Microsoft Revamps SQL Data Services (SDS) Cloud Database
A pure relational database in the cloud will be offered by Microsoft via its forthcoming Azure Services Platform.
After months of criticism that the test build of its cloud-based SQL Data Services (SDS) lacked the horsepower for enterprise-grade applications, Microsoft is scrapping the effort and moving to a pure relational model instead.
Microsoft made the surprise move just one week before its annual MIX09 conference in Las Vegas last month. Ironically, The company announced its cloud-based database offering at MIX08, one year before. The test version of SDS that critics panned was based on REST and SOAP Web services interfaces.
Eliminating Web Services
Under the new plan, Microsoft will offer a pure relational database in the cloud by exposing its Tabular Data Stream (TDS) over-the-wire protocol for accessing SQL Server via its forthcoming Azure Services Platform.
Microsoft will support traditional relational database capabilities, including SQL queries and support for relational schema and stored procedures. While the company maintains that was always the plan, it said it will no longer expose those capabilities by requiring the REST Web services interface.
The move eliminates the Web services layer and exposes TDS to SQL Server over the cloud. As a result, says Microsoft, developers will be able to move data-driven code developed in Microsoft’s T-SQL language for SQL Server over to SDS.
“We were calling it SQL Server but it really was not similar to a SQL Server-type experience,” explains Niraj Matrani, senior product manager for SDS. “Customers preferred more with the traditional T-SQL-based support, so we decided to go in this direction.”
Microsoft is suggesting that those who want to use REST-based programming for applications that do not require a relational database management system can use the Azure table storage. “You can still access your relational data (located on premises or in the cloud) via HTTP/REST using the ADO.NET Data Services framework,” Microsoft Senior Program Manager David Robinson wrote in a blog posting.
Indeed it was the similarity between Azure Tables for storage and SQL Server tables using the Entity Attribute Value (EAV) model that may have helped render Microsoft’s first test build of SDS dead on arrival. “The two were virtually identical,” says consultant and Microsoft MVP Benjamin Day of Brookline, Mass.-based Benjamin Day Consulting.
The four-month test of SDS failed to meet the performance expectations of enterprise developers such as OakLeaf Systems Principal and VSM contributor Roger Jennings, who found it was unable to offer the scalability of the existing SQL Server. “They offered some pseudo-relational features but they didn’t offer the ones that people really wanted,” Jennings explains.
It also became clear that developers weren’t going to redevelop their T-SQL application for SDS. “It highlights that challenge [that], as people transition to the cloud, they’re absolutely not going to accept two different programming models—one for inside the firewall and one for outside the firewall,” says Forrester Research Inc. analyst Jeffrey Hammond.
The new offering should be welcomed by developers, says Day: “It looks like it’s going to make it extremely easy for you to deploy existing code in the cloud.”
But he cautions that until the test release is available, it’s too early to say how well it will perform compared to on-premises versions of SQL Server. “We’ll have no idea what it will be until we run it,” he notes.
Jennings is skeptical. He pointed out that TDS was designed to run over high-speed LANs and that it’s not an Internet-friendly protocol. But Matrani insists Microsoft has already conducted extensive benchmarking and testing and performance shouldn’t be a problem. “We think it’s appropriate for what we are doing and the direction we are talking it,” he says. “As we get more early-adopter customers and we look at the type of workloads they’re building, they’ll keep modifying and tweaking our protocols so it’s more workload-friendly.”
Matrani adds that SDS will lend itself well to CRM applications, content management, product lifecycle management, supply chain and collaboration. But he acknowledges that it won’t initially be suited to data warehousing and OLTP.
Day and Jennings also warn that relational data has its performance limitations. “There are theoretical limits on how scalable you can make a relational database,” Day explains.
One of the unanswered questions is that of cost. While Microsoft still hasn’t disclosed how it is pricing SDS, rival Amazon Web Services has already lowered the bar. The company slashed the price of its EC2 service by offering those who sign one- or three-year commitments to usage-based pricing that ranges from 3 cents to 24 cents per hour, depending on configuration. That represents a 30 percent to 50 percent reduction, the company says.
Microsoft says SDS with TDS support will be available for testing by mid-year and commercially available by the end of the year.
Jeffrey Schwartz is editor of Redmond magazine and also covers cloud computing for Virtualization Review's Cloud Report. In addition, he writes the Channeling the Cloud column for Redmond Channel Partner. Follow him on Twitter @JeffreySchwartz.