In a display of marketing fanfare at its AWS re:Invent conference in late 2016, Amazon Web Services rolled out -- literally -- an 18-wheeler. The truck heralded Amazon's intention to move relational data from on-premises databases, like Oracle and SQL Server, to the cloud, even if doing so requires a vehicle containing systems with up to 100 petabytes of storage capacity.
The onstage demo of the AWS Snowmobile, as the truck is known, was over the top, but it was also emblematic of a growing trend in enterprise data management. In many organizations, on-premises databases are giving way to cloud database services.
Nearly all cloud-based databases moved up in popularity rankings calculated by the DB-Engines website during the last 12 months. Meanwhile, research firm MarketsandMarkets has forecasted that worldwide revenue from cloud databases and database-as-a-service platforms will top the $14 billion mark in 2019, up from just over $1 billion back in 2014.
The heightened cloud focus isn't lost on relational database market leaders Oracle and Microsoft. Both are now working to ensure that customers opt for their cloud database services instead of running Oracle Database or SQL Server on the AWS cloud. "We can bring in a truck, too. But it says 'UPS,'" quipped Robert Greene, Oracle's senior director of product management and strategy for cloud data management services.
Whole lotta planning goin' on
The original knock on putting databases in the cloud was that they wouldn't be secure there. But cloud providers have taken steps to lock down data, to the point where many analysts and experienced users now conclude that cloud data centers surpass most individual IT shops on security protections.
As the security concerns have eased, the questions for many IT managers are no longer why or whether to use cloud databases; now, they're how much, how soon and which one. The answers to those questions often start with: "It depends."
Carl Olofsonanalyst, IDC
"People first have to analyze how and which cloud they should go to," said Mike Walsh, founder and owner of Straight Path IT Solutions, a SQL Server consultancy and managed service provider in Milton, N.H. In the case of SQL Server, for example, the choice of whether to go with Microsoft's Azure SQL Database service may revolve around how committed an organization is to the overall Microsoft technology stack, according to Walsh.
"If you're a Microsoft shop, meaning you support both .NET development and SQL Server, Azure makes sense for a lot of reasons," he said. On the other hand, an organization that simply wants to move some virtualized SQL Server instances to the cloud -- "just lifting and shifting" -- might have a preference for doing so on AWS, he added.
The same holds in the Oracle world. The ease of changing platforms versus using the vendor's own Oracle Database Cloud Service is partly dependent on the extent to which an organization's business applications and middleware tools are interwoven with Oracle databases.
Survival of the fittest
Pricing could be a decider. AWS has aggressively priced cloud database services for hosting SQL Server and Oracle Database, although its options for deploying the latter without having to buy separate licenses from Oracle support only Oracle Database Standard Edition, not the full-featured Enterprise Edition. Companies can also use existing Oracle or SQL Server licenses to run the two databases on the AWS cloud -- a deployment model that Amazon calls Bring Your Own License.
But Oracle and Microsoft are responding by increasing their technology and marketing investments in Oracle Database Cloud Service and Azure SQL Database, respectively. In Oracle's case, the counterthrust includes hints of increases in licensing prices for customers that move Oracle Database to AWS -- or, for that matter, Azure.
"Oracle is going to do everything it can to get customers to its cloud," IDC analyst Carl Olofson said. "This is a survival issue for them."
The Oracle Cloud at Customer offering represents another step by Oracle to boost its own cloud platform. Introduced in May 2016, Cloud at Customer lets users run cloud-based versions of Oracle Database and other technologies in their own data centers, with Oracle handling management and maintenance of the system infrastructure.
Creating cloud-ready combinations of databases, applications and hardware on premises is a logical first step toward the public cloud for organizations, said Jim Czuprynski, a consultant at systems integrator Vion Corp. in Herndon, Va. Beyond that, how much companies move to the cloud depends on factors such as required service-level agreements and, at times, security issues, he said.
Easy does it on database management
For Oracle, much depends on how it moves forward on building tools that make deploying databases on Oracle Database Cloud Service more manageable, in Czuprynski's view. "Its interfaces and ease of use for provisioning and orchestration need some work," he said, noting that Oracle has to answer Amazon's portrayal of its own cloud database management capabilities as resembling "a big 'easy' button."
The way expenses are accounted for when using cloud database services can require a further shift in mindset for users, Straight Path's Walsh said. A lot of companies "are used to spending a boatload of money every few years" on capital expenditures to upgrade their systems, as opposed to paying monthly or hourly usage fees, he explained.
In that light, cloud database deployments might seem to be more expensive in the long run. But Walsh said organizations need to look more deeply.
"What you gain is not just a server with power and cooling -- you're also gaining access to a database platform that is already built," he said, pointing to a couple of functional areas as examples: "With Azure SQL Database, you get high availability, and disaster recovery only takes a few minutes to set up." Cloud costs, he surmised, may be more economical when viewed from that perspective.
Learn about hurdles to cloud database implementation
Find out why it's good to be paranoid about security on the cloud
Discover how big data changes the database status quo