carloscastilla - Fotolia

News Stay informed about the latest enterprise technology news and product updates.

Azure SQL Data Warehouse brings MPP to Microsoft cloud

As Azure SQL Data Warehouse reaches general availability, it brings massively parallel processing to a wider audience, and augers more competition in the cloud data space.

It joins the fray several years after cloud market leader Amazon's first data warehouse in the cloud, but Microsoft's Azure SQL Data Warehouse is now open for business. The analytics-oriented service scales elastically, and supports massively parallel processing (MPP) capabilities that once were available only to the most advanced IT shops.

"It is an extremely high-performance MPP service, with column store indexing," according to Andrew Snodgrass, analyst at Directions on Microsoft in Kirkland, Wash. "Azure SQL Data Warehouse can put numerous processors to work on queries, returning results much faster than any single server."

For many smaller companies, a data warehouse is still new, and dedicating staff to nurse and feed the warehouse is a burden. Cloud can be a benefit there. But even large companies with established data warehousing programs are currently reviewing their options.

That is one reason Microsoft is promoting the new offering as an alternative to on-premises data warehouses, especially those that focus on producing monthly reports. Today, these systems may have low utilization over much of the month, and then find high use as monthly reports come due.

That is a perfect case for cloud, a Microsoft data leader asserts. In an online blog last week discussing Azure SQL Data Warehouse's move to general availability, Joseph Sirosh, corporate vice president overseeing Microsoft's Data Group, described Azure elastic cloud computing as a means to efficiently marry processing with workload requirements for data warehousing. The service has been available in preview releases since June of last year.

Broad shoulders of column architecture

While some column-based analytics systems go back over 20 years, broad use of the column data architecture was still fairly new in 2012 when Amazon tapped such technology for its Redshift data warehouse in the cloud. Such software was principally useful when data warehouses were required to support large numbers of user queries against their data stores.

Redshift seriously upped the ante in cloud data warehouses, making them more than commodity type products. As Redshift became a prominent part of Amazon's cloud portfolio, it put pressure on Microsoft to add similar capabilities to its Azure line, while at the same time supporting basic SQL Server compatibility. While it has taken a while to achieve, viewers said this release of a Microsoft cloud-based data warehouse is still timely.

"Microsoft is playing catchup to some extent. But it was required to make some significant changes. Their strategy now is cloud first," said Ben Harden, principal for data and analytics at Richmond, Virg.-based services provider CapTech Ventures Inc.

Harden said cloud computing is a very influential trend and that CapTech is now seeing demand for both Amazon and Microsoft cloud implementations.

The wait for Microsoft's cloud data warehouse may have been worthwhile, according to Joe Caserta, president at Caserta Concepts LLC, a New York-based data consultancy that has partner agreements with both Amazon and Microsoft analytics.

"I am kind of glad they waited until they were ready," he said. "They now have a good core set of tools."

Scalability on a scale of one to ten

The ability to scale up to handle peak data loads is a plus for Azure SQL Data Warehouse, according to Paul Ohanian, CTO at Pound Sand, an El Segundo, Calif.-based electronic game developer that has worked with the new Microsoft software. A major use has been to produce analytics to track players' behavior, identify trends and create projections. Scaling up was a concern, he said.

"Our game was featured in the iOS App Store for seven weeks over Christmas. We went from testing a game with about 1,000 people overseas to all of a sudden getting half-a-million users in six days. But Azure SQL Data Warehouse allowed us to very easily scale from something like 1,000 users to 100,000 users," he said. "Literally, we saw that rise in eight hours."

Handling such issues was his team's original goal, according to Ohanian. "When we shipped our game, the scaling totally worked," he said.

Ohanian said his group has been an Azure cloud user for a number of years, but that it looked at other cloud and analytics alternatives. Affinities between APIs already in use and APIs for Azure SQL Data Warehouse were a factor in choosing the Microsoft software, he said.

By his estimation, the cloud data warehouse service is not too late to the fair, and it is full-fledged in important areas such as management.

"Maybe it's because it is coming later, but it seems to have streamlined some of the complexity," Ohanian said. "It is easy to get things up and going, and to manage."

Ohanian found favor with Azure SQL Data Warehouse pricing, which separates expenses for storage and computing. Starting Sept. 1, data storage will be charged based on Azure Premium Storage rates of $122.88 per 1TB per month; at that time, compute pricing will be about $900 per month per 100 data warehouse units, according to Microsoft. Data warehouse units are the company's measure for underlying resources such as CPU and memory.

It is important for Microsoft to field data warehouse products suited for its established users, so, naturally, C# and .NET developers continue to be a target of Azure cloud updates.

"We are seeing pretty equal demand between Amazon and Azure clouds. It is boiling down to what skill sets users have today," said consultant Harden. "People are taking the path of least resistance where skills are concerned."

Path of least resistance leads me on

The availability of Microsoft's massively parallel processing cloud data warehouse augers greater competition in cloud data, which is welcome by some.

An example is Todd Hinton, vice president of product strategy at RedPoint Global in Wellesley, Mass., a maker of data management and digital marketing tools. He is not alone in saying greater competition in the cloud data space could be good, or being unwilling to pick a winner just yet.

"I think they are going to be head-to-head competitors. You have Amazon fans and you have Microsoft fans. It's almost like the old operating system battle between Linux and Windows. For our part, we are data agnostic. We will be interested to see how Azure SQL Data Warehouse shakes out."

Like others, his company's software supports both Amazon and Azure. The company offers direct integration with AWS Redshift already, and he said he expects it will offer native support for SQL Azure Data Warehouse later this year.

Competition in cloud data warehouses goes further too, with players ranging from IBM, Informatica, Oracle and Teradata to Cazena, Snowflake Computing, Treasure Data and others. While it is not as hot as the Hadoop or Spark data management cauldrons, releases like Microsoft's show the cloud data warehouse space is heating up.

Next Steps

Get the DBA's perspective on Azure SQL Data Warehouse

Review implementation details for SQL Server upgrading

Learn about Azure machine learning

Dig Deeper on SQL Server Data Warehousing

Join the conversation

3 comments

Send me notifications when other members comment.

By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy

Please create a username to comment.

What criteria guide your use of data warehouses in the cloud?
Cancel
Nice article. I'm not using the cloud but want to know more.  Currently our department has no future plans for the cloud and I'm try to figure out why they are against it.
Cancel
Price performance ratio and security features could be paired as two vital aspects to evaluate the mpp data warehouses in cloud as a PAAS solution. The products should be benchmarked based on pricing and performance of their minimum compute and storage units.
Cancel

-ADS BY GOOGLE

SearchBusinessAnalytics

SearchDataCenter

SearchDataManagement

SearchAWS

SearchOracle

SearchContentManagement

SearchWindowsServer

Close