RFC1246

Network Working Group J. Moy, Editor Request for Comments: 1246 July 1991

                  Experience with the OSPF protocol

Status of this Memo

This memo provides information for the Internet community. It does not specify any Internet standard. Distribution of this memo is unlimited.

Abstract

This is the second of two reports on the OSPF protocol. These reports are required by the IAB/IESG in order for an Internet routing protocol to advance to Draft Standard Status. OSPF is a TCP/IP routing protocol, designed to be used internal to an Autonomous System (in other words, OSPF is an Interior Gateway Protocol).

OSPF is currently designated as a Proposed Standard. Version 1 of the OSPF protocol was published in RFC 1131. Since then OSPF version 2 has been developed. Version 2 has been documented in RFC 1247. The changes between version 1 and version 2 of the OSPF protocol are explained in Appendix F of RFC 1247. It is OSPF Version 2 that is the subject of this report.

This report documents experience with OSPF V2. This includes reports on interoperability testing, field experience, simulations and the current state of OSPF implementations. It also presents a summary of the OSPF Management Information Base (MIB), and a summary of OSPF authentication mechanism.

Please send comments to [email protected].

1.0 Introduction

This document addresses, for OSPF V2, the requirements set forth by the IAB/IESG for an Internet routing protocol to advance to Draft Standard state. This requirements are briefly summarized below. The remaining sections of this report document how OSPF V2 satisfies these requirements:

o The specification for the routing protocol must be well written such

  that independent, interoperable implementations can be developed
  solely based on the specification. For example, it should be possible
  to develop an interoperable implementation without consulting the
  original developers of the routing protocol.

o A Management Information Base (MIB) must be written for the protocol.

  The MIB must be in the standardization process, but does not need to
  be at the same level of standardization as the routing protocol.

o The security architecture of the protocol must be set forth

  explicitly. The security architecture must include mechanisms for
  authenticating routing messages and may include other forms of
  protection.

o Two or more interoperable implementations must exist. At least two

  must be written independently.

o There must be evidence that all features of the protocol have been

  tested, running between at least two implementations. This must
  include that all of the security features have been demonstrated to
  operate, and that the mechanisms defined in the protocol actually
  provide the intended protection.

o There must be significant operational experience. This must include

  running in a moderate number routers configured in a moderately
  complex topology, and must be part of the operational Internet. All
  significant features of the protocol must be exercised. In the case
  of an Interior Gateway Protocol (IGP), both interior and exterior
  routes must be carried (unless another mechanism is provided for the
  exterior routes). In the case of a Exterior Gateway Protocol (EGP),
  it must carry the full complement of exterior routes.

This report is a compilation of information obtained from many people. The reader is referred to specific people when more information on a subject is available. People references are gathered into Section 10.0, in a format similar to that used in [4].

1.1 Acknowledgments

The OSPF protocol has been developed by the OSPF Working Group of the Internet Engineering Task Force. Many people have contributed to this report. They are listed in Section 10.0 of this report.

2.0 Documentation

Version 1 of the OSPF protocol is documented in RFC 1131 [1]. OSPF Version 2, which supersedes Version 1, has been documented in RFC 1247 [2]. The differences between OSPF Version 1 and Version 2 are relatively minor, and are listed in Appendix F of RFC 1247 [2]. All information presented in this report concerns OSPF V2 unless explicitly mentioned otherwise.

The OSPF protocol was developed by the OSPF Working Group of the Internet Engineering Task Force. This Working Group has a mailing list, [email protected], where discussions of protocol features and operation are held. The OSPF Working Group also meets during the quarterly Internet Engineering Task Force conferences. Reports of these meeting are published in the IETF's Proceedings. In addition, two reports on the OSPF protocol have been presented to the IETF plenary (see "Everything You Ever Wanted to Know about OSPFIGP" in [5] and "OSPF Update" in [6]).

The OSPF protocol began undergoing field trials in Spring of 1990. A mailing list, [email protected], was formed to discuss how the field trials were proceeding. This mailing list is maintained by Ross Veach of the University of Illinois [rrv]. Archives of this list are also available. There has been quite a bit of discussion on the list concerning OSPF/RIP/EGP interaction.

A OSPF V2 Management Information Base has also been developed and published in [3]. For more information, see Section 3.0 of this report.

There is a free implementation of OSPF available from the University of Maryland. This implementation was written by Rob Coltun [rcoltun]. Contact Rob for details.

3.0 MIB

An OSPF Management Information Base has been published in RFC 1248 [3]. The MIB was written by Rob Coltun [rcoltun] and Fred Baker [fbaker]. The OSPF MIB appears on the mgmt subtree as SMI standard-mib 13.

The OSPF MIB was originally developed by Rob Coltun of the University of Maryland, under contract to Advanced Computer Communications. A subset of his proposal was implemented to facilitate their development, and represents operational experience of a sort.

The MIB consists of a general variables group and ten tables:

TABLE 1. OSPF MIB Organization

   Group Name           Description
   ________________________________________________________________
   ospfGeneralGroup     General Global Variables
   ospfAreaTable        Area Descriptions
   ospfStubAreaTable    Default Metrics, by Type of Service
   ospfLsdbTable        Link State Database
   ospfAreaRangeTable   Address Range Specifications
   ospfHostTable        Directly connected Hosts
   ospfIfTable          OSPF Interface Variables
   ospfIfMetricTable    Interface Metrics, by Type of Service
   ospfVirtIfTable      Virtual Links
   ospfNbrTable         (Non-virtual) OSPF Neighbors
   ospfVirtNbrTable     Virtual OSPF Neighbors

As MIBs go, the OSPF MIB is quite large; 105 objects. The following are some statistics describing the distribution of the MIB's variables:

o 11 define the above Group and Tables

o 10 define the Entry in a Table

o 7 are Counters

o 6 are Gauges

o 68 objects mandated by the OSPF Version 2 Specification

Section D.2 of the OSPF V2 specification [2] lists a set of required statistics that an implementation must maintain. These statistics have been incorporated into the OSPF MIB. The MIB's thirteen Counters and Gauges enable evaluation of the OSPF protocol's performance in an operational environment. Most of the remainder of the MIB's variables parameterize the many features that OSPF provides the network administrator.

For more information on the MIB contact Fred Baker [fbaker].

4.0 Security architecture

In OSPF, all protocol packet exchanges are authenticated. The OSPF packet header (which is common to all OSPF packets) contains a 16-bit Authentication type field, and 64-bits of Authentication data. Each particular OSPF area must run a single authentication scheme, as indicated by the Authentication type field. However, authentication keys can be configured by the system administrator on a per-network basis.

When an OSPF packet is received from a network, the OSPF router first verifies that it indicates the correct Authentication type. The router then authenticates the packet, running a verification algorithm using the configured authentication key, the 64-bits of Authentication data and the rest of the OSPF packet data as input. The precise algorithm used is dictated by the Authentication type. Packets failing the authentication algorithm are dropped, and the authentication failure is noted in a MIB-accessible variable (see [3]).

