Microsoft's latest additions to its public cloud exemplify how it plans to make cloud technologies more accessible...
to a broader audience.
Microsoft made two moves this week to bring its advanced cloud services to a wider variety of enterprises. The company acquired Cycle Computing, with plans to incorporate its high-performance computing (HPC) technology into the Azure public cloud. Microsoft also issued an Azure update to give users more control to deploy serverless architectures.
Largely the domain of computer scientists, HPC is seen as a natural fit for public cloud because of the ability to quickly spin up and down instances for one-off projects that require massive amounts of resources.
Most of Cycle Computing's customers are in industries with heavy compute needs, such as life sciences and insurance. The firm works with clients both on premises and across the major public cloud platforms, with particular growth lately in the latter. This acquisition reflects the desire of Microsoft and other major cloud providers to tailor their platforms for specific vertical markets, said Gordon McKenna, CEO of Inframon, an Ensono company and U.K.-based IT consultancy.
"If you look at the real low-hanging fruit for cloud, it's around high-processing workloads," McKenna said. "A lot of banks are looking to move risk management systems to Azure because they're struggling on premises with the number crunching."
All the major cloud providers have invested heavily in big computing and related capabilities around artificial intelligence and internet of things. Microsoft has added machine types tailored to these projects. And in the past, Cycle Computing executives have praised Azure in particular for its use of GPUs and support for FDR InfiniBand networking standards for high throughput and low latency.
HPC has always been a key driver for cloud adoption, and Cycle Computing carved out an early lead within this category, said Dave Bartoletti, an analyst with Forrester Research. Few platforms have a native, HPC-specific service, so this will help funnel more big-compute jobs to Azure.
"The market for huge compute workloads is also aligning with the large-scale analytics solutions market," he said. "This helps Azure build out a packaged offering to target life sciences, financial services and other data- and compute-intensive industries ... to help encourage migration of more workloads to the cloud."
This move is a critical element of Microsoft's big compute roadmap, which is built around HPC capabilities, said Chris Woodin, senior director of Microsoft and software asset management at Softchoice, an IT consultancy and managed service provider in Toronto.
It also will strengthen Microsoft's support for Linux and make higher-level services more attractive to general enterprises in the same way it did with technologies such as Windows and SQL, he added.
"Microsoft has always done one thing in their history: They democratize advanced technologies," Woodin said.
Very few of Softchoice's Azure customers use the platform for advanced computing capabilities, Woodin said. Mostly, they use it for backup and recovery, test and development, or basic VMs migration, but that will change as Microsoft makes more moves like this.
"They're all definitely intrigued by how they can use advanced capabilities, but they don't know how, and they don't feel like those capabilities are easily accessible to them," he said.
Serverless apps get better connections in Azure
Speaking of advanced services, Microsoft also added Azure Event Grid, a fully managed service that handles event routing and filtering, for users that build serverless architectures. Azure Event Grid has a narrow set of services, but the eventual goal is to handle events from any source to any destination in Azure. It also incorporates web hooks to publish events to third-party services.
Jeffrey Hammondanalyst, Forrester
Azure Functions, a service for running small bits of code, and Azure Logic Apps, a workflow orchestrator, are two examples of serverless tools that remove the need for users to manage the underlying infrastructure. Both services are integrated event handlers for Azure Event Grid, along with Azure Automation and Webhooks.
Azure Event Grid also works with event publishers Blob Storage, Resource Groups, Azure Subscriptions, Event Hubs and Custom Topics, and Microsoft plans to add more event sources and destinations later this year. The first 100,000 operations per month are free; beyond that, it costs $0.30 per million operations while in preview.
The most comparable service in Amazon Web Services is CloudWatch Events, which became available earlier this year. The biggest difference between the two is the ability to set up custom events, said Jeffrey Hammond, an analyst with Forrester.
"That's really important if you're trying to use the service as a basis for an event-driven API architecture, which is kind of what we're seeing the serverless folks do," he said. "It's good to have a list to choose from, which is what you have in [AWS], but it's even better to have that list and add your own things to it."
Vendors such as Salesforce are also going in the direction of embedding events into their platform architectures, and this is an important building block for Microsoft's efforts, he added.
"It's another example of the tit for tat of competition we're seeing from the major cloud platforms to add all the elements of a robust serverless architecture," Hammond said.
Trevor Jones is a senior news writer with SearchCloudComputing and SearchAWS. Contact him at firstname.lastname@example.org.
Get better acquainted with Azure Functions
Find out if your apps are ready for Azure's public cloud
Respond to security threats with Azure Security Center