library, we are ready to model any FEÆs packet processing behavior by a directional graph where each node is an instance of an FE block, representing a processing stage in the packet datapath. This section describes the details behind such a ôdirected graphö FE model. 4.3.1. Basic Concepts An FE stage is simply an instance of an FE block within an FE's datapath. As a packet flows through an FE along a datapath, it flows through one or multiple distinct stages, with each stage instantiating a certain FE logical function. Each FE allocates an FE-unique stage ID to each of its stages and passes the stage ID along with the corresponding block type as part of the FE stage information. This allows multiple instances of the same block present in a FE's datapath. Using NAT as an example, one NAT function is typically performed before the forwarding stage (packets arriving externally have their public addresses replaced with private addresses) and one NAT function is performed after (for packets exiting the domain, their private addresses are replaced by public ones). So there are three stages (NAT, forwarding, and NAT again) in this example datapath, with two NAT instances present in two different stages. A static FE can be modeled by a directed graph interconnecting all the stages present in the FE. Each node in the graph corresponds to a stage. In order to represent the directed interconnection between two consecutive stages along a datapath, each stage contains a ônext stageö pointer that is simply the stage ID of its next stage in the graph. Therefore, the following defines an FE stage (i.e., a node in the FE gragh): - stage identifier which uniquely identifies the node within this FE graph; - block type which identifies the block function that this stage is an instance of; - number of downstream stages which corresponds to the number of downstream nodes connected to this stage; - downstream stage identifiers which corresponds to the set of downstream nodes connected to this stage. With such information defined for each FE stage, it is now possible for CE to query the state of the static FE graph by querying for the initial (ingress) stages of the graph and then traversing the whole graph in a node-by-node fashion. 4.3.2. Topological versus Encoded State Approaches As pointed out in Section 4.2.1, there are potentially two different approaches to model the nodes and the connections between the nodes in the FE graph, namely, the topological approach and the encoded state approach. Yang, et. al. Expires May 2003 [Page 13] Internet Draft ForCES FE Functional Model Nov 2002 +------------+ +------------+ +------------+ input | Ethernet | | | | Ethernet |output ------->| Ingress |-->| IPv4 L3 LPM|-->| Egress |-----> | Port Mgr | | Forwarder | | Port Mgr | +------------+ +------------+ +------------+ {stage ID=1, {stage ID=2, {stage ID=3, type= type= type= Enet-IngP-Mgr, IPv4-L3-LPM-fwd, Enet-EgP-Mgr, #downstream=1, #downstream=1, #downstream=1, downstream={2} downstream={3} downstream=none } } } Figure 4. A simple example of an FE graph using encoded state approach. Input +------------+ +------------+ output ------->|Ingr-Port #1|-->| | +------------+ | | +------------+ ------->|Ingr-Port #2|-->| |-->|EgressPort#1|-----> +------------+ | | +------------+ ------->|Ingr-Port #3|-->|IPv4 L3 LPM |-->|EgressPort#2|-----> +------------+ |Forwarder | +------------+ ------->|Ingr-Port #4|-->| |-->|EgressPort#3|-----> +------------+ | | +------------+ ------->|Ingr-Port #5|-->| |-->|EgressPort#4|-----> +------------+ | | +------------+ ------->|Ingr-Port #6|-->| | +------------+ +------------+ {stage ID=1 {stage ID=7, {stage ID=8, type= type= type= Enet-Ing-port, IPv4-L3-LPM-fwd, Enet-Eg-port, #downstream=1, #downstream=4, #downstream=1, downstream={7} downstream= downstream=none } {8,9,10,11} } . . . } . . . {stage ID=6 {stage ID=11, type= type= Enet-Ing-port, Enet-Eg-port-Mgr, #downstream=1, #downstream=1, downstream={7} downstream=none } } Figure 5. The same example as in Figure 4 using topological approach. Using the topological approach as exemplified by DiffServ model, there are N connections between a fan-out node of 1:N (e.g., a classifier) and its next stages. Using the encoded state approach, Yang, et. al. Expires May 2003 [Page 14] Internet Draft ForCES FE Functional Model Nov 2002 fewer connections are typically needed between the same fan-out node and its next stages, because each packet carries some state information as metadata that the next stage nodes can interpret and invoke different packet treatment. Pure topological approaches can be overly complex to represent because they force on to build elaborate topologies with a lot more connections. An encoded state approach is nicer in that it allows one to simplify the graph and represent the functional blocks with more clarity. But it does require extra metadata to be carried along with the packet, like the preamble in the QDDIM model. For example in Figure 4, stage #2 (IPv4 L3 LPM Forwarder) generates some metadata at its output to carry information on which port the packets should go to, and #3 (Enet-Egress-port-Manager) uses this meta data to direct the packets to the right egress port. Figure 5 shows how the FE graph looks like when using the pure topological approach instead, assuming 6 ingress and 4 egress ports. It is clear that Figure 5 is unwieldy compared to Figure 4. Queue1 +---+ +--+ | A|------------------->| |--+ +->| | | | | | | B|--+ +--+ +--+ +--+ | | +---+ | | | | | | | Meter1 +->| |-->| | | | | | | | | | +--+ +--+ | | Counter1 Absolute Queue2| +--+ +---+ | Dropper1 +--+ +--->|A | | A|---+ | |------>|B | -------->| B|------------------------------>| | +--->|C |------> | C|---+ +--+ | +->|D | | X|-+ | | | +--+ +---+ | | +---+ +---+ Queue3| | Scheduler Classifier1 | | | A|------------>|A | +--+ | | | +->| | | |->| |--+ | | | B|--+ +--+ +->|B | | | | | +---+ | | | | +---+ +--+ | | Meter2 +->| |-+ Mux1 | | | | | | +--+ Queue4 | | Marker1 +--+ | +---------------------------->| |----+ | | +--+ Figure 6. An FE example with multiple datapath. Note that the FE graph can represent largely arbitrary topologies of the stages, regardless which approach (topological or encoded state) is taken. For example, Figure 6 shows an FE implementing QoS functions via a combination of logical functions like classifier, Yang, et. al. Expires May 2003 [Page 15] Internet Draft ForCES FE Functional Model Nov 2002 meter, marker, queue, scheduler, etc. Both approaches are able to represent such an FE graph. The only restrictions on topology relate to the source and sink nature of ingress and egress port functions respectively. For example, egress port functions must not have any downstream stages whereas no other stage may refer to an ingress port function as one of its downstream stages. 