[prev in list] [next in list] [prev in thread] [next in thread] 

List:       cassandra-dev
Subject:    what's a reasonable upper bound on PreparedStatement caching?
From:       Michael Cohen <mcohen01 () gmail ! com>
Date:       2013-08-19 6:34:36
Message-ID: CAPrQMJQ+Wrsw4km=GTbLYmv35sLam4+U7AeTwVdH8xkmX=g-=Q () mail ! gmail ! com
[Download RAW message or body]


If i'm doing some CQL query like:

select * from table
where id in (?,?,?,?.......)

but the bind parameters are variable length, because the keys are actually
stored in a set<timeuuid> column of another CF, I would potentially have
many, many PreparedStatements being compiled (and cached I assume),
potentially as many as I have rows in the other CF which contains the keys.
Is this a problem? Will the Java driver just evict them as needed from
something like an LRU cache? (If it helps, the 90th percentile will have <
500 keys.)

(The reason I've modeled things this way is because I want to avoid
creating a secondary index on the CF.)


Thanks!
Michael


[prev in list] [next in list] [prev in thread] [next in thread]