On Thu, Jun 23, 2011 at 12:43 PM, mcasandra <mohitanch...@gmail.com> wrote:
> How should one go about creating a data model from RDBMS ER into Big Table > Data model? For eg: RDBMS has many indexes required for queries and I think > this is the most important aspect when desiging the data model in Big > Table. > > I was initially planning to denormalize into one CF and use secondary > indexes. However I also read that creating secondary indexes have > performance impact. So other option is to create inverted index. But it > also > seems to be bad to have too many CFs. We have requirements to support high > volume min of 500 writes + 500 reads per sec. > > What would you advise? > > -- > View this message in context: > http://cassandra-user-incubator-apache-org.3065146.n2.nabble.com/How-to-create-data-model-from-RDBMS-ERD-tp6509041p6509041.html > Sent from the cassandra-u...@incubator.apache.org mailing list archive at > Nabble.com. > >From a high level prospective secondary indexes are identical to creating your own inverted index. Indexes store data in a different ordering to accelerate certain types of searches. Every extra index requires extra disk space, and each insertion requires updating more structures. In an RDBMS world indexes are usually placed on foreign keys or things in the where clause. You do not want to have unnecessary indexes because that slows the write path. A good quote is "design your data around your access requirements". Thus if you need to be able to quickly search the data on a certain column, you need the index. There is no way around that, if it hurts performance you need more hardware. If secondary indexes in Cassandra do what you need you should use that will save you design time. If they do not do exact what you need you may have to create your own inverted index that involves a second Column Family.
