Joyent, a unit of Samsung that provides cloud computing services, is teaming up with LunchBadger to deploy a serverless computing framework based on an event-driven architecture on top of Kubernetes clusters.
LunchBadger CEO Al Tsang says the company has developed a “shim” framework that can support multiple open source serverless computing frameworks. The plan is to employ the framework to support one or two of those frameworks by default. But IT organizations would have the option of swapping in other open serverless computing frameworks over time, he says.
The LunchBadger serverless computing frameworks is comprised of Express Gateway, an API gateway built entirely on Express.js, microservices composition and integration tooling, developer portal and IDE for writing functions that all gets deployed using a Kubernetes Runtime.
Shubhra Kar, vice president of product for Joyent, says the cloud service provider initially will make instances of the LunchBadger serverless computing framework available on the Triton Kubernetes instance on its private cloud. Over time, Joyent will extend that support to instances of Triton Kubernetes running in various public clouds, in addition to allowing IT organizations to deploy Triton Kubernetes in their own data center, he says.
Kar says Joyent expects IT organizations to employ serverless computing frameworks in specific asymmetrical use cases such as bursting workloads to the cloud or to take advantage of graphical processor units (GPUs) that are expensive to employ for hours at a time. Those services will be accessed using both functional programming languages directly as well as via legacy programming languages that will be able to call a function to spin up a stateless process, he notes.
While there is no shortage of serverless computing frameworks, Kar says frameworks that are optimized for Kubernetes should win the day because that approach makes it possible for IT organizations to take advantage of multiple clouds. Rather than getting locked into a proprietary implementation of a serverless computing framework such as AWS Lambda, many IT organizations are already moving to ensure they can port a serverless computing framework between clouds even before such frameworks are widely employed.
Kar acknowledges that as cloud computing continues to evolve, developers are being presented with a dizzying array of options for building and deploying application workloads. Over time, more instrumentation will be injected into application workloads that will enable cloud service providers to better optimize them using various classes of infrastructure resources such as GPUs, field programmable gate arrays (FPGA), virtual machines, bare metal servers and serverless computing frameworks, he says.
It’s not clear to what degree serverless computing frameworks built on top of containers will be employed. Developers are keenly interested in serverless computing frameworks because the event-driven architecture they are built on allows them to seamless invoke additional IT infrastructure resources as their application scales up and down. That should make many of the manual DevOps tasks associated with managing IT infrastructure disappear. But not every application workload lends itself to an event-driven architecture. The challenge facing DevOps teams now will be figuring out where precisely to apply serverless computing frameworks in way that make the most sense.