Have you considered creating a composite key for the various permutations of values that you need to query? For example "good.ok" could contain a column for each entity with the value specifying whether the entity is "good" or "ok". You could the. Have another composite key for "good.bad", "bad.ok" etc.
I've found that in order to reduce reads with Cassandra you often need to write the same data more than once in multiple places to suit your queries. I actually started implementing a bitmap operator at one point but it turned out to require more reads on the backend than it was worth. Cassandra already provides the bloom filter for probabilistic reads based on row keys, I think you want to try to leverage that as much as possible. Jason On Aug 19, 2011, at 10:43 AM, Benjamin Padgett <benjamin.padg...@cision.com> wrote: > What would you suggest for handling such a situation? I come from RDMS > world... > > Is it typical for a Cassandra implementation to make multiple slices to get > the desired data? > > Thanks for the quick response! > > ____________________________________________ > Benjamin Padgett > benjamin.padg...@cision.com > > -----Original Message----- > From: Jonathan Ellis [mailto:jbel...@gmail.com] > Sent: Friday, August 19, 2011 10:38 AM > To: dev@cassandra.apache.org > Subject: Re: Bitmask operators exist? > > On Fri, Aug 19, 2011 at 10:08 AM, Benjamin Padgett > <benjamin.padg...@cision.com> wrote: >> Have a column family with a few secondary indexes and I am trying to find >> records where multiple flags are set using a bitmask. I do not see >> documentation on bitmask operators so I am unsure if they exist. > > They do not, sorry. > > -- > Jonathan Ellis > Project Chair, Apache Cassandra > co-founder of DataStax, the source for professional Cassandra support > http://www.datastax.com
