All, I was up late last night trying to work out the tests etc. My initial testing may have been flawed but I think Lonnie is on the right track.
At 42 calls (all sent to the echo test application) my 4801 is maxed out and will NOT allow me access via the already connected console. No further access to the machine is possible. I think Douglas Adams was wrong about 42 being the answer to the question of Life, The Universe and Everything ;} This state persists until the concurrent call level drops to 16 at which time one can use the console and do functions. Perhaps a little about my environment? I'm using a Net4801-60 running 0.4.2 as the receipient. A 3.1GHz desktop running Debian "etch" as the load generator (don't want the generator running out of horses!!) and SIPp (sipp.sourceforge.net) as the testing software. Finally, I have a single BT101. All of this is connected via 5 port Netgear 100MB switch which in turn is connected to port 2 of my WRAP based m0n0wall (finally found a use for that port!!). The test is running entirely within it's IP subnet. The test network is connected to the m0n0wall merely so as to allow me to access the net for extra tools etc. It also allows my client to connect in as see my progress. Kris: I like your idea about testing different code bases. I can leave the rig built when I'm done with my paid testing. Finally I've found something that I can do that adds value to the AstLinux project. Et al: Since I'm not an heir to the Rothschild fortune, all donations of hardware would be greatly received. Thanks Mark On Tue, 2006-08-22 at 22:18 -0500, Lonnie Abelbeck wrote: > Mark, > > > I've discoverd that the NET4801 can only sustain 45 concurrent > > calls with a churn rate of 10 seconds before the software crashes. > > That is *only* 4Mbits/s upstream to handle those uLaw calls! > Quite impressive. > > Is this with 0.4.2 AstLinux? If so, I suspect 0.3.0 would *only* > give you 30 concurrent calls. > > I think we can safely say the net4801 can handle 10 simultaneous uLaw > calls with a wide safety/quality margin. > > Great work, Mark. > > Lonnie > > > > On Aug 22, 2006, at 9:34 PM, Mark Phillips wrote: > > > Hi all, > > > > Before I go boring you all with my findings has someone already done > > proper load testing of either the NET4801 or the Gumstix? > > > > I'm not planning to test the NET4801 per-se but rather am using my > > spare > > one as a learning tool for a consultancy job I've landed. In learning > > about the SIPp tool I will have tested the Soekris box to death. For > > example, I've discoverd that the NET4801 can only sustain 45 > > concurrent > > calls with a churn rate of 10 seconds before the software crashes. > > > > Now that I've built the test bed system/lan/etc It would be trivial to > > swap the Soekris for my Gumstix and give that a try. I'm expecting > > about > > double the capacity purely based on processor size. > > > > If there's any interest in this I'd be happy to publish AFTER I've > > done > > my consultancy job (about another 3 weeks). > > > > Mark > > > > _______________________________________________ > > Astlinux-users mailing list > > [email protected] > > http://lists.kriscompanies.com/mailman/listinfo/astlinux-users > > > > Donations to support AstLinux are graciously accepted via PayPal to > > [EMAIL PROTECTED] > > > > _______________________________________________ > Astlinux-users mailing list > [email protected] > http://lists.kriscompanies.com/mailman/listinfo/astlinux-users > > Donations to support AstLinux are graciously accepted via PayPal to [EMAIL > PROTECTED] _______________________________________________ Astlinux-users mailing list [email protected] http://lists.kriscompanies.com/mailman/listinfo/astlinux-users Donations to support AstLinux are graciously accepted via PayPal to [EMAIL PROTECTED]