There are currently few Authentication types in use. The current assignments are:

TABLE 2. Current OSPF Authentication types.

 Type code   Algorithm
 ____________________________________________________________________
 0           No authentication performed.
 1           Simple (clear) password.
 2-255       Reserved for assignment by the IANA ([email protected])
 > 255       Available for local (per-AS) definition.

For more information on OSPF's authentication procedures, see Sections 8.1, 8.2, and Appendix E of [2].

5.0 Implementations

The are multiple, interoperable implementations of OSPF currently available. This section gives a brief overview of the five implementations that have participated in at least one round of interoperability testing. (For a detailed discussion of OSPF interoperability testing, see Section 7.0 of this report.) Other implementations do exist, but because of commercial realities (e.g., the product is not yet announced) they unfortunately cannot be listed here.

The five implementations that have participated in the OSPF interoperability testing are (listed in alphabetical order):

o 3com. This implementation was wholly developed by 3com. It has

  participated in all three rounds of interoperability testing. It is
  also the only implementation of OSPF's TOS routing..  The 3com
  implementation consists of approximately 9000 lines of C code,
  including comments but excluding user interface and MIB code.
  Consult Dino Farinacci [dino] for more details.

o ACC. This implementation is based on the University of Maryland code.

  It participated in the last two rounds of interoperability testing.
  It also contains the only implementation of (a precursor to) the OSPF
  MIB (see Section 3.0 for details), which it uses for monitoring and
  configuration. The ACC implementation consists of approximately
  24,000 lines of C code, including its OSPF MIB code. Consult Fred
  Baker [fbaker] for more details.

o Proteon. This implementation was wholly developed by Proteon. It has

  participated in all three rounds of interoperability testing. It is
  also the only implementation that has a significant amount of field
  experience (see Section 6.0 for details). The Proteon implementation
  consists of approximately 9500 lines of C code, including comments
  but excluding user interface code.  Consult John Moy [jmoy] for more
  details.

o Wellfleet. This implementation has participated in all three rounds

  of interoperability testing.  Consult Jonathan Hsu [jhsu] for more
  details.

o University of Maryland. This implementation was developed wholly by

  Rob Coltun at the University of Maryland. It has formed the basis for
  a number of commercial OSPF implementations, and also participated in
  the latest round of interoperability testing. The University of
  Maryland implementation consists of approximately 10,000 lines of C
  code. Consult Rob Coltun [rcoltun] for more details.

Note that, as required by the IAB/IESG for Draft Standard status, there are multiple interoperable independent implementations, namely those from 3com, Proteon and the University of Maryland.

6.0 Operational experience

This section discusses operational experience with the OSPF protocol. Version 1 of the OSPF protocol began to be deployed in the Internet in Spring of 1990. The results of this original deployment were reported to the mailing list [email protected]. (Archives of this mailing list are available from Ross Veach [rrv].) No protocol bugs were found in this first deployment, although several additional features were found to be desirable. These new features were added to the protocol in OSPF Version 2.

The OSPF protocol is now deployed in a number of places in the Internet. In this section we focus on three highly visible systems, namely the NASA Sciences Internet, BARRNet and OARnet. The dimensions of these three OSPF systems is summarized in the following table:

TABLE 3. Three operational OSPF deployments

        Name      Version 1 date   Version 2 date   # routers
        _____________________________________________________
        NSI       4/13/90          1/1/91           15
        BARRNet   4/90             11/90            14
        OARnet    10/15/90         not yet          13

All the above deployments are using the Proteon OSPF implementation. There is one other deployment worth mentioning in this context. 3com has started to deploy OSPF on their corporate network. They have 8 of their routers running OSPF (the 3com implementation), and are planning on cutting over the remaining routers (20 in all). Currently they have two operational routers running OSPF and RIP simultaneously. One converts OSPF data to RIP data, and the other RIP data to OSPF data. For more details, contact Dino Farinacci [dino].

6.1 NSI

The NASA Science Internet (NSI) is a multiprotocol network, currently supporting both DECnet and TCP/IP protocols. NSI's mission is to provide reliable high-speed communications to the NASA science community. The NASA Science Internet connects with other national networks including the National Science Foundation's NSFNET, the Department of Energy's ESnet and the Department of Defense's MILNET. NSI also has international connections to Japan, Australia, New Zealand, Chile and several European countries.

For more information on NSI, contact Jeffrey Burgan [jeff] or Milo Medin [medin].

6.1.1 NSI's OSPF system

NSI was one of the initial deployment sites for OSPF Version 1, having deployed the protocol in April 1990. NSI has been running OSPF V2 since 1/1/91. They currently have 15 routers in their OSPF system. This system is pictured in Figure 1. It consists of a nationwide collection of serial lines, with ethernets at hub sites. The numbers associated to interfaces/links in Figure1 are the associated OSPF costs. Note that certain links have been weighted so that they are less preferable than others.

Many of NSI's OSPF routers are speaking either RIP and/or EGP as well as OSPF. Routes from these other routing protocols are selectively imported

into their OSPF system as externals. The current number of imported externals is 496.

All NSI externals are imported as OSPF type 2 metrics. In addition, NSI uses the OSPF external route tag to manage the readvertisement of external routes. For example, a route learned at one edge of the NSI system via EGP can be tagged with the number of the AS from which it was learned. Then, as the OSPF external LSA describing this route is flooded through the OSPF system, this tagging information is distributed to all the other AS boundary routers. A router on the other edge of the NSI can then say that it wants to readvertise (via EGP) routes learned from one particular AS but not routes learned from another AS. This allows NSI to implement transit policies at the granularity of Autonomous Systems, instead of network numbers, which greatly reduces the network's configuration burden.

NSI has also experimented with OSPF stub areas, in order to support routers having a small amount of memory.

6.1.2 NSI - Deployment analysis

NSI ran a couple of experiments after OSPF's deployment to test OSPF's convergence time in the face of network failures, and to compare the level of routing traffic in OSPF with the level of routing traffic in RIP. These experiments were included in NSI status reports to the OSPF plenary.

The first experiment consisted of running a continuous ICMP ping, and then bringing down one of the links in the ping packet's path. They then timed how long it took OSPF to find an alternate path, by noticing when the pings resumed. The result of this experiment is contained in Milo Medin's "NASA Sciences Internet Report" in [8]. It shows that the interrupted ping resumed in three seconds.

The second experiment consisted in analyzing the amount of routing protocol traffic that flow over an NSI link. One of the NSI links was installed, but did not have any active users yet. For this reason, all traffic that flowed over the link was routing protocol traffic. The link was instrumented to continuously measure the amount of bandwidth consumed, first in the case where RIP was running, and then in the case of where OSPF was running. The result is shown graphically in Jeffrey Burgan's "NASA Sciences Internet" report in [9]. It shows that OSPF consumes many times less network bandwidth than RIP.

