Closed clintkitson closed 8 years ago
Start with a discussion on how we've used stateless and stateful apps for years.
Show examples of stateless apps, both new and legacy (Wordpress, Drupal, SugarCRM) Show examples of stateful apps (Postgres, LogStash, Memcached/Redis)
Continue with the new way of running applications, without known IPs or ports. Explain Service Discovery (Consul, etcd, Zookeeper, focus on Consul)
Start a Postgres DB for a certain legacy service. See it auto-register in Consul. Show IP and port. Start a legacy service. Auto-register and get info from Consul. Show IP and port. Show loadbalancer, have it auto-load frontend app ^. Go to LB hostname for app. Restart app and backend service, see it all re-attach and get reloaded.
Show a scale-out app with a distributed backend (CouchDB/Cassandra). Verify auto-registration, show working app.
Pull the plug on a few servers, show the applications and backing services all redistribute.
How do you deal with infrastructure resiliency when your modern apps are mostly stateless? Do you care at all if your infrastructure goes down? Of course you do! In this session we will demonstrate several concepts: stateless vs stateful applications and how to build proper Cloud Native Infrastructure for them; data persistence in distributed environments; service discovery; and orchestration layers.