Network Working Group Vishwas Manral
Internet Draft Netplane Systems
Russ White
Cisco Systems
Aman Shaikh
Expiration Date: December 2002 University of California
File Name: draft-ietf-bmwg-ospfconv-applicability-00.txt June 2002
Benchmarking Applicability for Basic OSPF Convergence
draft-ietf-bmwg-ospfconv-applicability-00.txt
1. 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. Internet Drafts may be updated, replaced, or obsoleted by
other documents at any time. It is not appropriate to use Internet
Drafts as reference material or to cite them other than as a "working
draft" or "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.
2. Abstract
This draft describes the applicability of [2] and similar work which
may be done in the future. Refer to [3] for terminology used in this
draft and [2]. The draft defines the advantages as well as
limitations of using the method defined in [2], besides describing
the pitfalls to avoid during measurement.
Manral, et. all [Page 1]
INTERNET DRAFTdraft-ietf-bmwg-ospfconv-applicability-00.txt June 2002
3. 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 [1].
4. Motivation
There is a growing interest in testing SR-Convergence for routing
protocols, with many people looking at testing methodologies which
can provide information on how long it takes for a network to
converge after various network events occur. It is important to
consider the framework within which any given convergence test is
executed when attempting to apply the results of the testing, since
the framework can have a major impact on the results. For instance,
determining when a network is converged, what parts of the router's
operation are considered within the testing, and other such things
will have a major impact on what apparent performance routing
protocols provide.
This document describes in detail the various benefits and pitfalls
of tests described in [2]. It also explains how such measurements can
be useful for providers and the research community.
5. Advantages of Such Measurement
o To be able to compare the iterations of a protocol implementa-
tion. It is often useful to be able to compare the performance
of two iterations of a given implementation of a protocol to
determine where improvements have been made and where further
improvements can be made.
o To understand, given a set parameters (network conditions), how
a particular implementation on a particular device is going to
perform. For instance, if you were trying to decide the process-
ing power (size of device) required in a certain location within
a network, you can emulate the conditions which are going to
exist at that point in the network and use the test described to
measure the perfomance of several different routers. The results
of these tests can provide one possible data point for an intel-
ligent decision.
If the device being tested is to be deployed in a running net-
work, using routes taken from the network where the equipment is
to be deployed rather than some generated topology in these
Manral, et. all [Page 2]
INTERNET DRAFTdraft-ietf-bmwg-ospfconv-applicability-00.txt June 2002
tests will give results which are closer to the real preformance
of the device. Care should be taken to emulate or take routes
from the actual location in the network where the device will be
(or would be) deployed. For instance, one set of routes may be
taken from an abr, one set from an area 0 only router, various
sets from stub area, another set from various normal areas, etc.
o To measure the performance of an OSPF implementation in a wide
variety of scenarios.
o To be used as parameters in OSPF simulations by researchers. It
may some times be required for certain kinds of research to
measure the individual delays of each parameter within an OSPF
implementation. These delays can be measured using the methods
defined in [2].
o To help optimize certain configurable parameters. It may some
times be helpful for operators to know the delay required for
individual tasks so as to optimize the resource usage in the
network i.e. if it is found that the processing time is x
seconds on a router, it would be helpful to determine the rate
at which to flood LSAs to that router so as to not overload the
network.
6. Assumptions Made and Limitations of such measurements
o The interactions of SR-Convergence and forwarding; testing is
restricted to events occurring within the control plane. For-
warding performance is the primary focus in [4] and it is
expected to be dealt with in work that ensues from [5].
o Duplicate LSAs are Acknowledged Immediately. A few tests rely
on the property that duplicate LSA Acknowledgements are not
delayed but are done immediately. However if some implementa-
tion does not acknowledge duplicate LSAs immediately on
receipt, the testing methods presented in [2] could give inac-
curate measurements.
o It is assumed that SPF is non-preemptive. If SPF is implemented
so that it can (and will be) preempted, the SPF measurements
taken in [2] would include the times that the SPF process is
not running ([2] measures the total time taken for SPF to run,
not the amount of time that SPF actually spends on the device's
processor), thus giving inaccurate measurements.
o Some implementations may be multithreaded or use a
Manral, et. all [Page 3]
INTERNET DRAFTdraft-ietf-bmwg-ospfconv-applicability-00.txt June 2002
multiprocess/multirouter model of OSPF. If because of this any
of the assumptions taken in measurement are violated in such a
model, it could lead to inaccurate measurements.
o The measurements resulting from the tests in [2] may not pro-
vide the information required to deploy a device in a large
scale network. The tests described focus on individual com-
ponents of an OSPF implementation's performance, and it may be
difficult to combine the measurements in a way which accurately
depicts a device's performance in a large scale network.
