What is service discovery?

  • A key problem in deployment is getting services to be able to find each-other

  • A database might be at 10.1.1.1:6543 while the webserver is at 10.1.1.2:8080

Level 1 Service Descovery: Baiscs

  • Everything is manually configured

You'll know you should start caring about service discovery when you start seeing one of the following:

  1. You want "zero downtime deployments" or to use other, more complex deploy strategies

  2. You have more than a couple of microservices

  3. You are deploying to several environments (dev, staging, prod) and it's getting unwieldy

Lever 2: Service Discovery via hash table

  • A straightforward approach: Store the service IPs in a hash table

  • The hash table in question in the config is here:

In simple term:

  • All you need to do is update your frontend to set the keys in the hash table to be the forntend's IP

  • Then make the backend do the same for the backend

Problems with this approach

  • It's complex

  • It's error prone

  • It requires to write custom config files

Level 3 Service Discovery via DNS

  • The idea for DNS is just to map hostnames to IPs. This is the industry standard

  • It would be ideal if we could connect to "http://frontend" from our reverse proxy and having DNS respond with the IP for the correct versions of the frontend

  • So instead of the first example where you use the IP

  • You would use mongo where mongo is a key in the key:value pairs

PreviousWhat is autoscaling & serverless?NextWhat is Log Aggregation?

Last updated