Hi Michael,
I would suggest testing:
throughput with 64 byte packets
throughput with 1500 byte packets
a distribution of whatever you deem an imix to be ... 64, 128, 512,
1500 with a Poisson distribution.
throughput of periodic packet stream when the control plane is being
affected (ie. either icmp to router, TTL expired, or some router alert
option)
max prefixes
number of lost packets when fiber is pulled to one of the
"uplinks" ... to determine the convergence time.
you can send 1000 packets per second and loop the receive back to the
test unit so you can track how many milliseconds of packet loss.
IPv6 throughput
number of filters / ACLs applied ... and throughput when these ACLs
are applied.
number of supported mac filters
Apply a filter to the control plane and then attack port 179 .. see
how the BGP sessions hold up.
saturate a GE card with transit traffic and see if the network control
packets suffer any loss (ie. session drop, retransmit, etc)
Kind regards,
Truman Boyes
On 22/07/2009, at 8:05 PM, Michael J McCafferty wrote:
All,
We are putting together a test plan to test a pair of Cisco 7206
VXR's,
each with with NPE-G2. The purpose of the test is just to make sure we
know where their realistic limits are with a real configuration, full
route tables from two providers, etc. We have one JDSU T-Berd 8000
test
system with interfaces and software to test a single stream through
multi-mode fiber interfaces. We plan to test through the interfaces on
the NPE and through PA-GE cards with a variety of packet sizes
(especially 64 Byte).
I'd be interested in any thoughts on additional testing or testing
methodologies we might want to do, to help us set our expectations for
this setup and to plan when we need to go bigger as we grow traffic,
hosts, etc.
We plan to get 1 to 3 additional full tables and peer with Any2
Easy on
this network within the next year. We want to determine how this
platform will behave under moderate DoS attacks, BGP updates, etc. Is
there anything else we need to be mindful of? Can we get a realistic
test of the routers with the T-Berd? What else should we test while we
have the maintenance window and the test system on hand?
Your thoughts and experience are appreciated!
Thanks !
Mike
--
************************************************************
Michael J. McCafferty
Principal, Security Engineer
M5 Hosting
http://www.m5hosting.com
You can have your own custom Dedicated Server up and running today !
RedHat Enterprise, CentOS, Ubuntu, Debian, OpenBSD, FreeBSD, and more
************************************************************