6.2 BARRNet

BARRNet is the NSFNet regional network in Northern California. At the present time, it serves approximately 80 member sites in an area stretching from Sacramento in the north-east to Monterey in the in the south-west. Sites are connected to the network at speeds from 9.6Kbps to full T1 using Proteon and cisco routers as well as a Xylogics terminal server. The membership is composed of a mix of university, government, and commercial organizations. BARRNet has interconnections to the NSFNet (peering with both T1 and T3 backbones at Stanford University), ESNet (peering at LLNL, with additional multi-homed sites at LBL, SLAC, and NASA Ames), and DDN national networks (peering at the FIX network at NASA Ames), and to the statewide networks of the University of California (peering at U.C. Berkeley) and the California State University system (peering at San Francisco State and Sacramento State).

Topologically, the network consists of fourteen OSPF-speaking Proteon routers, which as a "core", with six of these redundantly connected into a ring. All "core" sites are interconnected via full T1 circuits. Other member sites attach as "stub" connections to the "core" sites. The bulk of these are connected in a "star" configuration at Stanford University, with lesser numbers at other "core" sites.

Contact Vince Fuller [vaf] for more information on BARRNet.

6.2.1 BARRNet's OSPF system

BARRNet was also one of the initial deployment sites for OSPF Version 1, having deployed the protocol in April 1990. BARRNet has been running OSPF V2 since November 1990. They currently have 14 routers in their OSPF system. The BARRNet OSPF system is pictured in Figure 2. It consists of a collection of T1 serial lines, with ethernets at hub sites.

Most of BARRNet's OSPF routers are speaking either RIP and/or EGP as well as OSPF. Routes from these other routing protocols are selectively imported into their OSPF system as externals. A large number of external routes are imported; the current number is1816. The bulk of these are the T1 NSFNet routes, followed by several hundred NSN routes, around 60-80 BARRNet routes from the non-OSPF system, and several dozen from ESNet.

All external routes are imported into the BARRNet system as external type 1 metrics. In addition, BARRnet, like NSI, uses the OSPF's external route tagging feature to help manage the readvertisement of external routes via EGP.

BARRnet is also using four stub OSPF areas in order to collapse subnet information. These stub areas all consist of a single LAN. They do not contain any OSPF routers in their interiors.

6.2.2 BARRNet - Deployment analysis

Initial deployment of OSPF Version 1 in BARRNet pointed to the need for two new protocol features that were added to OSPF V2, namely:

o Addition of the forwarding address to OSPF external LSAs. This

  eliminated the extra hops that were being taken in BARRNet when only
  routers BR5 and BR6 were exchanging EGP information with the NSS (see
  Figure 2). Without the forwarding address feature, that meant that
  NSFNet traffic handled by routers BR10, BR16 and BR28 was taking an
  extra hop to get to the NSS.

o Addition of stub areas. This was an attempt to get OSPF running on

  some of the BARRNet routers that had insufficient memory to deal with
  all of BARRNet's external routes.

6.3 OARnet

OARnet, the Ohio Academic Resources Network, is the regional network for the state of Ohio. It serves the entire higher education community, providing Ohio schools access to colleagues worldwide. The Ohio Supercomputer Center and the NSF Supercomputer Centers are reached through OARnet. Libraries, databases, national and international laboratories and research centers are accessible to faculty, helping make Ohio schools competitive.

OARnet was established in 1987 to provide state-wide access to the CRAY at the Ohio Supercomputer Center in Columbus, Ohio. Since then it has evolved into a network supporting all aspects of higher education within Ohio. A primary goal of OARnet is to facilitate collaborative projects and sharing of resources between institutions, including those outside the state. OARnet connections are available to Ohio academic institutions and corporations engaged in research, product development, or instruction. Colleges, universities, and industries currently use OARnet connections to communicate within the state and with colleagues around the country.

OARnet uses the Internet (TCP/IP) and DECNET protocols. OARnet participants using TP/IP protocols are connected to the worldwide Internet, which includes all the major networks open to non-classified research. OARnet is also connected to NSFNet, the national research and

education network sponsored by the National Science Foundation. It has gateways to BITNET, CSNET, CICNet (a network connecting the Big Ten universities), and the NASA Science Internet.

For more information on OARnet, contact Kannan Varadhan [kannan].

6.3.1 OARnet's OSPF system

OARnet has been running OSPF Version 1 since October 15, 1990. They currently have 14 routers in their OSPF system. The OARnet OSPF system is pictured in Figure 3.

There are 29 sites connected directly to the OARnet backbone. All 13 of OARnet's OSPF routers act as ASBRs. There are 40 OSPF internal routes on OARnet's network, and they import about 120 routes from RIP. OARnet runs EGP on the DMZnet at Columbus, which connects them to CICNet. The router connecting OARnet to DMZnet (OAR1 in Figure 3) runs EGP on the DMZnet side, and OSPF and RIP on the OARnet backbone. No EGP routes are imported into the OSPF system. The OAR1 router is configured to generate a default when EGP routes are available. The OAR1 router is the keystone for routing on OARnet's network, in that it acts as an intermediary for all of OARnet's RIP centric routers.

OARnet uses the Event Logging System on its Proteon routers to generate traps for "interesting" events related to routing. They have these traps sent to an SNMP management station, where the logs are collected for later perusal.

6.3.2 OARnet - Deployment analysis

OARnet is monitoring their OSPF system via collection of traps on their SNMP management station. The following is a report on their observations. It has been edited slightly to conform better with the other text and maps presented in this report. For more information, contact Kannan Varadhan [kannan]:

3 of our 10 DS1 circuits are on digital microwave, and these tend to flap occasionally. Our observations indicate that the routers bring up links, and adjacencies, on average, in about 2 seconds. Routes fallback to alternate backup paths instantly. Whole blocks of routes cut over the instant the adjacencies are formed.

In contrast to this, our RIP routes would take about 3-6 minutes to cutover, and, on occasion, would not cut back to the preferred paths. This was our prime motivation in switching to OSPF.

We attempted to duplicate Milo Medin's ping test to dramatically illustrate the performance of RIP over OSPF. To do this, we selected a host on the farthest point from our workstation, and ran a continuous ping to it. We would then bring down a primary DS1 circuit, and watch the time it took to switch to the fallback route. Following this, we would bring the circuit back up, and study the time it took to re-sync to the new path. With RIP, we were unable to fully complete the experiment, because the farthest point was exactly equal to the new (and preferred) primary path, and therefore, RIP would never choose it on it's own, until the path it was currently using failed. With OSPF, it took about 2 seconds to synchronize over a new, much slower 56kb path, and less than a second when the DS1 circuit came back up.

