Key enabling products technologies

Storage

BioMart 0.8: discarded on long term, it has serious scalability issues hit in ICGC and BLUEPRINT.
HDF5: very efficient at storage level, but discarded, very difficult to build queries as it does not have a query processor
Relational databases with SQL'99 ARRAY type (e.g. PostgreSQL): discarded, not so sparse on dumps or updates
Apache Cassandra: Tested one year ago, with a custom query language called CQL. Although scalable, it was slow storing the methylation test data, so discarded.
MongoDB: It is based on binary JSON documents paradigm (i.e. a hierarchical model). Very scalable, very fast data loads, lot of specialized indexes. As it doesn't have a query language as such, but API's to build streamlined queries, there are some queries easy to be built in SQL (i.e., joins) which much be rewritten in map ... reduce style.
RethinkDB: A promising NoSQL database, which, as MongoDB and other NoSQL databases, is based on JSON documents paradigm. But as it is still buggy, it has been discarded (for now).
MySQL: discarded, slow index builds
tabix tools: fastest than anyone, manages BED, VCF, GFF, SAM, focused on chromosomal coordinate searches; but discarded, very difficult to build queries as it does not have a query processor.
Apache Solr: based on Lucene, very scalable and promising, but as it is focused on incremental results fetch (like in search engines), it is slow when you ask for all the results.
[Elastic Search] (http://www.elasticsearch.org/)