One of the buzzwords to emerge over the past year is that of "serverless" computing or architecture, which, as the term suggests, involves the provisioning of key information technology resources to users without the fuss and muss of acquiring and activating additional hardware, which not only means servers, but disk space as well. Let the cloud vendors worry about the messy details of protocols, security, resource provisioning, processor speeds, and memory allocation, and focus on the applications business users need to run their organizations.
Serverless is, for all intents and purposes, another name for Platform as a Service. There are vendor tools and environments suited for such a purpose, including Amazon Web Services Lambda, IBM BlueMix OpenWhisk, and Microsoft Azure Functions, Buzzwording aside, full-throttle adoption of serverless platforms may even stir rethinking of optimal hybrid cloud architectures, and what it means for IT teams to serve as brokers of needed business services.
That's the experience of Gojko Adzic, a highly regarded thought leader in the IT space and partner at Neuri Consulting, who recently explored his journey down the serverless computing path with his MindMup project. In his post, Adzic provides some food for thought as to the best way to structure the delivery of cloud-based back-end services to a dynamic user base.
MindMup, which offers mind mapping tools, first piloted the AWS Lambda platform in February 2016, and moved entirely over to Lambda at the beginning of 2017, Adzic relates. The site has seen positive outcomes so far in its one-year journey, with a user base increasing by 50% while hosting costs have dropped 50%, he says. Plus, scaling to meet demand is now relatively painless.
Serverless computing is about a platform approach, not just services. Organizations thinking that moving applications to serverless platforms will save money will be disappointed, Adzic says. Even if an infrastructure is deployed across the resources of a cloud provider such as AWS, as it may involve making multiple duplicate payments for connecting web requests, he explains."By far, the biggest lesson for me was to really embrace the platform, not just the service," he relates.
Adopting a platform approach can be accomplished three ways: through the use of distributed authorization: letting clients orchestrate workflows: or allowing clients to directly connect to AWS resources, Adzic explains. MindMup went with the third, direct-to-client, approach, as the first two options have limitations within AWS environments. Enabling direct access to platform services has helped to reduce latency and costs.
As Adzic observes, this model of direct cloud-to-client architecture represents the most expedient way to deliver hybrid services, and he goes on to suggest that this may even change the way enterprises think about hybrid cloud architecture. That is, open up back-end services directly to clients and client applications, rather than structuring layers of services between users and cloud functions:
"When client applications can directly connect to 'back-end' resources, there's very little benefit orchestrating that from anywhere else. Coordination, workflows and many other aspects of an application can move directly to the client application. Only the parts that really need to be locked down for security reasons or to use specialist resources need to go to AWS. The hybrid-cloud of the future isn't going to be a mix of AWS and Google, or AWS and on-premise. It will be a mix of AWS and client machines."