July 7, 2015

Basho, a scalable data platform

Basho Technologies (www.basho.com), an innovator in distributed database, participated recently to the IT Press Tour last June and made a very good refresh of their portfolio and company history and background. Founded in 2008 and inventor of Riak, a distributed database, Basho leverages the large market adoption to address different needs with variety of Riak. In EMEA, the expansion is driven from London and 40% of total company bookings come from Europe. Riak KV is the famous NoSQL database and Riak S2 for Simple Storage, renamed from Cloud Storage, is the object storage offering built on KV. Basho has more than 200 customers worldwide. With a new team on board for several months now, we feel a new wave at Basho with new initiatives and product development. Same root as other distributed database approaches, all came from internal development made by internet giants that didn't dedicate VC money to buy expensive Oracle DB and didn't find any commercial products that matched their demanding needs. Scalability was and is one of the top difficult attribute to deliver. The capability to scale horizontally the data structure is for several years a de-facto standard with the capability to anticipate failures with proactive mechanisms: the resiliency of the whole system is a key attribute.
In term of product, Basho, like a few other players, uses a ring topology aka as a master less or symmetric architecture. With a ring scalability and high availability are immediately solved. For data protection as Riak KV is a distributed DB, the replication with 3 copies is used to ensure maximum durability of data. The ring model offers a near-linear scalability able to deliver a predictable response time and workloads as nodes number increases. The other good point for Basho is the open development environment with various APIs and Clients in multiple languages. Riak S2 is the new name of Cloud Storage, built on KV, with the capability to split files in 1MB chunks. Each S2 nodes streams chunks to KV nodes.
Recently, Basho made a new iteration with the introduction of the Basho Data Platform with clear services models and instances with Riak KV and S2 but also a new one to be introduced in the future. The services instances will be things such Spark, Solr or Redis and the storage instances are KV or S2. Between these 2 layers, Riak Core Services offer multiple features such replication and synchronization, message routing, cluster management or logging. The idea behind the data platform is to offer a base layer with advanced services for enterprise data applications.


0 commentaires: