I have a client creating a new system that will have 2 sites with rf/sf per site being 1 and the total rf/sf being 2. Each site will have 3+ indexers.
My question is how this effects bucket distribution over the indexers if you leave maxbucket as the default of 3 and is there any performance implications
They're going rf/sf as 1 due to disk costs.
Thanks in advance
"rf/sf per site being 1 and the total rf/sf being 2"
What do you mean by that?
origin:1,total:2? Or site1:1,site2:1,total:2? Or something else?
the actual config has not been decided yet.
I'm trying to find the best one to spread the buckets across the indexers.
"The best one" actually depends on your:
1) Business needs
2) Budget (and other) constraints.
Thanks for that,
as I stated rf/sf is 1 per site so total of 2 searchable copies due to costs.
I want to know the best way to spread the buckets of an index over as many indexers as possible to get the best bandwidth out of the I/O sub-system.
Cheers
OK. In other words you _do_ want
site_replication_factor = origin:1,site1:1, site2:1,total:2
and the same for site_search_factor.
This will give you one copy at each site for total of two copies. The distribution of buckets within a single site will be managed by CM.
Just remember that within the "source" site the data will _not_ be moved from the indexer that you will be sending it to from your forwarders.
Also - if you happen to lose one indexer within a single site, the cluster will try to replicate the buckets from the other site which might end badly (you will run out of space since you'll be merging data from two indexers into a single one).
Thanks for that information