Serverless computing is gaining attention as the latest way for organizations to drive greater efficiency from their cloud deployments. However, Capgemini notes significant hurdles remain in seeing greater adoption of serverless compute platforms.
Serverless computing in general is designed to reduce the amount of overhead associated with offering services in the cloud. This includes the ability for a cloud provider to dynamically manage server resources.
Charlie Li, chief cloud officer at Capgemini, noted the business case and pricing models are among many challenges stymying adoption by organizations. Li explained that clients were initially getting into the cloud business as a way to take cost out of their operations. Many of these initial forays were hastily done, leaving enterprises with programs that did not save as much money as expected.
“I would argue that from what I have seen, poor governance and policies have actually increased the cost for some of these organizations,” Li said.
In addition, there are so many options in the market that enterprises aren’t sure which way to go. Li said organizations that are dedicated to a single cloud platform and have no desire to add a hybrid component to their operations can overlook this issue.
“If you have a full Microsoft shop and are all-in on Azure, then there is no problem,” Li said. “But with multi-cloud deployments you run into issues. It can be done, but it’s not cheap.”
With serverless computing, an organization can become even more reliant on the performance scaling and monitoring capabilities of a particular vendor. Cloud vendors offering serverless compute platforms include Amazon Web Services (AWS), Microsoft Azure, IBM, Google Cloud Platform (GCP), and Oracle.
Li said this confusion has led to slower than expected adoption of container and serverless computing platforms.
A recent Cloud Foundry Foundation survey showed the percentage of respondents using containers at their company had increased from 22 percent in 2016, to 25 percent this year. Those evaluating use of containers increased from 31 percent to 42 percent.
However, when compared to a similar survey conducted by Evans Data earlier this year, the Cloud Foundry Foundation report indicated slower-than-expected adoption of containers over the past year.
“There has been no dramatic increase in broad deployment of containers by companies over the past year,” the report noted. “We see a significant increase in interest among Cloud Foundry Foundation survey participants (Evans does not), but actual adoption and deployment has seen either marginal growth or even a significant drop, based on Evans data.”
Serverless Pricing Challenges
Li said serverless computing is also being hampered by a lack of flexibility on the side of cloud vendors in developing pricing models to drive adoption. In reducing potential waste and taking advantage of their diminutive resource needs, serverless computing allows for pricing models that typically break down access by tenths of a second.
“The public cloud providers want to eek out as much margin as possible, so they have not been as efficient in the backend with their serverless options,” Li said. As an example, he said there is little in the way of choice for organizations looking to use serverless computing as a backup solution to broader container deployments.
“There has to be a pricing model for something that is operating only as a backup, but not being charged for having those services on all the time,” Li said.
A report earlier this year from 451 Research found that it was difficult to nail down exact costs of serverless computing, which highlights the confusion cited by Li. But, in general, 451 Research found IBM to be less expensive than rival services from Microsoft, Google, and AWS.
IBM’s cost advantage was on usage based on tenths of a second, up to around two seconds of total usage per script. Helping to bolster IBM’s value is that the company does not round up usage to the nearest usage marker, which 451 Research found to be in 128 megabyte buckets.
“IBM has a simple model,” said Owen Rogers, research director for 451 Research’s Digital Economics Unit. “For very small duration scripts, it’s pretty much cheapest all the time, or at least in our parameters.”
For longer usage scenarios, 451 Research found Microsoft’s Azure platform could offer a less expensive product.
Li explained this market confusion has handicapped the adoption curve of serverless compared with that of containers.
“We will see in a year,” Li said. “It may even be two years before there is mass adoption of serverless. All of our clients are expressing interest in trying serverless, but outside of a few new applications being launched and some PoCs, mass adoption is still some time away.”