Assessing Workloads for Serverless Suitability
A common use case that became popular during the early days of serverless was image processing with Amazon S3 and Lambda functions. It showed how easy it was to upload a file and execute code in the cloud without manual intervention. Countless trivial serverless use cases and how-to articles soon flooded the tech media. While such simple examples give firsthand reports on the possibilities of serverless, enterprises have bigger challenges to overcome (both functional and nonfunctional). These include:
- The complexity of business logic tucked inside legacy monoliths
- Engineering teams spread across multiple departments
- The volume of users flocking to the website to consume the services
- Unpredictability of visitor traffic during certain times, days of the week, and special occasions
- Making sure the services are available around the clock and globally
- Depending on your business domain and priorities, you may also consider principles such as DevOps-first, security-first, DevX-first, observability-first, etc.
- Releasing new features and fixes without impacting users
- Securing customers’ personal information and confidential business data
- Protecting the business from malicious users, bots, and malware
This list is already long, but the suitability assessment should consider all aspects of the application, not just the implementation part. While a serverless enthusiast might assure you that it’s suited to every workload, there are some fundamental questions you should ask. For example:
- Will this application perform efficientlyon serverless?
- Will it be cost-effective?
- Can I trust the serverless expert to be objective?
Efficientand cost-effective are tricky to explain in a cloud and serverless context. In distributed computing, being efficient doesn’t always relate to speed. Sufficiently is an equivalent term that you can use to evaluate suitability in many cases. Similarly, cost-effectiveness does not always mean being cheaper than your on-premises or cloud-hosted container applications in like-for-like comparisons.
When you assess the suitability of serverless technology with serverless-first thinking, you consider serverless as your first tech‐ nology choice. This does not mean you overrule other technologies with a serverless-must mindset.