Further research is required in this area.
7. Observations on the Tests Described in [2]
Some observations taken while implementing the tests described in [2]
are noted in this section.
7.1. Measuring the SPF Processing Time Externally
The most difficult test to perform is the external measurement of the
time required to perform an SPF calculation, since the amount of time
between the first LSA which indicates a topology change and the
duplicate LSA is critical. If the duplicate LSA is sent too quickly,
it may be received before the device under test actually begins run-
ning SPF on the network change information. If the delay between the
two LSAs is too long, the device under test may finish SPF processing
before receiving the duplicate LSA. It is important to closely inves-
tigate any delays between the receipt of an LSA and the beginning of
an SPF calculation in the device under test; multiple tests with
various delays might be required to determine what delay needs to be
used to accurately measure the SPF calculation time.
7.2. Noise in the Measurement Device
The device on which measurements are taken (not the device under
test) also adds noise to the test results, primarily in the form of
delay in packet processing and producing outout from which measure-
ments are taken. The largest source of noise is generally the delay
between the receipt of packets by the measuring device and the infor-
mation about the packet reaching the device's output, where the event
can be measured. The following steps may be taken to reduce this sam-
pling noise:
o Take lot of samples. The more samples which are taken, the less
Manral, et. all [Page 4]
INTERNET DRAFTdraft-ietf-bmwg-ospfconv-applicability-00.txt June 2002
that noise in the measurements will impact the overall measure-
ment, as noise will tend to average out over a large number of
samples.
o Try to take time-stamp for a packet as early as possible.
Depending on the operating system being used on the box, one
can instrument the kernel to take the time-stamp when the
interrupt is processed. This does not eliminate the noise com-
pletely, but at least reduces it.
o Keep the measurement box as lightly loaded as possible, unless
the loading is part of the test itself.
o Having an estimate of noise can also be useful.
The DUT also adds noise to the measurement. The first and third
points also apply to the DUT.
7.3. Gaining an Understanding of the Implementation Improves Measure-
ments
While the tester will (generally) not have access to internal infor-
mation about the OSPF implementation being tested using [2], the more
thorough the tester's knowledge of the implementation is, the more
accurate the results of the tests will be. For instance, in some
implementations, the installation of routes in local routing tables
may occur while the SPF is being calculated, dramatically impacting
the time required to calculate the SPF.
7.4. Gaining an Understanding of the Tests Improves Measurements
One method which can be used to become familiar with the tests
described in [2] is to perform the tests on an OSPF implementation
for which all the internal details are available, such as GateD.
While there is no assurance that any two implementations will be
similar, this will provide a better understanding of the tests them-
selves.
Manral, et. all [Page 5]
INTERNET DRAFTdraft-ietf-bmwg-ospfconv-applicability-00.txt June 2002
8. Acknowledgements
Thanks to Howard Berkowitz, (hcb@clark.net) and the rest of the BGP
benchmarking team for their support and to Kevin
Dubray(kdubray@juniper.net) who realized the need of this draft.
9. References
[1] Bradner, S., "Key words for use in RFCs to Indicate Requirement
Levels", RFC2119, March 1997.
[2] Manral, V., "Benchmarking Methodology for Basic OSPF Convergence",
draft-bmwg-ospfconv-intraarea-00.txt, May 2002
[3] Manral, V., "OSPF Convergence Testing Terminology and Concepts",
draft-bmwg-ospfconv-term-00.txt, My 2002
[4] Bradner, S., McQuaid, J., "Benchmarking Methodology for Network
Interconnect Devices", RFC2544, March 1999.
[5] Trotter, G., "Terminology for Forwarding Information Base (FIB)
based Router Performance", RFC3222, October 2001.
10. Authors' Addresses
Vishwas Manral
Netplane Systems
189 Prashasan Nagar
Road number 72
Jubilee Hills
Hyderabad, India
vmanral@netplane.com
Russ White
Cisco Systems, Inc.
7025 Kit Creek Rd.
Research Triangle Park, NC 27709
riw@cisco.com
Aman Shaikh
Manral, et. all [Page 6]
INTERNET DRAFTdraft-ietf-bmwg-ospfconv-applicability-00.txt June 2002
University of California
School of Engineering
1156 High Street
Santa Cruz, CA 95064
aman@soe.ucsc.edu
Manral, et. all [Page 7]