Showing results for 
Search instead for 
Did you mean: 
Create Post

DPA show fragmented indexes


I'm fairly new to using DPA, and I'm not finding what I need. For a very specific scenario, we've got a DPA instance running for our developers. I know in Orion, you can drill down in to individual databases and see some information on index fragmentation %.

Am I completely overlooking where in DPA I can find a similar view? The developers don't and won't have access to Orion, only DPA. I've browsed around and through all the menus I could find and don't see it as a view.

Currently using DPA 11.1.457. Thanks for the assistance!

0 Kudos
3 Replies
Level 14

Here is a custom alert you can set up for fragmentation: Custom Alert - Fragmentation

From a performance point of view, DPA doesn't focus much on fragmentation because it's not that big of a performance hit in modern data centers. In fact, it could even improve performance in some cases. My intent is not to start a debate, but there are a LOT of great articles out there concerning external fragmentation vs internal...


0 Kudos

Thank you, I'll take a look at this as an option. So from what I'm seeing, this would be an alert that would trigger and let people know if any monitored indexes breached a specific fragmentation %? Would there be any way to have this display as a widget for instance to show the top 10 fragmented indexes?

0 Kudos
Level 14

You are exactly correct on the alert. Because of the old charting engine, DPA is more limited than the Orion platform and can really only add a custom metric. You can create one (there's an example out on Thwack) for page splits, but for an index, not sure it makes as much sense. If you knew which larger indexes are prone to fragmentation, you could monitor each with a custom metric and even create a new category for it under resources... Plot the % fragmentation of let's say the top 10 over time to watch the fragmentation grow and then even alert on those new custom metrics...

I don't think the fragmentation queries for the metrics would have to run that often either - maybe even once a day or a couple times a day to keep overhead minimal.