On Fri, May 4, 2012 at 3:57 PM, Mark Reynolds <marey...@redhat.com> wrote: > Deigo, > > In the meantime, you should get a performance boost if your top "tree" > suffix(dc=company,dc=com) has the same attributes indexed as all the other > sub-suffixes(db's). Even if the db is empty, this will still help when you > search on the top node. > > Mark > > > On 05/04/2012 10:15 AM, Rich Megginson wrote: >> >> On 05/04/2012 07:44 AM, Diego Woitasen wrote: >>> >>> On Fri, May 4, 2012 at 10:24 AM, Rich Megginson<rmegg...@redhat.com> >>> wrote: >>>> >>>> On 05/04/2012 06:47 AM, Diego Woitasen wrote: >>>>> >>>>> I didn't know how to title this mail. I think this should be a feature >>>>> request in Track when I want to discuss this here first. >>>>> >>>>> I have 389DS with 150 DBs with an structure similar to this: >>>>> >>>>> dc=company,dc=com >>>>> ou=Headquarters,dc=company,dc=com >>>>> ou=Branch1,dc=company,dc=com >>>>> ou=Branch2,dc=company,dc=com >>>>> . >>>>> . >>>>> . >>>>> ou=Branch150,dc=company,dc=com >>>>> >>>>> Each one of this subtrees are in separate DBs because I have subtree >>>>> replication between the 150 branches of the companies. >>>>> >>>>> 80% of the objects are in the ou=HeadQuarters. I've noticed that the >>>>> performance is definetely better when I use base ou=Headquarters in my >>>>> applications. >>>>> >>>>> I have indexes on each DB but I think that the problem is that 389DS >>>>> doesn't have a master index or something to improve the searchs in >>>>> scenarios like mine. >>>> >>>> >>>> Can you explain more about what you mean by "master index"? >>> >>> An index that includes all the DBs. May be "global index" is a better >>> name. Right now, when you search for something, 389DS queries all the >>> DBs, one by one and with 150 DBs is a problem. There should be a way >>> to avoid that. >> >> >> Ok, I see. Yes, might be useful too for doing simple paged searches, >> server side sorting, vlv, etc. across multiple databases. >> >>> >>> >>>> >>>>> May be the solution is to implemen another replication code that >>>>> doesn't required separate DBs for subtree replication. >>>>> >>>>> Shall I file a ticket? Or there is a solution now? >>>>> >>>>> Regards, >>>>> Diego >>>>> >>> >>> >> >> -- >> 389 users mailing list >> 389-us...@lists.fedoraproject.org >> https://admin.fedoraproject.org/mailman/listinfo/389-users
I've filed a ticket https://fedorahosted.org/389/ticket/357 -- Diego Woitasen -- 389 users mailing list 389-us...@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/389-users