Here are some more observations of the OARnet OSPF system's behavior:

o 131.187.36.0 is the 56kb line to Kent State University. Kent also has

  a DS1 circuit leading into ASP, the Akron Pop. Likewise, UAkron.edu
  has a similar configuration. A roundabout backup path exists when
  traffic heads up to Cleveland over a couple of DS1 circuits, and then
  down a 56kb backup path used by another school in the Cleveland area.

  Some statistical information:

          1. 09:55:17: SPF.37: new route to Net 131.187.36.5,
                       type SPF cost 32
          2. 09:55:18: SPF.37: new route to Net 131.187.36.6,
                       type SPF cost 22
          3. 09:55:20: SPF.21: State Change, nbr 131.187.27.6,
                       new state <Full>, event 9
          4. 09:55:21: SPF.37: new route to Net 131.187.36.5,
                       type SPF cost 31
          5. 09:55:22: SPF.37: new route to Net 131.187.36.6,
                       type SPF cost 21
          6. 09:55:28: SPF.21: State Change, nbr 131.187.21.5,
                       new state <Full>, event 9
          7. 09:55:29: SPF.21: State Change, nbr 131.187.51.6,
                       new state <Full>, event 9
          8. 09:55:31: SPF.37: new route to Net 131.187.36.5,
                       type SPF cost 22
          9. 09:55:33: SPF.37: new route to Net 131.187.36.5,
                       type SPF cost 11

  The Akron router restarts, and has to re-sync with all the lines.
  This restart is confirmed when one looks at the traps from gwCSP1.
  Traps from gwASP1 still do not get through to us, because traps are

  sent via UDP, and gwASP1's routing tables are not fully configured
  yet.

  Events 1 and 2 are route changes routing traffic via Cleveland,
  across 2 DS1 circuits and a 56kb line.

  When the DS1 circuit to UAkron came up, routes instantly cut over to
  use this as a better least cost path. This is shown in events 3, 4
  and 5.

  In a few seconds, the line to Columbus is the next one up. This is
  event 6. Event 8 relates to this cutover, and is the best path yet.
  When the DS1 circuit to Kent is up, the link is used instantly.

  We are able to make such a definitive conclusion of these traps on
  the basis of the topological information that we have about the
  network and the means used to monitor them.

o To illustrate the time required to fully synchronize a database, we

  piece together a few adjacency forming traces...

  Please bear in mind that these time stamps are the time stamps on the
  management station, and are not to be taken as the absolute truth.
  Things we haven't taken into account are        transit times of
  messages,       ordering of events (SNMP traps are sent using UDP),
  loss of event reports (recall that an entire synchronization sequence
  of gwASP1 on the ASP-CSP link is missing),     etc.

  The trace below corresponds to the Akron router, gwASP1 bring up the
  link in the previous section. This is as observed on the other end of
  the line, gwCSP1.

          REPORT DATE: 02/26/91   ROUTER: gwcsp1
          09:55:06: SPF.15: State Change, ifc 131.187.22.6,
                     new state <Point-To-Point>, event 1
          09:55:06: GW.xxx: Link Up Trap: 09:55:07: SPF.37:
                    new route to Net 131.187.22.5, type SPF cost 1
          09:55:07: SPF.21: State Change, nbr 131.187.22.5,
                    new state <Init>, event 1
          09:55:09: SPF.37: new route to Net 131.187.27.5,
                    type SPF cost 22
          09:55:11: SPF.21: State Change, nbr 131.187.22.5,
                    new state <ExStart>, event 14
          09:55:11: SPF.21: State Change, nbr 131.187.22.5,
                    new state <2-Way>, event 3
          09:55:12: SPF.21: State Change, nbr 131.187.22.5,
                    new state <Exchange>, event 5

          09:55:12: SPF.21: State Change, nbr 131.187.22.5,
                    new state <Full>, event 9
          09:55:12: SPF.21: State Change, nbr 131.187.22.5,
                    new state <Loading>, event 6

  Below, is another trace of the same router restart sequence, where
  the router is proceeding to bring up other DS1 circuits. Bringing up
  the first adjacency took about 5 seconds. Subsequent adjacencies take
  the router less than a second as seen below.

          REPORT DATE: 02/26/91   ROUTER: gwasp1
          09:55:20: SPF.15: State Change, ifc 131.187.27.5,
                    new state <Point-To-Point>, event 1
          09:55:20: GW.xxx: Link Up Trap: 09:55:20: SPF.21:
                    State Change, nbr 131.187.27.6, new state <Init>, event 1
          09:55:20: SPF.21: State Change, nbr 131.187.27.6,
                    new state <ExStart>, event 14
          09:55:20: SPF.21: State Change, nbr 131.187.27.6,
                    new state <Exchange>, event 5
          09:55:20: SPF.21: State Change, nbr 131.187.27.6,
                    new state <Full>, event 9
          09:55:21: SPF.21: State Change, nbr 131.187.27.6,
                    new state <Loading>, event 6
          09:55:24: SPF.21: State Change, nbr 131.187.51.6,
                    new state <Init>, event 1
          09:55:24: SPF.21: State Change, nbr 131.187.21.5,
                    new state <Init>, event 1
          09:55:25: SPF.37: new route to Net 131.187.21.6, type SPF cost 13
          09:55:25: SPF.37: new route to Net 131.187.51.5, type SPF cost 22
          09:55:28: SPF.21: State Change, nbr 131.187.21.5,
                    new state <ExStart>, event 14
          09:55:28: SPF.21: State Change, nbr 131.187.21.5,
                    new state <2-Way>, event 3
          09:55:28: SPF.21: State Change, nbr 131.187.21.5,
                    new state <Exchange>, event 5
          09:55:28: SPF.21: State Change, nbr 131.187.21.5,
                    new state <Full>, event 9
          09:55:28: SPF.21: State Change, nbr 131.187.21.5,
                    new state <Loading>, event 6
          09:55:29: SPF.37: new route to Net 131.187.51.6, type SPF cost 1
          09:55:29: SPF.37: new route to Net 131.187.21.5, type SPF cost 1
          09:55:29: SPF.21: State Change, nbr 131.187.51.6,
                    new state <Exchange>, event 5
          09:55:29: SPF.21: State Change, nbr 131.187.51.6,
                    new state <ExStart>, event 14
          09:55:29: SPF.21: State Change, nbr 131.187.51.6,
                    new state <2-Way>, event 3
          09:55:29: SPF.21: State Change, nbr 131.187.51.6,

                    new state <Full>, event 9
          09:55:29: SPF.21: State Change, nbr 131.187.51.6,
                    new state <Loading>, event 6

  A transient fault on a DS1 circuit, causes the line to flap. All
  routers quickly reroute around the flap, and the router itself takes
  about 2 seconds to bring up the adjacency once more.

