NetApp Community Update
This site will enter Read Only mode on July 23 as we prepare to move to a new platform. You will still be able to view content, but posting and replying will be temporarily disabled. We're excited to launch our new Community experience on July 30 and more information will follow soon. Stay connected during the transition - Join our Discord community today.
Is it best practice (or should I ask how do I get the maximum performance) to have a smaller number of large aggregates or to have specialized "types of data" based aggregates such as SQL Database Aggregates and carve out datastores and only put SQL Databases in those datastores. Then have SQL Transaction Logs aggregates and carve out datastores and only put SQL Transaction logs in those datastores. Then have ORACLE aggregrates.etc....etc.....Or should I have aggregates based on sequential and non-sequential read/writes and place my datastores in those aggregates? Various storage admins have told me that you should only have one or two very large 64bit aggregates and dump everything in them and this is just as good as anything else and you get the most IOPS. Then other admins state you should have aggregates based on the type of traffic. So I am not sure who to believe.
In my point off view best practice are separate os ( same) with dedup , database with compression. If you have Only one ctrl create 2 agregate to separate database and log but more simple is your configuration more better she Will be