Re: [Engine-devel] DB performance optimization

Hi Laszlo, I agree with both of your comments. 1. Duplicate data cache. Right now it is definitely possible. 2. Updates. Right now there is only a time based invalidation, but that will be changed. We will add a feature to map between stored procedures and the tables they update, and will invalidate relevant data. Caching on the SQL level is hard (harder than Hibernate 2nd level cache for instance) but that's what we have right now... (And I'm not sure that JPA/Hibernate will be better). ----- Original Message ----- From: "Laszlo Hornyak" <lhornyak(a)redhat.com&gt; To: "Liran Zelkha" <lzelkha(a)redhat.com&gt; Cc: "engine-devel" <engine-devel(a)ovirt.org&gt; Sent: Monday, May 13, 2013 12:28:59 PM Subject: Re: [Engine-devel] DB performance optimization Hi Liran, Thanks for this patch, I have been asking for caching for a long while and I hope to see caching in the engine soon. Beyond the comments I have made to your patch, I have some questions about the idea just to start the discussion. So your cache solution is caching data based on SQL statements? What will the cache do if we query the same data with two different SQL statements? e.g. you can query from views like vms and sometimes we only query the vm_static and we should get the same vm_static. Also, how will you deal with updates? e.g. when you update the vm_static record, that vm_static record must be updated or invalidated but also the vm record from Thx, Laszlo ----- Original Message -----