          REPORT DATE: 02/26/91   ROUTER: gwasp1
          14:33:43: GW.xxx: Link Up Trap:
          14:34:19: SPF.15: State Change, ifc 131.187.22.5,
                    new state <Down>, event 7
          14:34:19: GW.xxx: Link Failure Trap:
          14:34:19: SPF.47: Net 131.187.22.6 now unreachable
          14:34:36: SPF.15: State Change, ifc 131.187.22.5,
                    new state <Point-To-Point>, event 1
          14:34:36: GW.xxx: Link Up Trap:
          14:34:37: SPF.37: new route to Net 131.187.22.6, type SPF cost 1
          14:34:45: SPF.21: State Change, nbr 131.187.22.6,
                    new state <2-Way>, event 3
          14:34:45: SPF.21: State Change, nbr 131.187.22.6,
                    new state <Init>, event 1
          14:34:46: SPF.21: State Change, nbr 131.187.22.6,
                    new state <ExStart>, event 14
          14:34:46: SPF.21: State Change, nbr 131.187.22.6,
                    new state <Exchange>, event 5
          14:34:47: SPF.21: State Change, nbr 131.187.22.6,
                    new state <Full>, event 9
          14:34:47: SPF.21: State Change, nbr 131.187.22.6,
                    new state <Loading>, event 6

o On the amount of time it takes for a router to restart, and become

  fully synchronized. Taking the logs in the previous instance, we
  notice that the CSP-ASP link comes up at 9:55:06. The last link is
  observed to be up at 9:55:29, which is less than a minute.

o On the RIP equivalent of the tests, it took us 3 minutes to cutover

  to the slower speed fallback route, and we lost countless many
  packets.  The routes never cutover to the higher speed paths when
  available, and we waited well over 30 minutes watching this,
  wondering why. Unfortu- nately, at this point, we seem to have lost
  the RIP statistics.

  On the OSPF version, we have...

          {nisca danw 51}
          ping 131.187.25.6 PING 131.187.25.6 (131.187.25.6):

          56 data bytes 64 bytes from 131.187.25.6:
          icmp seq=0 ttl(255-ttl)=54(201). time=20 ms
                  [...]
          icmp seq=10 ttl(255-ttl)=54(201). time=20 ms
                   ||             T1 down
          icmp seq=14 ttl(255-ttl)=54(201). time=180 ms
          icmp seq=15 ttl(255-ttl)=54(201). time=60 ms
                  [...]
          icmp seq=38 ttl(255-ttl)=8(247). time=1300 ms
          icmp seq=39 ttl(255-ttl)=54(201). time=820 ms
                   ||             Tl Up
          icmp seq=40 ttl(255-ttl)=54(201). time=20 ms
          icmp seq=41 ttl(255-ttl)=54(201). time=20 ms
          131.187.25.6 PING Statistics
          51 packets transmitted, 48 packets received, 5% packet loss
          round-trip (ms) min/avg/max = 20/277/1300

6.4 Features exercised during operational deployment

In operational environments, all basic mechanisms of the OSPF protocol have been exercised. These mechanisms include:

o Designated Router election. There have been operational deployments

  have as many as 8 OSPF routers attached to a single broadcast
  network.

o Database synchronization. This includes OSPF's adjacency bringup and

  reliable flooding procedures. Large operational OSPF link state
  databases (e.g., BARRNet) have provided a thorough test of these
  mechanisms.

o Flushing advertisements. The procedure for flushing old or

  unreachable advertisements (the MaxAge procedure) has been tested
  operationally.  It is interesting to note that flushing of
  advertisements does occur more during interoperability testing
  (because of the constant restart- ing of routers) that it does
  operationally. For example, in a week of BARRNet statistics, 9650
  advertisements were flushed, while 688,278 new advertisements were
  flooded.

o Import of external routes. All options of external LSAs have been

  tested operationally: type 1 metrics, type 2 metrics, forwarding
  addresses and the external route tag.

o Authentication. The OSPF authentication procedure has been tested

  operationally.

o Equal-cost multipath. Operational deployments have included

  topologies with equal-cost, redundant paths.

o Stub areas. These have been deployed both in BARRNet and NSI.

6.5 Limitations of operational deployments

The following things have not been tested in an operational environment:

o Multi-vendor deployments. So far all deployments have used a single

  implementation. However, extensive interoperability testing of OSPF
  has been done (see Section 7.0 of this report).

o Regular OSPF areas. These have however been tested in all three

  rounds of the OSPF interoperability testing.

o Virtual links. These have however been tested in OSPF's

  interoperability testing.

o Non-broadcast networks. However, OSPF interoperability testing has

  been performed over X.25 networks.

o TOS routing. However, this has been tested in OSPF's interoperability

  testing.

6.6 Conclusions

All basic features of the OSPF protocol have been exercised. Very large OSPF link state databases (e.g., BARRNet's OSPF system) have been deployed, providing a thorough test of OSPF's database synchronization mechanisms. No OSPF protocol problems have been found in operational deployments.

Most of the hassles in operation deployments has to do with the OSPF/RIP interchange. Many of these issues have been ironed out on the ospf-tests mailing list (see Section 2.0). However, the interaction between OSPF, RIP, and EGP continues to be an active area of research.

7.0 Interoperability Testing

There have been three separate OSPF V2 interoperability testing sessions. Five separate implementations have participated in at least one session: implementations from the companies 3com, ACC, Proteon and Wellfleet, and the publicly available implementation from the University of Maryland.

Each of the testing sessions is described in a succeeding section. For each session, the participants are identified, and the testing topologies are described along with the particular protocol features that were exercised. Any protocol problems that were encountered during the testing are also described. In addition, for the second and third rounds testing reports were sent to the ospf mailing lists. These reports are reproduced in this document.

There is quite a bit of commonality in the features that have been tested from session to session. There are several reasons for this commonality. First, in each testing session an attempt has been made to increase the size of the OSPF system under test. For example, the number of external routes imported has doubled each session. Secondly, the interoperability sessions have been debugging sessions as well as protocol sessions. Many things tested in the third round were to ver- ify that implementations had successfully fixed problems found in earlier sessions. A brief overview of the testing session is presented in the following table:

TABLE 4. OSPF interoperability testing at a glance.

Site Week Routers Externals Implementations _____________________________________________________________________________ Proteon 9/25/90 6 20-30 3com, Proteon, Wellfleet SURAnet 12/17/90 10 96 3com, ACC, Proteon, Wellfleet 3com 2/4/91 16 400 3com, ACC, Proteon, Wellfleet, UMD

For more information on the interoperability testing, the following people can be contacted: Fred Baker [fbaker], Rob Coltun [rcoltun], Dino Farinacci [dino], Jonathan Hsu [jhsu], John Moy [jmoy], and William Streilein [bstreile].

7.1 Testing methodology

In the interoperability tests, the routers have been interconnected using ethernet, serial lines (PPP and proprietary), X.25 and 802.5 token ring. Monitoring of the routers has been done through connecting terminals (either directly or via telnet) to the router consoles. Each implementation has a different user interface, which makes monitoring somewhat difficult. As explained earlier in this document, there is now an OSPF MIB, which in the future will enable a common monitoring interface to all implementations.

