Of course none of the other vendors had a chance to fix, twiddle, etc., so
I'm hoping we just let the Data Comm stuff rest now. Data Communications
will be running new tests every few weeks as vendors discover bugs and fix
things, no?
Our results were based on Data Communications installing, configuring, and
running the Gauntlet Internet Firewall, w/o any changes by TIS.
Fred
At 09:17 AM 3/14/96 -0500, Brien Wheeler wrote:
>Philippe Cayphas (pc @
godel2 .
bim .
be) writes:
>> Data Communications has performed a very interesting comparison of a lot of
>> firewalls.
>> Let's say that Raptor product was not the best. Far away.
>>
>> This study can be found at :
>> http://www.data.com/Lab_Tests/Firewalls.html
>
>The test tools used by Data Communications in the preparation of the article
>that you cite created an unnaturally high connection rate when attempting
>to configure all virtual clients. This high connection rate exposed a minor
>bug in Raptor's Generic Service Passer (one that had never impacted any
>customer), causing the _test configuration_ to fail. The actual tests,
>which generated a more realistic (but still high) connection load were not
>performed during that testing.
>
>Raptor fixed the bug and retested at NSTL (where the original testing was
>performed) on the same hardware, using the exact original test tools and
>scenarios. On retest, the Eagle (running on a SPARC 5) performed better
>than all firewalls tested for the original article. A SPARC 20 Eagle
>that was tested for comparison purposes was literally off the chart,
>serving up to 1200 "transactions per minute", while the chart published
>in the original article stopped at 900 tpm.
>
>An update which mentions all this appeared in a recent (March 1996)
>Data Communications issue.
>
> Brien Wheeler
> Software Engineer
> Raptor Systems, Inc.
>
>
|
|
