While in my previous blogs I provided insights to the benefits and challenges of Serverless Architecture. Some use cases fall into a default category of being Serverless, however some are driven primarily based on the characteristics of Serverless and hence it is essential to understand that first and apply accordingly.
Following principles of Serverless must be taken into consideration while deciding on the suitability of use cases.
Less computing power
Ideal for those use cases that require less computational power but does its job without impacting overall user experience. For e.g., On upload of a video, an image processing job kicks in to generate thumbnail images out of it and then the images are stored in a storage system such as S3. This kind of use case doesn’t really make an impact on the overall user experience.
Event driven in nature
We live in an era of modern applications that rely heavily on multiple cloud providers including 3rd party SaaS applications/services such as Salesforce CRM, Google Maps, S3, Twitter Analytics, GitHub for various purposes. Most of these providers will not let you run your applications/services on their systems but trigger events that you capture and process as per your application needs. On the other hand, your applications can send events to event processors such as API Gateway/S3 deployed on Cloud.
Similarly, database changes triggering events to internal/external systems, addressing eventual consistency in distributed applications.
Not ideal for long running processes
Serverless computing is not ideal for applications that has long running processes or workloads. For e.g., it is not ideal for web applications or web services.
Scheduled batch jobs/workloads
If you have built batch jobs that runs on regular basis which requires dedicated infrastructure round the clock. Since the batch job runs for a scheduled period of time leading to under-utilization of resources thus you can leverage Serverless computing model to save resources and cloud infrastructure to utilize intense parallel computation, IO and network access. Similarly, it can be leveraged for ETL processes that are batch in nature.
Automation needs
You can leverage less computing power of Serverless on cloud to automate some repetitive tasks such as backups, resources monitoring, generating reports, etc. Few more use cases that you can think of is automating the CI/CD pipeline by eliminating the need for Provisioning and Operation overheads.
Auto-scaling needs
Most of the applications have such use-cases that gets expected surge in requests are good candidates to be implemented using Serverless approach. Serverless computing model as stated earlier leverages cloud infrastructure providing inherent auto-scaling capability.
Pay-as-you-go subscription model
As the Serverless model uses the computing power only for a required time and then destroyed right after the usage thus utilizing the resources efficiently and reducing the costs drastically. Thus you are paying only to the resources that are essential for you.
API Development
With auto-scaling option in Serverless, scaling APIs which is one of the common task, is trivial. And when API is not in use, then you pay nothing.
Single-Page Application
Such type of applications usually relies on APIs to serve dynamic content from disparate systems. APIs through Serverless functions can serve dynamic content wherein static content is served by means of cloud provided storage services.
Mobile BaaS
Through APIs you can build backend services using Serverless approach that serves data to mobile applications from disparate systems.
Data Processing and Analytics
ETL Processes
Image Processing
Data Enrichment
Real-time Data Streaming and Processing
Hosting Machine Learning models
Webhooks
Most webhook-type actions don’t need much of code to get to work, thus they are ideal for function-oriented approach provided by Serverless computing.
Automation Needs
Housekeeping Tasks
Auditing and Notification
Respond to Alerts
Periodic Jobs
IoT Backends
Streaming of data from various devices through Pub/Sub model, that triggers Serverless functions for processing and enrichment of data.