Clustrix Discusses the Significance of the Development of Google Spanner
Clustrix, provider of the first scale-out relational database designed for the elastic scaling requirements of high-transaction, high-value workloads of today’s Web applications, recently commented on the beta launch of Google Spanner (read more at http://www.clustrix.com/bettersql/google-spanner-good-better).
According to Clustrix CEO Mike Azevedo, “Google is clearly thinking about the same problem that we’ve been helping companies address: making a SQL database scale both writes and reads past a single server. This is perhaps one of the most important technical issues that companies must grapple with, as it’s a key cause of site outages, and directly affects growth and bottom lines.”
The Challenge of Scaling Writes
According to Clustrix, Google built Cloud Spanner to address an internal reality that many Clustrix customers also face: the unsustainability of sharding MySQL. Like Clustrix, Google Cloud Spanner leverages 2-Phase Locking, MVCC for transactional commit consistency and Paxos for consensus across database servers to provide the horizontal scale-out ability of NoSQL with the relational consistency of an RDBMS. The benefits of such a system are numerous, but as with any RDBMS, IT architects are asking what the tradeoffs with Spanner will be. In an article posted on the company’s site, Clustrix’s Dave Anselmi offered insight on a number of issues related to Cloud Spanner, and how Clustrix addresses them:
Latency. Spanner provides cross-node transactions with ACID guarantees, but the more nodes you add, the more latency climbs (the slower your transactions get). ClustrixDB, as shown by a recent InfoWorld benchmark, scales out with linear latency, so you retain the same performance running on large clusters as you would on a single server.
SQL compatibility. Where Spanner uses custom client libraries making it difficult, if not impossible, to migrate from existing SQL-based databases, ClustrixDB is completely MySQL compliant. Simply import your MySQL dumpfile, point your application endpoint at ClustrixDB, and it will handle your MySQL workload at scale.
Untested territory. While it will be fascinating to see what companies are able to do with Spanner, ClustrixDB is not beta. It’s a production-ready database powering hundreds of applications worldwide, including some of the world’s largest social and e-commerce sites.
Vendor lock-in. Cloud Spanner relies on Google’s computing resources, meaning that if you want to use it, you must do it through Google Cloud Platform. On the other hand, you can run ClustrixDB on any cloud or your own data-center, avoiding vendor lock-in.