In general, each implementation has an error logging capability, and this is often how problems are discovered. LAN protocol analyzers are

also used to capture OSPF protocol packet exchanges that are causing problems. These packet traces are available for analysis either during of after the testing sessions.

Verification of routing was done through visual inspection of implementations' routing table and link state databases (via the console interface), and through network debugging tools such as "ping" and "traceroute".

7.2 First round (Proteon, 9/25/90 - 9/29/90)

The first round of OSPF protocol testing took place at Proteon Inc.'s Westborough facility, the week of September 25, 1990. Three implementations participated, from the vendors 3com, Proteon and Wellfleet.

There were two 3com routers, two Wellfleet routers and two Proteon routers available for testing. These routers were interconnected with ethernets and serial lines. External routes were imported from the Proteon company internet. In addition, during off hours we were able to connect the routers under test to the Proteon company internet, forming one fairly large OSPF system.

The testing at Proteon proceeded as follows:

o All routers were connected to a single ethernet. Then, as routers

  were taken up and down, the Designated Router election algorithm and
  the Database Description process were tested. Also OSPF's reliable
  flooding algorithm was tested in this configuration.

o Twenty to thirty external routes were imported into the OSPF system

  by a Proteon router (which was simultaneously running RIP). It was
  then verified that these external routes were installed into the
  router's routing tables.

o One of the 3com routers was configured to originate an OSPF default

  route. This tested OSPF default route processing, and also tested the
  behavior of the system when multiple routers were importing external
  routes.

o The OSPF system was split into areas. Both regular OSPF areas (non-

  stub) and stub areas were tested.

o The six routers under test were connected to the Proteon company

  internet (which was also running OSPF), forming an OSPF system of
  eighteen routers. This configuration was shortlived, due to a
  disagreement between the 3com and Proteon routers concerning how to

  represent an OSPF default route.

Unfortunately, incomplete records were kept of this testing, so that no maps of the testing configurations can be reproduced for this document.

7.2.1 Problems found in the First round testing

A couple of OSPF protocol/specification problems were uncovered in the first round of testing. First, it was noticed that there was a window in the Database Description process where concurrently flooded MaxAge advertisements could prevent database synchronization from completing. This required a change in the specification's handling of MaxAge LSAs.

Secondly, it was found that the OSPF specification did not specify how the Network Mask field should be set in external LSAs that were advertising the DefaultDestination. This was a minor problem, but caused difficulties because of assumptions made in one implementation on how the mask should be set.

7.3 Second round (SURAnet, 12/17/90 - 12/21/90)

The second round of OSPF protocol testing took place at SURAnet's College Park facility, the week of December 12, 1990. Four implementations participated, from the vendors 3com, ACC, Proteon and Wellfleet.

There were two 3com routers, two ACC routers, two Wellfleet routers and four Proteon routers available for testing. These routers were interconnected with ethernets, serial lines and token rings. External routes were imported from SURAnet by one of the Proteon routers.

The testing at SURAnet proceeded as follows. Initially nine routers were configured as a single backbone area, with six of the routers connected to a single ethernet. This is pictured in Figure 4. In this configuration, the Designated Router transition and database synchronization process were tested. Ninety-six external routes were imported from SURAnet to stress the flooding algorithm. By restarting the router that was importing the routes, the flushing of advertisements from the routing domain was tested. Additionally, variable-length subnets and OSPF's optional TOS routing capability were tested in this configuration.

Next the routers were configured into four separate OSPF areas, with each area directly connected to the OSPF backbone (which was a single ethernet). There were no virtual links in this configuration. Inter-

area routing was tested, including having AS boundary routers internal to a non-backbone area. Also tested was the case where a single router was both an area border router and an AS boundary router.

For more details of the testing, see the "Official report of the Second Round Testing" listed below.

7.3.1 Official report of the Second round testing

The following report was sent to the ospf, ospf-tests, and router- requirements mailing lists after the second round of interoperability tests:

The second round of OSPF multi-vendor testing was done in College Park, Maryland the week of 12/17/90. The facilities were provided by SURAnet. Four major router vendors were present, Advanced Computer Communications (ACC), Proteon, 3Com, and Wellfleet. A press conference and presentation was provided for 3 different data communication magazine representatives.

Each vendor provided at least 2 routers. Each vendor had a device connected to a common Ethernet. This Ethernet was configured as the OSPF backbone area. The rest of the routers were attached to the various backbone routers via Ethernet, Token Ring, proprietary serial line, PPP serial line, and X.25 type media. The following test scenarios were performed and completed in the following order:

o Intra-area routing. All routers were configured to reside in the

  backbone area. Designated Router election was performed various
  number of times so each vendor could be designated router for a
  period of time. Packet data was captured on a Sniffer for later
  observation.

o Variable Length Subnet Masks. Some of the serial lines in the

  configuration were configured to be on the same IP network but with
  different subnet masks. It was observed that all routers stored
  routes for the different length subnets.

o Import SURAnet routes. The Proteon router was listening for RIP

  routes generated by the SURAnet routers. These routes were imported
  into our OSPF test system. 96 external link state advertisements were
  generated as a result. Many scaling type implementation problems
  surfaced for each vendor during this exercise.

o Type of Service generation. While the test setup was still configured

  as a single area, the 3Com router generated Type of Service link

  state advertisements. It was observed how the other vendor
  implementations reacted to it. Some problems were found.

o Inter-area routing. Multiple areas were configured. Common non-

  backbone areas were shared by Proteon and Wellfleet and by ACC and
  3Com. It was observed that the correct Intra-area and Inter-area
  routes appeared in each router's routing table. At this point in the
  test setup, the Proteon router regenerated the 96 SURAnet routes into
  the configuration. It was observed that the routes were learned and
  propagated over area boundaries. Some problems occur at this point.
  More emphasis on this scenario will occur at the next round of
  testing.

o OSPF over X.25. A point-to-point link was connected between the

  Proteon router and the 3Com router. The X.25 packet level was
  configured to run over the link. OSPF was enabled over the link to
  verify that multi-vendor OSPF over X.25 was performed. Both of these
  routers were in the same area.

o MaxAge advertisements. Link state advertisements were flushed from

  the routing domain using the MaxAge procedure. We verified that all
  routers removed the advertisements from their databases, after they
  were properly acknowledged by the flooding procedure. Some problems
  were found in this test, although not nearly as many as in the first
  round of testing.

Each vendor agreed that this sort of testing was extremely valuable and that it should occur again. 3Com has offered for the third round of testing to occur in Santa Clara sometime in February. 3Com will encourage other OSPF implementations to join in the testing. Items that will be tested are:

o Intra-area routing with loops (as well as equal-cost multipath).

o Inter-area testing including Stub and Transit area support, with both

  Intra-area and Inter-area loops.

o Virtual link testing in the looped Inter-area configuration.

o RIP/OSPF route interchange including testing forwarding address

  capability in external link state advertisements.

o EGP/OSPF router interchange including the use of the route tag field

  in external link state advertisements.

o More than two routers connected to an X.25 network. We would like to

  test OSPF's non-broadcast multi-access capabilities by attaching more
  than two vendor's routers to an X.25 packet switch.

o Several vendors running OSPF and RIP simultaneously. This will

  further test the OSPF/RIP interactions.

o Test processing of links with cost LSInfinity. These links should be

  treated as unreachable.

Furthermore, we hope that in future rounds of testing an OSPF MIB would allow us to also use a network management station to gather test data.

In summary, the stability of implementations were better this time more so than the first round of testing. No problems with the protocol design were encountered. The exchange of ideas and the cooperation among implementors that occurred during this test effort, continues the spirit that OSPF is truly an open protocol.

7.3.2 Problems found in the Second round testing

No problems were found in the OSPF protocol during the second round of testing.

7.4 Third round (3com, 2/4/91 - 2/8/91)

The third round of OSPF protocol testing took place at 3com's Santa Clara facility, the week of February 4, 1991. Five implementations participated, from the vendors 3com, ACC, Proteon and Wellfleet and the publicly available University of Maryland implementation (running on a SUN workstation).

There were five 3com routers, four ACC routers, three Wellfleet routers, three Proteon routers and the UMD Sun workstation available for testing (giving a total of 16 routers available). These routers were interconnected with ethernets, serial lines and X.25. External routes were imported from BARRNet by one of the 3com routers.

The initial testing configuration is shown in Figure 5. Three areas were configured, along with a non-contiguous backbone. The backbone was then joined by configuring two virtual links. In this configuration the following OSPF functionality was tested: inter-area routing and virtual links.

The system was then reconfigured so that twelve of the routers were connected to a single ethernet. This configuration is pictured in Figure

By bringing routers up and down, this configuration tested Designated

Router election, database synchronization and reliable flooding. To see how this functionality, and also the implementations, scale, 400

external routes were imported from BARRNet.

7.4.1 Official report of the Third round testing

The following report was sent to the ospf, ospf-tests, and router- requirements mailing lists after the third round of interoperability tests:

The third round of OSPF interoperability testing was held at 3com Corporation in Santa Clara the week of February 4-8. Four router vendors came to the testing: 3com, ACC, Proteon and Wellfleet. In addition, Rob Coltun brought the University of Maryland implementation, which was run on a Sun Workstation.

Testing was performed over ethernet, point-to-point networks (using PPP) and X.25. In all we had 16 routers available: five 3com routers, four ACC routers, three Proteon routers, three Wellfleet routers and Rob's SUN. We also were able to import external routes from BARRNet.

Specific tests performed included the following:

o Initially we configured the routers into three separate areas and a

  physically disconnected backbone. The backbone was then reconnected
  through configuration of several virtual links.  These tests verified
  the generation and processing of summary link advertisements, as well
  as the operation of virtual links.

o We connected multiple routers to an X.25 packet switch, testing

  OSPF's non-broadcast network capability. OSPF was successfully run
  over the X.25 network, using routers that were both DR eligible and
  DR ineligible. Some problems were encountered, but they all involved
  running IP over X.25 (i.e., they were not X.25 specific).

o We also connected a 3com router, Proteon router, and Rob's SUN to an

  ethernet, and then treated the ethernet as a non-broadcast network.
  This allowed us to connect Rob's SUN into the rest of the routing
  domain without installing the IP multicast modifications to the SUN
  kernel. It also further tested the OSPF's non-broadcast network
  capability.

o We then reconfigured the OSPF system so that all but three of the

  routers were connected to a single ethernet. This tested the
  Designated Router functionality (12 routers were synchronizing with
  the DR). We then also tested the DR election algorithm, by
  selectively restarting the DR, or sometimes both the DR and the
  Backup DR. This also tested OSPF's Database Description process.

o In this configuration, we then imported 400 external routes from

  BARRNet (one of the 3com routers ran both OSPF and EGP). Some
  problems were encountered in implementations' buffer allocation
  strategies, and problems were also found in the way implementations
  avoid IP fragmentation. But overall, this system was fairly stable.

The following problems we found in the OSPF specification:

o The specification should say that the "Network mask" field should not

  be verified in OSPF Hellos received over virtual links.

o The specification should say that on multi-access networks neighbors

  are identified by their IP address, and on point-to-point networks
  and virtual links by their OSPF Router ID. This eliminates confusion
  when, for example, a router is restarted and comes up with the same
  IP address but a different Router ID.

Thanks to 3com for providing the testing facility, cables. terminals, X.25 switch. etc. Also thanks to Vince Fuller of BARRNet for helping us import the BARRNet routes.

7.4.2 Problems found in the Third round testing

A couple of specification/protocol problems were found in the third round of interoperability testing. First, it was noticed that the specification did not specify the setting of the Network Mask field in Hellos sent over virtual links. This caused some initial difficulty in bringing up virtual links between routers belonging to different vendors. Secondly, it was noticed that the specification was not strict enough in defining how OSPF neighbors are identified on multi-access networks. This caused difficulties in one implementation when another vendor's router was restarted with the same IP address but a different OSPF Router ID. This is discussed more fully in the above "Official Report of the Third Round Testing".

7.5 Overall: Features tested

All significant protocol features and mechanisms have been tested in the three rounds of interoperability testing. In particular, the following basic pieces of the protocol have been tested:

o Designated Router election. With as many as thirteen routers attached

  to a single LAN, the election of Backup Designated Router and
  Designated router was verified by bringing routers up and down,

  singly and in pairs.

o Adjacency bringup. The Database Description process was verified,

  with databases having over 400 LSAs. Adjacency bringup was also
  verified in times when flooding was taking place simultaneously.

o Reliable flooding. It was verified that OSPF's flooding algorithm

  maintains database synchronization, both in the presence of loops in
  the topology, and with large databases (over 400 LSAs).

o Flushing advertisements from routing domain. OSPF's procedure for

  flushing old or unreachable LSAs from the routing domain was
  verified, both in the presence of topology loops and with many LSAs
  being flushed at once. This is also referred to as OSPF's MaxAge
  procedure.

o OSPF routing hierarchy. The OSPF four level routing hierarchy:

  intra-area, inter-area. type 1 externals and type 2 externals was
  tested.

o Import of external routing information. The importing of external

  routes has been tested, with as many as 400 imported at once. Also,
  the varying options in external LSAs has been tested: type 1 or type
  2 metrics and forwarding addresses.escribe all options. In addition,
  test setups were utilized having AS boundary routers both internal to
  non-backbone areas and also being simultaneously area border routers.

o Running protocol over various network types. In the test setups, OSPF

  has been run over ethernet, point-to-point serial lines (both PPP and
  proprietary), 802.5 token ring and X.25.

o Non-broadcast, multi-access networks. OSPF has been tested over X.25.

