App Modernization – Serverless Approach

In the past several years, computing conventions have transformed drastically – shifting from on-premise models to cloud computing, which has now given way to the serverless computing model.

In the serverless model, a developer can take a business requirement and deploy a coded solution straight into the serverless cloud—and it will run to provide business services. Serverless computing completely hides the management of underlying servers and its load balancing, failover, auto scaling, etc. from admins and developers of serverless computing applications.

Value for Business

In this world, traditional admin teams tasked with server management, maintenance, upgrades, and patches are not needed.?While there is still a need for admins in this model, their day to day tasks are quite different. Admins will focus on user and security management, key management, configurations, and DevOps. ?The cloud provider is responsible for managing machines, resource allocation, and providing a reliable computing platform.

Serverless computing also makes it easier for organizations to move closer towards the No Operations (NoOps) model. In a pure NoOps model, there is no need for IT operational resources to provision and maintain servers, deploy code, etc. These operational tasks are completely automated using continuous integration and continuous deployment (CI/CD). Although achieving the purest form of NoOps seems like a wishful task, the latest tools and services from serverless computing vendors make a very high level of IT process automation possible. With this, a virtuous efficiency cycle is created: IT organizations can shift their resources to create and maintain automated process with fewer IT operational resources, dedicating more resources to improve the agility and quality of enterprise applications.?

The biggest change associated with serverless computing is cost. Computing and maintenance costs are both reduced dramatically versus equivalent traditional server-based systems.?As we mentioned in an earlier post, customers are charged only for the computing resources applied to a given service. For example, if a microservice is used sparingly during the day on serverless computing, its cost will be the total of the computing resources utilized while the service invoked– not the 24x7x365 cost of server resources dedicated for that service.?Companies will save even more when they go serverless due to the lack of need to manage maintenance of servers, upgrades, patches, scaling, etc.

Serverless computing also opens the door to microservices development . Developers can deploy self-containing, modular units of business logic as microservices independent of other microservices that collectively provide a macro application for the end users.

Serverless computing also provides developers and architects a previously unattainable agility to test out new concepts, to trial POCs, or just a new piece of code.?With the ease of a few clicks, developers can setup new functions, then write the function, and test it before incorporating it into a microservice or existing application. Going serverless also makes things easier for admins, freeing time and resources to focus on user management, security, and DevOps, rather than mundane server maintenance tasks.

Right-sizing serverless computing in your organization requires an equal focus on understanding the negative implications. One is the potential latency of infrequently used microservices and code debugging.?Most serverless services are designed to be put to rest if they remain unused, requiring some ramp-up time before the service responds to a request.?

Serverless also introduces unforeseen challenges when it comes to code debugging. the major vendors are working on solutions, it is not easy to integrate debugging on a developer’s traditional IDE. ?AWS is solving this with a local serverless module (or container) called SAM Local or using their cloud based IDE, Cloud9.

The Approach

Here are a few of the things you’ll want to keep in mind as you go serverless. As with any new technology, serverless computing introduces a learning curve.?We believe that people are the most critical component of the transition. There will be a need to train or hire resources with new skill sets, particularly in identifying competent administrators and architects for the platform itself. When it comes to development, existing professionals can be trained relatively easily since most serverless computing architecture supports the use of existing programming languages such as Java, c#, Node js, Python, etc.

A great way to adopt serverless computing into your computing environment is to identify a few use cases and conduct proof of concepts (POCs) around them.?Since a serverless computing architecture does not require and infrastructure configuration, most leading vendors provide a free tier account for a substantial amount of time, such that conducting a POC is relatively easy.??Most vendors provide a set of tools to build continuous integration and continuous delivery for development, testing, and deployments. Identifying tools from cloud vendors and/or other third-party vendors, and incorporating them during the POCs can help solidify DevOps best practices within the organization.

It’s time for IT organizations to free themselves from the servers and legacy asset conventions brought by the world of traditional computing. Since all providers provide a robust free tier to play with their serverless components, what are you waiting for? Start small and scale up today. There is no reason not to spin up a proof of concept, and evaluate where it works in your organization. ?