Serverless Applications – Is it the way forward?

Lakshmi H Shastry • November 26, 2019
Share This Article


Businesses are adopting innovative cloud technologies, serverless computing and microservices architectures be it for IT modernization, mobile apps, data-driven decisions or IoT landscape.  Serverless applications are growing in popularity from the past few years.  Market research forecasts Serverless applications growth is driven by harnessing cloud technologies for rapid application development, dynamic scalability, optimal use of resources and cost-efficiency.

Serverless applications are more flexible and amenable to change, as individual functions have an architecturally aware role, a key benefit for the dynamism of today’s businesses. Event-driven applications and environments emphasize on choreography than orchestration.  Serverless approach helps to focus on robust business logic, immersive user experience and responsive, reliable API ecosystem.

Serverless applications, accelerates the pace of innovation and productivity with the ability to release independent functions in faster iterations to address evolving business needs without having to manage servers, storage, networking, or provisioning of resources. A serverless framework is comprised of distributed microservices that self assembles deployments with installations, security updates and function calls in response to an external event.  Ease of deployment & agility drastically reduces dependency on IT Ops, DBAs etc.  With the serverless application, costs depend on usage of RAM, CPU, network or number of requests.  When there is no consistent usage, as charges are applied only when resources are used, makes them more cost-effective.  Scaling architecture with high availability managed by its provider is a great strength.

Serverless techniques and design patterns are contextual to the purpose being fulfilled, but following best practices, some as listed, helps in building strong serverless applications or to integrate them with existing architectures:

  • Switch statements in functions tend to make them large & complex. This pattern ends up in scaling the entire application, rather than scaling the specific element. Each function should do only one thing
  • Embedded functions cannot be easily broken down. Functions should push data to a data store or queue and if more work needed, function triggers must be event-based
  • Cold starts impact performance on starting up and scaling. Use libraries and dependencies absolutely needed features to fulfill function core responsibility. Optimize build and keep deployment packages small
  • Database Connections take time, and while function is scaling up, each function environment needs a connection, adding bottleneck and I/O wait into initialization. Handle connections with autoscaling servers
  • Decouple function co-ordination and failures with queues and state machine for chaining and choreography of asynchronous invocations for passing data, managing global state, retries and failed events
  • Rethink data layer significantly for scale as low latency is a pretty big bottleneck and causes significant overhead. Treat all data is in motion, and not at rest at any point

Serverless computing is potentially a huge development, but there are caveats impacting rapid adoption and growth scale of serverless market.

  • Going serverless requires an enormous rethink of traditional development and production paradigm. Without local data storage, application must be stateless, although good, needs reengineering.  Typically uses a large ecosystem relying on other hosted services such as databases, authentication & authorization, encryption to name few.  It’s a new way to build, deploy and manage code with a new mindset
  • Flexibility is a trade-off for more complicated moving parts. Managing a complex application with many endpoints can be challenging in microservices and Event-Driven architectures.  Absence of system control has several consequences such as cold starts, loss of functionality, forced upgrades, unexpected limits, and cost changes
  • Multitenancy problems as serverless architecture can have concerns with security, robustness, and performance. Dependency on vendors for debugging and monitoring tools is a pitfall, also debugging distributed systems is difficult and requires a significant amount of relevant metrics to identify the root cause.  Monitoring and testing are critical as badly written functions increases costs a lot
  • Without defined standards, there are lock-in problems, compliance, security concerns, and vendor dependency. Switching serverless offering to another vendor requires significant time and efforts.  Serverless applications must be developed as secure application and follow best practices of platform offerings.
  • New technologies always have risks associated with lack of tools and the stability of frameworks
  • Serverless is not the right fit for always-on applications requiring very predictable and consistent performance, and reading and writing a lot of data into back-end systems

As Serverless technologies are moving towards maturity, certain direction towards establishment of standards for portability, kick-starter tools with auto-configuration services, premium plans with custom resources limits for deployment speed of updated functions,  fixes to cold starts without need of pre-warming and maturity of tools with support for local, offline debugging, testing and monitoring supports needs of today’s large scale enterprises.

With so many thrilling trends in the vast, ever changing cloud computing world, best days for serverless applications are still ahead in well-designed and well-considered solutions…

To know more, please download the whitepaper on Serverless Architecture with AWS Lambda : Beyond Functions.

Let’s create something amazing together!

Contact us Next
Latest Blog
LinkedIn Instagram Facebook Twitter