Google Cloud SQL debuts, following Amazon's RDS lead

Two and a half years after offering a limited preview, Google has finally opened up Cloud SQL, its managed database service, to all comers. What’s more, it includes a service-level agreement — a sign Google wants users to think of this as a serious business product, not a toy for running a WordPress site.

Many other Cloud SQL features are also eye-opening. The smallest instances of the service can support databases up to 500GB, previously limited to 250GB, with multizone replication, backup, and encryption both inside and outside (via SSL) Google’s networks.

Dig into the FAQ for Google Cloud SQL, and you’ll get the skinny on that SLA, complete with some generous terms. “Downtime” for the service is described as any period with a server-side error rate of greater than 20 percent. If your instance of MySQL doesn’t deliver at least 95 percent uptime for the month, you could get as much as 50 percent of your bill credited back to you. However, this doesn’t happen automatically in the event of an outage; you have to notify Google technical support within a 30-day window to receive those credits.

But several details about Google Cloud SQL jump out; the first and most striking is Google’s choice of database engine. Google stuck with MySQL itself, albeit version 5.5 rather than the 5.6 update Oracle is currently shipping. Granted, MySQL 5.5 is the most widely tested and best-understood version, but why Google opted for MySQL over its increasingly popular and fully binary-compatible fork MariaDB — which isn’t troubled by Oracle’s looming shadow — is another question. Also, some MySQL features aren’t supported, such as user-defined functions.

Another possible issue: The lowest instance sizes barely seem worth the money. The smallest instance, the D0 size, costs a mere 2.5 cents per hour (that’s $18 per month for those keeping score at home), but comes with a pitiful 125MB of RAM and 500MB of storage — and a limit of only 25 concurrent connections and queries.

The big competition here is with Amazon, of course, which already peddles its own database-in-the-cloud services through its RDS offerings. RDS went live for all users last June, although its cheapest MySQL instances are comparable to the cost of Google’s, at 2.5 cents per hour. But like Cloud SQL after it, RDS also offers a 99.95 percent uptime guarantee and a broader selection of databases: MySQL, Oracle, Microsoft SQL Server, and as of last November, a PostgreSQL-based solution.

For Google to really compete, it’ll need to do that on the value of its entire package, rather than instance price alone. The more it can automate all the duties normally left to database admins — such as “sharding as a service” — the better a chance Google stands of looking like a leader and not a follower.

This entry was posted in SQL Server and tagged . Bookmark the permalink.