Monday, June 18, 2012 5:04 PM
Ok so for version 6.5 of the tool, in the getting started guide I can read:
- When you collect performance data in MAP, you want to collect performance data for the same period of time for all machines
If I read this correctly, I have to run performance data gathering for all machine at the same time for the same amount of time. So let's say that I'm planning for 7 days of data to get a significant sample, all my 300+ machines in prod will be queried at the same time.. that sounds a bit risky.
Or do we need to interpret the time constraint as a common lenght of time. (i.e. I can gather perf data for say 50 machine a week for 6 week and they'll each have a week's worth of data.)
The way I see it the only thing thing that matters is the lenght of the time serie and it needs to be without gaps.
Am I understanding this correctly?
Wednesday, June 20, 2012 12:31 AMModerator
You could do either. For the all at once approach, MAP wouldn't able to query all of them at exactly the same time but they would be polled over the span of a minute or 2, so close to each other in terms of time. The data transferred is initially high at 1-2 Mb since MAP rechecks each machine's inventory for any missing data. After that though it is only 18-20 kb every 5 minutes per machine.
For the staggered approach, as long as you're confident that the usage is similar from week to week, then it shouldn't be a problem. MAP's recommendations are only as good as the data it can collect.
Please remember to click “Mark as Answer” on the post that helps you, and to click “Unmark as Answer” if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread.
Thursday, June 21, 2012 3:00 PM
Thanks a bunch for answering, I figured as much. I'll try to match the data gathering period so they line up with typical workload.