RE: [bmwg] WLAN switch controller I-Ds for BMWG to consider

Instead of arguing with Scott Bradner, would you mind reviewing the two WLAN
Switch Benchmarking drafts and argue with us?  One of the reasons Tom
Alexander is bringing the draft to the BMWG is to avoid these pitfalls.
Your comments on the draft will help us make sure we have repeatable
benchmarks, while keeping sight of real world traffic.

Claims and crashes are not limited to Cisco.  Last year I worked with the
esteemed David Newman on a public "Enterprise Wi-Fi Test" for Network World.
When running WLAN Switch Benchmarking Methodology section 5.1.1, one of the
vendors crashed repeatedly.  Before you start faulting the vendor, realize
that they did not have a copy of WLAN Switch Benchmarking Methodology when
verifying their beta code.

Enterprise Wi-Fi needs to be tested.  Duke University is currently having
problems with iPhones on their network
(http://www.networkworld.com/news/2007/071607-duke-iphone.html).  My focus
is to find these problems with a simulated stimuli in a laboratory
environment and to avoid follies like you mentioned.

And everyone else on the BMWG reflector is invited to comment on the WLAN
Switch Benchmarking drafts.  They can be found at:

http://www.ietf.org/internet-drafts/draft-alexander-bmwg-wlan-switch-term-00
.txt

http://www.ietf.org/internet-drafts/draft-alexander-bmwg-wlan-switch-meth-00
.txt

Jerry Perser

> -----Original Message-----
> From: Curtis Villamizar [mailto:curtis@occnc.com]
> Sent: Thursday, July 19, 2007 8:12 PM
> To: Jerry Perser
> Cc: bmwg@ietf.org
> Subject: Re: [bmwg] WLAN switch controller I-Ds for BMWG to consider
> 
> 
> In message <200707191853.l6JIrf4v010095@mail124c25.carrierzone.com>
> "Jerry Perser" writes:
> >
> > You comment's have me puzzled.  They don't match my experiences with the
> > BMWG over the last several years.
> 
> 
> I remember arguing with Scott Bradner circa 1992/1993 regarding
> "Bradner Test consdiered harmfull" because Cisco was using the Bradner
> test to claim "fastest router in the world" at a time when Cisco
> routers fell over - literally crashed - under the loads that the
> Internet provided.  The Cisco 7000 with its "route cache" design was
> the height of this folly where Cisco designed specifically to get high
> benchmark numbers on less than worthless benchmarks.
> 
> When routes changed the Cisco 7000 dropped packets like crazy due to
> the route cache design.  At high enough loads BGP and IGP adjacencies
> dropped and a network could become unstable.  At even higher loads the
> router would crash.  But in the Bradner test the routes never changed.
> This was the problem with high repeatability but no relevance to real
> world traffic forwarding capability.
> 
> maybe it was just that some of the ISP people tried to push the focus
> in the direction of high end but it never really was there.
> 
> Curtis
> 
> _______________________________________________
> bmwg mailing list
> bmwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/bmwg
> 
> No virus found in this incoming message.
> Checked by AVG Free Edition.
> Version: 7.5.476 / Virus Database: 269.10.11/909 - Release Date: 7/20/2007
> 4:39 PM
> 

No virus found in this outgoing message.
Checked by AVG Free Edition. 
Version: 7.5.476 / Virus Database: 269.10.11/909 - Release Date: 7/20/2007
4:39 PM

_______________________________________________
bmwg mailing list
bmwg@ietf.org
https://www1.ietf.org/mailman/listinfo/bmwg