For more information regarding the security incident at F5, the actions we are taking to address it, and our ongoing efforts to protect our customers, click here.

Forum Discussion

sundogbrew's avatar
sundogbrew
Icon for Altocumulus rankAltocumulus
Aug 15, 2014

TPS question...

So I was monitoring the dashboard during a load test and I saw a spike which was like 538 TPS (my license is for 500) but there was no message in /var/log/ltm that I had exceeded my license... I know that it goes by a percentage like in my case it would be 50 per 10ms but if it exceeded 500 it seems it would have exceeded that limit? Can anyone shed some light on this? This is one of those things that I fail to understand. Since it is so critical you would think they would have a better way of telling you how you are doing. Like tell you your high-water mark and then your average. When I saw that I exceeded my license and then didn't get a message it makes me wonder just what I am getting when I look at that graph. Thanks in advance Joe

 

7 Replies

  • I think TPS is licensed per CPU core, so for example a Viprion 8-core would be 4,000 TPS.

     

  • That's good info, but what does that mean in terms of where I am at with my limits? The dashboard says something about the busiest CPU but how does it handle the load, do I have another 400 on the other core?

     

  • Peter, Thanks for the response. I understand the load sharing but I guess I am having trouble wording my question. How is this, when I look at the dashboard and it says you are using 400 of your available 500 TPS, and as previously stated you get 500 for each core. So with two cores am I actually seeing 800 of 1000 because I have two cores and they are evenly distributed, or am I seeing 400 of 1000 and there are like 200 on each core? Thanks Joe

     

  • have you read this SOL: http://support.f5.com/kb/en-us/solutions/public/6000/400/sol6475.html

     

    also you might have just one core, what hardware platform are you using?

     

  • Imperically, I bet you're seeing '400 of 1000'. I think the wording in the dashboard is confusing. using 'tmsh show sys performance throughput' you'll see similar numbers without the confusing reference to 500.

     

    I say 'Imperically' above because we see over 1,000tps and don't experience any problems (no /var/log/ltm warnings about ssl licenses) and no service impact.

     

  • I read the SOL which sort of beats around the same bush. Tells you about how the graph represents a full second and the TPS is measured in 10ms so there is rounding off. But it is more like you may get the error and not see that many connections but not you may see the connections and not get the error? To get 530 in a second I would imagine it was averaging 5+/ms which again goes back to do I actually have 10/ms to work with and it just won't tell me?! So I did the command that pete suggested which gave me this... (it's lunch time so the number is pretty low.)

     

    SSL Transactions Current Average Max(since 08/18/14 10:46:31)

    SSL TPS 60 65 94

     

    Also in the article it says to run this

     

    grep -i "perf_SSL_total_TPS" /config/bigip.license SSL, 500 TPS Per Core perf_SSL_total_TPS 500 key KJSAWMJ-KFYUCRP

    perf_SSL_total_TPS : 500

     

    So that makes you think you only have one core... So then I did this to see how many cores I have...

     

    log grep -i core /proc/cpuinfo core id: 0 cpu cores: 2 core id: 1 cpu cores: 2

     

    So according to that I have a dual core right? So I should have 1000 TPS to use up and according to Pete's command I am averaging 65 for the last couple of hours? So I need to run Pete's command during the load and see what I get for numbers? I HATE when things are vague like this...

     

    Thanks Joe