Any questions? Post'em

Hi Zafer,

 

 

I encourage you to work with your local field engineer to help size deployments. When a product does more work, it requires more resources, and understanding the performance of a combined feature-set (SSL + compression + ...) is a fairly involved multi-dimensional problem (lots of variables, many things to consider).

 

 

Having said that, the relative advantage of the BIG-IP product versus competitors is still strong. If BIG-IP can handle more L4, L7, SSL, compression, and so on for individual tests, that also means BIG-IP can handle more in combined tests. For example, if BIG-IP handles 6Gbps of compression and 6Gbps of SSL in separate tests, where a competitor may handle 3Gbps of compression and 3Gbps of SSL in separate tests, neither vendor will achieve both metrics at the same time, but perhaps BIG-IP will achieve 4Gbps of SSL+compression, and the competitor will achieve 2Gbps of SSL+compression -- BIG-IP's advantage is the same whether you look at individual metrics or combined metrics.

 

 

About iRules performance, I've performed extensive tests of L7 performance on both BIG-IP and competitive platforms (including Alteon, Redline, and Foundry). If you compare similar platform models and feature-sets between products, BIG-IP's performance with iRules is consistently higher. However, if you're using iRules to do something uncommon, something that the competitors are unable to do at all, then there's of course no way to compare this directly to the competitors. For the "L7" tests that vendors in our market use as the baseline, it's inspecting an HTTP URL to select between different groups of servers. All vendors support this basic L7 feature-set, so it's a good baseline comparison. The advertised L7 performance of BIG-IP is based on this same sort of test, using iRules. As you've seen in the report(s), BIG-IP's performance using iRules for this task is very competitive against our competitors. The real key in judging L7 performance is to make sure there's a similar feature-set in use between platforms that are being compared. Based on my extensive testing, I'm very confident BIG-IP will come out ahead. :)

 

 

For common tasks like selecting a pool of servers based on the HTTP URI you don't even need iRules -- I suggest using the httpclass profile (HTTP Classification profile) instead. For more advanced tasks, my experience testing Alteon, Redline, Foundry, and more has shown that BIG-IP's performance is very competitive. If you're seeing 60% CPU on the BIG-IP, then you'll see more than 60% CPU on similar competitive hardware. If you see higher utilization on the BIG-IP compared to a similar competitive platform, then I suggest a different set of features must be in use (perhaps BIG-IP is acting as a full proxy, whereas the competitor is not, in which case it's appropriate to use a simpler non-full proxy mode on the BIG-IP).

 

 

Hope this helps, good luck!

 

 

Mike Lowell