Serverless technology has infiltrated startups with a significant cloud presence, but in 2018, look for increased adoption by mainstream enterprises.
AWS Lambda has led the serverless market since its 2014 launch, but today the offering is aggressively pursued by Google Cloud Functions, IBM Cloud Functions and Microsoft Azure Functions.
In addition to the commercial activity, there is an open source movement to create a standard for serverless computing that works across all clouds. OpenFaaS, or open source function as a service, aims to make the best of serverless without the lock-in. Other open source initiatives include Apache's OpenWhisk, which underlies IBM's Cloud Functions and Fission, which is a framework for serverless functions for Kubernetes.
The overall serverless technology market is nebulous. For one thing, it's hard to directly compare serverless to other cloud services, and not all serverless activity is priced the same. In general, prices are based on compute cycles used. So when a particular task is compute-intensive, such as image processing, it costs more. If a task merely kicks off a simple queue service such as sending an email, it would cost less, said Larry Carvalho, research manager at IDC in Framingham, Mass.
In terms of adoption, Amazon claims a 300% year-over-year growth. It's safe to say that serverless technology has grown by leaps and bounds, although the evidence is largely anecdotal.
"We are seeing an increasing number of startups going on a fully functional-based computing model and enterprises using functions for new applications," Carvalho said. "Those customers already on AWS are using it where it makes sense when adding functionality to existing applications."
Where does serverless belong in enterprise IT?
The cloud provider is responsible for server resources behind a serverless service. The customer's application consumes only the resources it requires. The application's code runs as a 'function' -- often in response to some event or trigger -- and customers are charged per executed function. It's particularly attractive for new product launches wherein throughput is difficult to estimate.
"It scales, so IT shops only pay for what they use," said Chris Moyer, CTO at Cannabiz Media, a marketing company that tracks marijuana licenses, among other things.
Serverless technology lets developers focus on code, and frees IT from having to manage an entire stack. Usual concerns about vendor lock-in remain, even though each provider has opened up to allow more types of code, which eases the ability to transport applications to another platform.
The majority of enterprises today are not as worried about lock-in, as most have limited workloads in the cloud. To see meaningful serverless computing uptake in the enterprise, IT pros want to see more than just code execution -- a lot more, such as application lifecycle management capabilities or security features, IDC's Carvalho said. "Once you have that, then you will see more adoption," he said.
For IT pros slow to appreciate the value of serverless computing, two recent Intel vulnerabilities, Spectre and Meltdown, gave an example of the technology's benefits.
IT pros patch their own cloud apps in traditional hosting scenarios, whereas a service provider automatically patches a serverless app when issues occur in the stack below app code. And the Spectre and Meltdown patches themselves caused further problems with specific systems, creating even more confusion and headaches for IT pros. Further, patching can degrade performance, but serverless users might avoid this issue by enforcing a service contract with the cloud vendor.
"It was a real 'aha' moment for IT, as responsibility for system patching shifted from IT to the cloud provider," Moyer said.
And that is true for any serverless service, since the developers don't control the underlying infrastructure.