  Some testing was also done treating ethernet as a non-broadcast
  network. Two separate situations were tested: when all routers
  attached to the non-broadcast network were DR-eligible, and when only
  some of them were.

o Authentication. OSPF's authentication procedure was tested for the

  two current authentication types.

o Equal-cost multipath. Much of the testing was done in configurations

  with redundant paths, and equal-cost multipath was verified through
  examination of implementations' routing tables.

o Variable-length subnet masks. It was verified that implementations

  paid attention to the network mask field present in OSPF LSAs.

Testing was also performed on the following pieces of OSPF's Area functionality:

o Extent of advertisements. It was verified that all advertisements

  except external LSAs were flooded throughout a single area only.

o Inter-area routing. The generation and processing of summary link

  LSAs was tested. Also tested were configurations having multiple area
  border routers attaching to a single area.

o Virtual links.

The following OSPF options were also tested:

o TOS routing. The interplay between TOS-capable and non-TOS-capable

  routers was tested, by configuring TOS-specific metrics in the only
  implementation (3com) supporting TOS routing.

o Stub areas. OSPF's stub area functionality was verified.

7.6 Testing conclusions

The interoperability testing has proven to be very valuable. Many bugs were found (and fixed) in the implementations. Some protocol problems were found (and fixed), and gray areas of the specification were cleared up. Implementations have also been "bullet-proofed" in order to deal with the unexpected behavior of other implementations. All participants in the testing now understand the maxim "be conservative in what you generate, and liberal in what you accept" (if they didn't already).

7.7 Future work

The one thing that has gone mostly untested at the interoperability sessions is the interaction between OSPF and other routing protocols (such as RIP and EGP). Each interoperability session generally had a router running multiple routing protocols in order to import external routing information into the OSPF system. However, simultaneously running multiple routing protocols between different vendors' routers has not been tested.

Each vendor has developed a slightly different architecture for the exchange of routing information between differing routing protocols. As the OSPF field testing has also shown, this exchange of routing information is an area of ongoing work and a candidate for future standardization.

8.0 Simulation

The OSPF protocol has been simulated by the Distributed Systems Research Group at the University of Maryland Baltimore County (UMBC). The two principal investigators for the OSPF simulation project are Dr. Deepinder P. Sidhu of UMBC and Rob Coltun. They have been aided by three graduate students: S. Abdallah, T. Fu and R. Nair. This section attempts to summarize their simulation setup and results. For more information, contact the Distributed Systems Research Group at the following address:

       Dr. Deepinder P. Sidhu
       Department of Computer Science
       University of Maryland Baltimore County
       Baltimore, MD 21228
       email: [email protected]

A demo was given of their OSPF simulation at the March 4-8, 1991 IETF in St. Louis. Details of the demo should be available in the IETF proceedings.

8.1 Simulator setup

The Distributed System Research Group uses a significantly enhanced version of the MIT network simulator. The simulator is event driven, and contains support for both point-to-point networks and ethernet links. It can simulate characteristics of both packet switches and hosts, and can simulate internet behavior under various types of data traffic load (e.g., Poisson, normal, exponential and uniform distributions). This latter ability could be used, for example, to simulate how a routing protocol works in a congested internet. Specific network topologies can be input into the simulator, or pseudo-random network topologies can be generated. Packet loss rates can also be simulated.

To simulate OSPF, Rob Coltun's OSPF implementation was incorporated into the simulator, essentially unchanged.

The output of the simulator can be displayed in a graphical manner (it uses X windows). Any variable in the implementation under test can be monitored. In addition, statistical reports can later be produced from logging files produced during the simulation runs.

8.2 Simulation results

The OSPF simulation has been run using the following topologies:

o The two sample topologies in the OSPF specification (Figure 2 and

  Figure 6 in [2]). The first of these topologies shows an Autonomous
  System without areas, the second the same AS with three areas and a
  virtual link configured.

o The 19-node hub topology from [7].

o A large network of over 50 nodes, all attached to a single ethernet.

o A large network of over 50 nodes, containing both ethernets and

  serial lines, pseudo randomly generated.

In these topologies, the correctness of the OSPF database synchronization was verified. This was done through examination of the nodes' OSPF link state databases and the nodes' routing tables. The implementation of multiple OSPF areas was also tested. Also, database convergence time was analyzed as a function of the network components' link speeds.

Also, some formal analysis of the OSPF protocol was undertaken. The neighbor and interface state machines were analyzed. In addition, the Designated Router election algorithm was verified for certain sets of initial conditions.

9.0 Reference Documents

The following documents have been referenced by this report:

[1] Moy, J., "The OSPF Specification", RFC 1131, October 1989.

[2] Moy, J., "OSPF Version 2", RFC 1247, July 1991.

[3] Coltun, R. and Baker, F., "OSPF Version 2 Management Information

   Base", RFC 1248, July 1991.

[4] Reynolds, J. and Postel, J., "Assigned Numbers', RFC1060, March

   1990.

[5] Corporation for National Research Initiatives, "Proceedings of the

   Thirteenth Internet Engineering Task Force", Cocoa Beach, Florida,
   April 11-14, 1989.

[6] Corporation for National Research Initiatives, "Proceedings of the

   Sixteenth Internet Engineering Task Force", Florida State
   University, February 6-9, 1990.

[7] Gardner, M., et al., "Type-of-service routing: modeling and

   simulation," Report 6364, BBN Communications Corporation, January
   1987.

[8] Corporation for National Research Initiatives, "Proceedings of the

   Seventeenth Internet Engineering Task Force", Pittsburgh
   Supercomputing Center, May 1-4, 1990.

[9] Corporation for National Research Initiatives, "Proceedings of the

   Eighteenth Internet Engineering Task Force", University of British
   Columbia, July 30-August 3, 1990.

10.0 People

The following people have contributed information to this report and can be contacted for further information:

TABLE 5. People references in this report

Tag Name Affiliation email __________________________________________________________________________________ bstreile William Streilein Wellfleet [email protected] dino Dino Farinacci 3com [email protected] fbaker Fred Baker ACC [email protected] jeff Jeffrey Burgan Sterling Software [email protected] jhsu Jonathan Hsu Wellfleet [email protected] jmoy John Moy Proteon [email protected] kannan Kannan Varadhan OARnet [email protected] medin Milo Medin Sterling Software [email protected] rcoltun Rob Coltun U. of Maryland [email protected] rrv Ross Veach U. of Illinois [email protected] vaf Vince Fuller BARRNet [email protected]

Security Considerations

The OSPF protocol's security architecture is described in Section 4.0.

Author's Address

John Moy Proteon Inc. 2 Technology Drive Westborough, MA 01581

Phone: (508) 898-2800 Email: [email protected]