BIG-IP 1500 Http TPS

If a client re-uses a connection to a VIP it would be more efficient than sending each HTTP request over a separate TCP connection. Similarly, if you use a OneConnect profile to have LTM re-use serverside TCP connections, it would be more efficient than LTM opening separate TCP connections for each HTTP request. I would assume the sizing numbers are best case estimates of what the platform can handle. Your actual experience will be highly dependent on the traffic patterns and client and server behavior. The best testing method is to simulate your user traffic patterns through LTM to the actual application you will be load balancing.

 

 

Aaron

Thanks Aaron, but I feel my questions are still not clear.

 

 

Some clarifications;

 

 

1) What traffic profile is refered to when stating the TPS ?

 

2) Can I use keep-alvie option and, if so, what additional configuration is needed?

 

3) Can I assume better performance by using keep-alive ?

 

 

BR

 

J.B

Hi JB,

 

 

1) What traffic profile is refered to when stating the TPS ?

 

 

It's a little dated by now, but F5's performance testing methodology is explained in the following document:

 

 

http://www.f5.com/reports/f5-performance-report.pdf

 

 

And they've provided suggestions for how to conduct your own testing here:

 

 

http://devcentral.f5.com/downloads/f5/creating-performance-test-methodology.pdf

 

 

2) Can I use keep-alvie option and, if so, what additional configuration is needed?

 

You can configure serverside keep-alive with a OneConnect profile:

 

 

SOL10067 - Tuning the OneConnect feature on the BIG-IP LTM

 

https://support.f5.com/kb/en-us/solutions/public/10000/000/sol10067.html

 

 

3) Can I assume better performance by using keep-alive ?

 

You would get better performance using keep-alives than without. But I'm not sure what exact method was used to determine the 2000 TPS rating for the 1500 platform.

 

 

Aaron