Now that Microsoft is, in theory, giving away part of its architecture via Azure Stack, Google is also electing to monetize a key ingredient of its infrastructure.
Google Cloud Bigtable is a productized version of the NoSQL database that stores Google's bits and bytes. The big selling point: It doesn't require the maintenance traditionally needed for compatible on-prem NoSQL solutions.
Bigtable may ring a bell for some users: It's the name for the storage technology Google described back in 2006, one of the key elements in the company's secret data-processing sauce. The performance claims made by Google for Cloud Bigtable include single-digit millisecond latency and "over twice the performance per dollar" of self-managed editions of comparable NoSQL solutions.
Aside from being a time-tested and Google-scaled technology, Cloud Bigtable offers another vaunted boon with a familiar data interface: the Apache HBase API. Google claims data can be moved back and forth between Cloud Bigtable and a private instance of HBase, implying that data stashed in Cloud Bigtable can be extracted with minimal fuss -- though, one would presume, with accompanying network egress costs. (Ingress to Cloud BigTable is free, although storage, networking, and processing nodes all incur separate bills.)
Likely its biggest appeal, Cloud Bigtable provides HBase-like storage without the management overhead. HBase may be massively scalable, but its power comes at the cost of great complexity. Thus, Cassandra has shaped up as a desirable alternative to HBase, and now Cloud Bigtable could provide competition, albeit from a different direction.
Cloud databases of massive scale that don't require manual maintenance are standard issue, whether or not they're backed by an existing cloud vendor. Microsoft has HDInsight; Amazon has Elastic MapReduce. Other managed solutions for massive data storage, within the conventional SQL space, are also emerging: Snowflake, a startup co-founded by ex-Microsoft vet Bob Muglia, provides a SQL-compatible data warehousing solution that can scale automatically based on demand.
HBase's developers are aware of the work ahead in making HBase easier to use. The project recently celebrated the release of its 1.0 revision after four years of development, with stability and a clean client API as key prerequisites. At this rate, Google and other cloud vendors might save them the trouble -- even if it comes at the cost of agreeing to Google's terms, rather than on Apache's.