Yea, it does. This is a suggested solution for solving one of the more
complex aspects of social graphs, but you would need serious cluster of
machines so solve that. It really depends on what you are going to run
against such an index. Most users can live with a 1 shard with 1 replica
index, How many shards can one machine sustain and still be operational
depends on what you do against it, very hard to give a number. But, even
with 1000 shards per machine, you will need 50 machines.
If you don't want to solve the social graph problem presented in that README
case, then all your users can be stored in a single index, or maybe several
indices just chunked differently.
On Fri, Aug 27, 2010 at 9:35 PM, Andrei firstname.lastname@example.org wrote:
The README file hints that ES can be used for social search by
creating an index for each user and then searching on multiple indices
with different boost levels per index. Is this suggestion practical?
Any serious social product is likely to have 50,000+ users, so what
sort of hardware would be needed to support 50,000 indices?
On Aug 27, 3:24 am, Shay Banon shay.ba...@elasticsearch.com wrote:
Which version were you using? In 0.9.0 the index creation was really
its much more lighter in 0.10. Just a note, each shard is a full Lucene
index, so its not very lightweight. The number of indices you can create
depends on the machine you are running on, the amount of memory allocated
elasticsearch, and your expectations from the created system (latency,