With a trio of new features, Microsoft Azure continues to build on its position as the top alternative to the king of public cloud, Amazon Web Services.
Microsoft followed through this month on new features in Azure around security, massive virtual machines (VMs) and containers. And with claims that it adds 10,000 new Azure customers a week, Microsoft has reached the point where it has a fully functioning cloud platform, said Carl Brooks, an analyst with 451 Research LLC, based in New York.
None of the new features stands out above major competitors, but they are all positive additions, Brooks said. The release also shows the historically slow-moving Microsoft is light years ahead of its typical rate of product development -- a sign the vendor has gotten something right with Azure.
"These are all signs of Microsoft hitting its stride and doing what it said it would do," Brooks said.
Carl Brooksanalyst, 451 Research, LLC, based in New York
Azure customers can now use Docker engine on Ubuntu images via the Azure Marketplace. Docker support was promised last year, but until now customers had to use an extension model. Microsoft also plans to offer Docker support in the latest version of Windows, which is expected to be released sometime later this year.
Google and Amazon have offered Docker support since last year, while smaller players such as Joyent, Inc. have based entire platforms around the container technology. But that lag shouldn't hurt Microsoft because while there's plenty of talk around Docker, there's not much evidence that customers are using it in production yet, said James Staten, an analyst with Forrester Research, Inc., based in Cambridge, Mass.
"They're behind in the hype cycle, but in terms of actual adoption of Docker they're not really behind at all," Staten said.
The ability to run Linux containers in Azure also points to Microsoft's competency to run non-Windows workloads, and backs up the rhetoric about embracing more open source products, Brooks said.
Internal Forrester survey data shows a growing number of non-Windows workloads going up in Azure, Staten said.
"There's still a bias in their audience of people already in the c-sharp camp, but what it shows is those people are able to convince their colleagues in the company and people building with Linux and open source they can play at that game, too," Staten said.
Big VMs and central key control
The Azure cloud additions are largely catch-up, which is to be expected as Amazon, Google and Microsoft continue to leapfrog each other with new features, Staten said. The new G-series VM sizes, however, are in some ways ahead of the competition with 32 vCPUs, 448 GB of RAM and 6.59 TB of local SSD space.
The most common customer for the G-series will be in the big data space, including partners such as Cloudera Inc., and XtremeData Inc., that need large amounts of fast local storage and a platform that can handle large-scale data replication, a Microsoft spokesperson said.
XtremeData, which operates a big data analytics platform based on an SQL engine and is available through most large public providers, was given an early preview of the G-series and has been working with it for several weeks. So far, the technology looks very promising, including networking speeds of almost 2 GB per second, said Ravi Chandran, CTO for the Chicago-based company.
"What we need is a combination of computing resources balanced with a sufficient bandwidth to the storage device and sufficient bandwidth on networking to exchange data between nodes," Chandran said. "G-series has abundant resources on all three fronts."
XtremeData has partnerships with most of the large cloud vendors, including Amazon, CenturyLink, IBM and Microsoft.
The third new Azure feature is Key Vault, which is in public preview. Key management is important to many enterprises, and customers will have centralized control over the security feature to encrypt data with asymmetric keys in Hardware Security Modules.
Previously, Azure customers could bring their own keys, but this is the first cloud-based key system on the platform.
All three of the new feature sets are of interest to XtremeData. And while there are areas Microsoft still must improve on to keep pace with AWS, it's been impressive to see the growth of Azure, Chandran said.
"What we're seeing in the last 12 months is a lot of the big guys are rapidly catching up to AWS," Chandran said. "I'd have to say Microsoft has probably done the most with how rapidly they're improving."
Curating cloud competition
Azure is still behind with some of its networking functions and identity management capabilities, but those gaps are already on the roadmap to be addressed, Staten said. And while Azure has gotten better at big data by embracing Hadoop and third-party tools, it also lacks a true competitor to Amazon Redshift, he added.
Azure also had a well-publicized outage in November and fell behind the competition in that area last year.
Still, Microsoft will generate the highest growth rate of any of the major cloud vendors in 2015, according to Forrester. Internal survey data shows it behind only Amazon Web Services (AWS) in the U.S. in new growth, while it's actually outpacing AWS by a considerable margin in Europe, if customer survey responses are to be believed, he added.
"They're a very solid number two," Staten said.
Microsoft has taken a curator approach to Azure, which can be seen with its video service pack, mobile backend as a service and developer tools, Staten said. AWS users that signed up four years ago can deal with the incremental additions, but customers that sign up today can be incredibly overwhelmed by navigating all the options once they get past the basics of EC2 instances and S3 storage.
"Azure is doing a better job of curating and guiding you to the right capabilities while not diminishing the breadth of services to you," Staten said.
That should be a concern for AWS this year because more developers are moving to public cloud, especially at large enterprises where employees have a mix of skill sets, Staten said. Most of those developers are just looking for best practices and don't want to build and configure workloads around DevOps, nor do they have the necessary skill sets.