AWS Lambda garners interest, production workloads as serverless world evolves

AWS Lambda is about to grab more case studies and production workloads as no-ops gains traction. Toss in other efforts from cloud vendors and even managing a cloud server seems absurd.

Amazon Web Services has the potential to do to cloud server instances what virtualization and cloud computing did to physical servers. In short, AWS may be rendering the management of any server--virtual or physical--obsolete.

Meanwhile, other cloud vendors are following AWS into the serverless world and ultimately you can see where managing instances won't make sense.

AWS' Lambda service hasn't been publicly available long enough for enterprise architecture planning cycles so it's a bit early to predict a groundswell of interest. However, cloud native consulting firms such as Appirio are seeing demand.

TechRepublic: AWS vs Microsoft Azure: Understanding the serverless application trend

In a nutshell, Lambda is an example of a so-called no-ops environment. A no-ops environment relegates server management--in an AWS case an EC2 instance--to something that just happens in the background. AWS handles provisioning and management based on triggers and demands for an application. There's no administration needed to run code and that appeals to mobile, analytics and Internet of things workloads.


Special Feature

Building the Software Defined Data Center

There are massive efficiencies, agility, and manageability benefits from virtualizing your data center and running it from software.

Read More

"It's very early, but the customer interest is real and we're recommending Lambda. We expect to see production workloads based on Lambda in the next six to nine months," said Glenn Weinstein, Appirio's senior vice president of global services and CIO. "We're about as excited about serverless as we were about moving from on premises to the public cloud. Serverless supports that same underlying advantage--getting rid of servers."

Weinstein argued that infrastructure as a service will increasingly look like software as a service. If you use Salesforce or Workday there's no thought given to the servers in the background. IaaS is headed in that same direction. Now that Lambda has traction Microsoft is prepping Azure Functions, which is a spin on a no-ops take on infrastructure, and IBM has OpenWhisk. Google is also testing a similar service.

Special Feature

Cloud Computing: Moving to IaaS

Infrastructure as a Service providers make a very compelling argument for businesses to stop running their own data centers and simply purchase server capacity on-demand and scale up and down as needed. This is our deep dive on IaaS strategy and best practices

Read More

A serverless approach may not work for all workloads, but the functions that are growing--IoT, mobile, analytics--are likely to be no-ops first.

Bustle CTO Tyler Love is a proponent of a no-ops approach. Bustle is a network targeting millennial women and the aim was to scale the site as it grew. AWS and Lambda enabled Bustle to scale quickly and going serverless made sense. "When you go no-ops you're never paying for unutilized capacity and servers shut off," said Love, who noted one of the biggest challenges of managing your own cloud servers is turning compute off when it's not needed.

The real win of going serverless to Love is focus. "Engineers without having a lot of domain knowledge in operations are able to build web sites without the expertise," said Love. "We're thinking about functions to execute something not using a tool set."

Bustle is an early reference customer, but AWS is building out its Lambda case studies now. VidRoll, Localytics and MLBAM along with the Washington Post and Zillow are all early Lambda customers.

Perhaps the biggest case study is already in the family. Alexa runs on Lambda, said Matt Wood, AWS' general manager of product strategy. "Lambda manages scale and availability and is useful for applications that need scale quickly or for functions that run infrequently," said Wood. "Maintaining a stack for things like that gets expensive."