Updates to Google cloud database services aim to bolster the platform's appeal as an alternative to on-premises deployments.
Managed databases are one of the most hotly contested subsets in the public cloud market. To stay competitive, Google rounded out its stable of offerings on Google Cloud Platform (GCP) this week with four important updates:
- commit time stamps in Cloud Spanner;
- a fully managed Redis cache;
- PostgreSQL support in Cloud SQL; and
- cross-zone replication in Cloud Bigtable.
Cloud Bigtable, Google's managed NoSQL database, has been available since 2015, but traditional enterprises have been reticent to adopt the service due to a lack of replication to protect against failures in a given zone. This new feature, currently in beta, improves availability, throughput and durability by automatically copying data across multiple zones within a region. It also can route traffic to different clusters to isolate workloads.
Bigtable is most frequently used for time-series data, monitoring and IoT, according to Google. It competes against Amazon DynamoDB and the more recently released Microsoft Azure Cosmos DB.
"These [NoSQL] databases are now being seen as mission-critical, so therefore, it really requires a lot of pieces that you take for granted," said Tony Baer, an analyst at Ovum. "Google is saying, 'We're making Bigtable a serious database.'"
Cloud Spanner is Google's globally distributed database that's based on internal Google technology. The ability to time stamp commits is one of the most requested features since the service became generally available last May, Google said. It gives users more direct access to the TrueTime API -- Google's distributed clock for all its servers -- to automatically identify when a transaction is committed. These time stamps can then be used to determine mutation orders and log any changes for auditing.
Similarly, the addition of PostgreSQL support, which was first disclosed more than a year ago, addresses the biggest feature request from Google cloud database customers. Google's version offers 99.95% availability guarantees, replication and instances with up to 416 GB of RAM.
Merv Adriananalyst, Gartner
PostgreSQL functionality is not on par with that of Oracle, but it has become the de facto alternative, Baer said.
"With each of the cloud players offering its own managed services, this database is really starting to come out on its own as opposed to being the white-labeled engine behind something else," he said.
It's particularly important for the Google cloud database portfolio to attract Oracle's massive install base to its public cloud. Unlike Microsoft Azure and AWS, Oracle's cloud databases aren't available on GCP. Forming a partnership with Oracle creates some headaches for Azure and AWS, but Google acknowledges it's important to make it available on its cloud at some point.
Cloud Memorystore for Redis, also in beta, is a fully managed, in-memory data store compatible with Redis protocol. It's a check box item for Google, as Redis is popular for memory caching.
GCP and the fight for database domination
These moves aren't exactly catch-up for Google, but rather the latest step in an arms race among the major cloud players, which have emphasized enterprise database migrations to their platforms, particularly to one of their managed services. The allure of these services is that they remove much of the grunt work related to infrastructure upkeep, which frees staff to build and deploy applications faster. AWS, the market leader in IaaS and PaaS, claims its Aurora database offering is its fastest growing service ever.
Google remains at the bottom of the so-called Big Three cloud vendors behind AWS and Microsoft Azure, though GCP receives plaudits for its technical prowess. These updates are the latest example of Google as it branches out to attract a more traditional business IT audience.
"What you're seeing is the continuing refinement and completion of a full portfolio of data management services, which is essentially what all the big cloud platform providers are doing," said Merv Adrian, a Gartner analyst.
IBM hasn't made much headway in this space, while Oracle will likely do fine maintaining its customer base, but may struggle to bring in new business, Adrian said. Organizations need to make new architectural choices as they move to cloud, so they're not turning to their old partners by default. Instead, customers are willing to examine alternatives -- and the Big Three, and possibly Alibaba, have the financial strength to separate from the rest of the market, Adrian said.
"We're looking at a generational shift here," he said.