4.3.3. Cascading FE Blocks An FE block may contain zero, one or more ingress port stages. Similarly, an FE block may contain zero, one or more egress port stages. In another word, not every FE block has to contain any ingress port or egress port stages. For example, Figure 7 shows two cascading FE blocks. Block #1 contains one ingress port function but no egress port function, while block #2 contains one egress port function but no ingress port function. It is possible to connect these two FE blocks together to achieve the complete ingress-to- egress packet processing function. This provides the flexibility to spread the functions across multiple FEs and interconnect them together later for certain applications. ------------------------------------------------------- | +---------+ +------------+ +---------+ | input| | | | | | output | ---+->| Ingress |-->|Header |-->|IPv4 |---------+--->+ | | port | |Decompressor| |Forwarder| FE | | | +---------+ +------------+ +---------+ Block #1| | ------------------------------------------------------| V | +-----------------------<-----------------------------+ | | |----------------------------------------- V | +------------+ +----------+ | | input | | | | output | +->--+->|Header |-->| Egress |---------+--> | |Compressor | | port | FE | | +------------+ +----------+ Block #2| -----------------------------------------| Figure 7. An example of two different FE blocks connected together. 5. Data Modeling and Representation A formal data modeling language is needed to represent the conceptual FE model described in this document and a full specification will be written using such a data modeling language. It is also necessary to identify a data representation method for over-the-wire transport of the FE model data. The following is a list of some potential candidates for consideration. For the moment, we intend to leave this as an open issue and much debate is needed in the ForCES WG before a decision Yang, et. al. Expires May 2003 [Page 16] Internet Draft ForCES FE Functional Model Nov 2002 can be made. Therefore, we only provide the candidate list and some initial discussion here without drawing a conclusion yet. - XML (Extensible Markup Language) Schema - ASN.1 (Abstract Syntax Notation One) - SMI (Structure of Management Information) [RFC1155] - SPPI (Structure of Policy Provisioning Information) [RFC3159] - UML (Universal Modeling Language) Most of the candidates here, with the notable exception of UML, are capable of representing the model in the document and over-the-wire. Of course, it is also possible to choose one data model language for specification in the document and later allow several over-the-wire representations to map the model into different implementations. XML has the advantage of being human and machine readable with widely available tools support. However, it is very verbose and hence less efficient for over-the-wire transport. It also requires XML parsing functions in both the CE and FE and hence may impose large footprint esp. for FEs. Currently XML is not yet widely deployed and used in network elements. XML for network configuration in general remains an open area that still requires substantial investigation and experiment in IETF. ASN.1 format is human readable and widely used in network protocols. SMI is based on a subset of ASN.1 and used to define Management Information Base (MIB) for SNMP. SPPI is the adapted subset of SMI used to define Policy Information Base (PIB) for COPS. Substantial investment has been made in SMI/MIBs/SNMP by IETF and the Internet community collectively has had many years of design and operation experience with SMI/MIBs/SNMP. However, it is also well recognized that SMI/MIBs/SNMP is not well suited for configuration and so SPPI/PIBs/COPS-PR attempts to optimize for network provisioning and configuration. UML is the software industryÆs standard language for specifying, visualizing, constructing and documenting the artifacts of software systems. It is a powerful tool for data modeling. However, it does not provide a data representation format for over-the-wire transport. 6. Security Considerations The FE model just describes the representation and organization of data sets and attributes in the forwarding plane. The associated communication protocol (i.e., ForCES protocol) will be defined in separate documents and so the security issues will be addressed there. 7. Intellectual Property Right The authors are not aware of any intellectual property right issues pertaining to this document. Yang, et. al. Expires May 2003 [Page 17] Internet Draft ForCES FE Functional Model Nov 2002 8. IANA consideration A namespace is needed to uniquely identify the FE block type for each FE logical function. 9. Normative References [RFC1812] F. Baker, ôRequirements for IP Version 4 Routers", June 1995. [RFC1155] M. Rose, et. al., ôStructure and Identification of Management Informationfor TCP/IP-based Internets", May 1990. [RFC3084] K. Chan, et. al., ôCOPS Usage for Policy Provisioning,ö March 2001. [RFC3159] K. McCloghrie, et. al., ôStructure of Policy Provisioning Information (SPPI)", August 2001. [RFC3290] Y. Bernet, et. al., ôAn Informal Management Model for Diffserv Routersö, May 2002. 10. Informative References [FORCES-REQ] H. Khosravi, et. al., ôRequirements for Separation of IP Control and Forwarding", work in progress, Oct 2002, . [DS-PIB] M. Fine, et. al., ôDifferentiated Services Quality of Service Policy Information Baseö, work in progress, June 2002, . [FRMWK-PIB] M. Fine, et. al., ôFramework Policy Information Baseö, work in progress, June 2002, . [QDDIM] B. Moore, et. al., ôInformation Model for Describing Network Device QoS Datapath Mechanismsö, work in progress, May 2002, . [QPIM] Y. Snir, et. al., ôPolicy Framework QoS Information Modelö, work in progress, Nov 2001,