🔭 Service Discovery

Updated at 2021-03-01 16:29

Service discovery solutions provide the following common functionalities:

  1. Allow services to register to the directory.
  2. Allow consumers to find where those services reside.
  3. There can be various other optional features such as service health checks.

You would use service discovery when you have many APIs that are inter-communicating. Managing all, possibly changing, endpoints by hand in various services is very tedious.

The most common ways service discovery is done:

  • DNS: simple but a bit tedious to update e.g. Add a good TTL value and point to a load balancer
  • Zookeeper: old but battletested technology. Overly complex for what it is but decent if you have it running already
  • Consul: excellent service discovery software with health checks, extra points for supporting DNS
  • Eureka: niche but usable
  • Custom Solution: e.g. by using AWS resource tags

I'd say:

  • start with cloud-provider tagging if everything is in the cloud
    • easy to implement and manage, and well secure if you setup strict permissions
  • start with DNS if working on multi-cloud or on-premises
    • just so you have a single point of truth and one way of finding some service
  • switch to Consul when these become too much to manage
    • more to manage but comes with great benefits