Container or Serverless – Which approach is better?
Published : March 23rd, 2021
When it comes to application deployment in the cloud, container and serverless architectures are two of the most commonly used strategies. When used the right way, they both help organizations deploy applications faster and more cost-effectively.
Though the features of containers and serverless architectures overlap to some degree, they are not interchangeable. Containers work better for some use cases, while in others, serverless is what you need.
In this blog, we discuss the similarities, differences, and key considerations between containers and serverless computing services and explain why an ‘ ‘organization’s team might choose one technology over the other.
Containerization refers to the set of technologies and tools that enable the automation of typical deployment procedures performed manually in the legacy world. Without containerization, the manual deployment process is usually error-prone and can take several days or even weeks to accomplish. But with containerization, all the required tools and frameworks can be systematically baked into scripts that can be repeated with one command or a click of a button.
In the past, a typical deployment of an enterprise application used Application Servers (AS) such as WebSphere or WebLogic. Getting these installed for the first time with all the bells and whistles on the identified infrastructure was in itself a project that could take several months. Once the AS was installed and before the application could be deployed, the configuration of data sources and other external dependencies had to be fed into the configuration files.
Once this necessary setup was complete, the executable would be placed in the configured destination, and the AS started. Unless the person doing this was an expert, this would seldom work in the first few attempts.
With DevOps, Docker, and containerization, this entire infrastructure, container images with dependent tools, and software with respective connectivity can be coded as executable commands that will ensure that the environment is ready at the click of a few buttons. The container instances are usually immutable, and every new release will bring up new container instances. The database and other persistence storage can have strategies for clean starts or can retain their schema depending on the business’s needs.
Serverless architecture is an innovative Software as a Service (SaaS) option supported by most cloud providers that enable organizations to focus on building the core functions with the programming language of their choice without having to worry about the container and infrastructure.
Serverless architecture gives a head start to the development team without having to sort out the infrastructure and provisioning solutions. AWS Lambda is one of the serverless (Function as a Service) options offered by AWS to develop a microservices architecture for organizations using AWS. Similarly, Azure has a serverless counterpart called Azure Function for organizations using Azure.
Typically, in the early stages, serverless architecture was used as a quick and inexpensive option to perform stateless operations to process data and pipelines involving data processing. Serverless is gaining momentum in developing stateless APIs because of its seamless integration with the cloud ecosystem. Compared to the container-based approach, serverless is yet to mature fully but is well-suited for rapid development.
Both containerization and serverless technologies have their strengths, weakness, and specific applicability. In this section, we will discuss some of them based on our experience with our customers.
One of serverless’s biggest disadvantages is vendor lock-in, especially when functions integrate with vendor proprietary services.
Following are some situations when you could choose serverless architecture over containerization:
- For organizations already using the cloud (AWS or Azure) and not using microservices, serverless is a great way to implement microservice architecture for newer business capabilities.
- Organizations with no cloud presence can leverage serverless architecture to experiment and develop microservices architecture from scratch for identified business capabilities.
- Organizations with multiple isolated solutions built over a period of time can explore opportunities to consolidate and modernize as long as:
- the overall system is not complex.
- the number of functions or services is limited and is not expected to grow exponentially.
- Building data lakes with serverless used for data processing and transformation.
- In large complex enterprises with containerization already in place, serverless can be used for observable business capabilities that are less frequent in nature.
- The user traffic for certain business capabilities is highly volatile and expected exponential traffic for a shorter duration.
- In cases when you choose a cloud provider for a specific use case that is not feasible in a primary cloud platform like the heterogeneous cloud.
Serverless architecture can be a great additional tool for enterprises and startups offering unprecedented time-to-market and low pay-per-execution costs.