Performance Monitoring

If you have questions or if you want to share your opinion about Aware IM post your message on this forum
Post Reply
intra
Posts: 279
Joined: Thu Oct 11, 2012 1:30 pm
Location: Australia

Performance Monitoring

Post by intra »

Hi all,

In an effort to keep my production environments in good shape, i keep fairly detailed logs regarding the overall system performance and other associated metrics to help troubleshoot issues when they arise. I was wondering how others in their environments monitor performance?

I keep fairly detailed metrics (per second) and aggregate up to a time range as needed (anywhere from 5 minutes to a year aggregated) as shown the attached screenshot, just wondering if people would be willing to share their methods to ensure that parts are functioning within spec?
Attachments
Screenshot from 2019-02-19 15-10-21.jpg
Screenshot from 2019-02-19 15-10-21.jpg (122.89 KiB) Viewed 5160 times
Avid Linux user....
customaware
Posts: 2391
Joined: Mon Jul 02, 2012 12:24 am
Location: Ulaanbaatar, Mongolia

Re: Performance Monitoring

Post by customaware »

Show Off! :-D
Cheers,
Mark
_________________
AwareIM 6.0, 8.7, 8.8, 9.0 , MariaDB, Windows 10, Ubuntu Linux. Theme: Default, Browser: Arc
Upcloud, Obsidian....
Image
hpl123
Posts: 2579
Joined: Fri Feb 01, 2013 1:13 pm
Location: Scandinavia

Re: Performance Monitoring

Post by hpl123 »

Well, that looks awesome and I wish I had something like that haha :). Any possibility you could share the BSV IF this is AIM?

Another cool idea would be to get better stats regarding AIM performance integrated here.
Henrik (V8 Developer Ed. - Windows)
RLJB
Posts: 914
Joined: Tue Jan 05, 2010 10:16 am
Location: Sydney, Australia

Re: Performance Monitoring

Post by RLJB »

Hey Pete, interesting topic, here's my position on this...

We have tried and set-up numerous monitoring systems over the years. Many look very similar to what you have. It's useful to an extent, we get notified by SMS when things get out of acceptable bands etc. We know certain things within our environment we should check when things peak (eg Execution Context, Database monitors for long running queries etc).

BUT...

The question is what to do with the info?

The Aware logs lack the specific info you need (in a format that is usable) to tie an episode of degraded server performance with an app/bsv event, especially if you are running multiple BSVs on the one server. I've banged on before about logging improvements required to do this (search forum for logging).

So, in my opinion you can get insight into things like:

"my server is peaking at this time of day"
"it's Aware server process (or Tomcat process) that is peaking"

BUT you can't answer the obvious next question "what in Aware is causing this?"

So, is it useful? Yes - you must have some server monitoring. But without improved logging it isn't THAT useful.

Happy to be proven wrong and someone can show me the way to do this better at conference. In fact it would make a great Conference topic in Portugal, because as you scale we have found this topic to be a major drain on resources as we scramble and try and figure out why a server is running slow/peaking.

btw - we have tried services like DataDog to compare server peaks with logs but it isn't useful due to the way Aware logs.
Rod. Aware 8.6 (latest build), Developer Edition, on OS Linux (Ubuntu) using GUI hosted on AWS EC2, MYSQL on AWS RDS
Post Reply