58 lines
2.3 KiB
Markdown
58 lines
2.3 KiB
Markdown
|
---
|
||
|
|
layout: "intro"
|
||
|
|
page_title: "Use Cases"
|
||
|
|
sidebar_current: "use-cases"
|
||
|
|
---
|
||
|
|
|
||
|
|
# Use Cases
|
||
|
|
|
||
|
|
At this point you should know [what Serf is](/intro/index.html) and
|
||
|
|
the high-level features Serf provides. This page lists a handful of
|
||
|
|
concrete use cases of Serf. Note that this list is not exhaustive by
|
||
|
|
any means. Serf is a general purpose tool and has infinitely many more
|
||
|
|
use cases. But this list should give you a good idea of how Serf
|
||
|
|
might be useful to you.
|
||
|
|
|
||
|
|
It is important to remember that all the use cases available below
|
||
|
|
require _no centralized state_, are masterless, and are completely
|
||
|
|
fault tolerant.
|
||
|
|
|
||
|
|
#### Web Servers and Load Balancers
|
||
|
|
|
||
|
|
Using Serf, it is trivial to create a Serf cluster
|
||
|
|
consisting of web servers and load balancers. The load balancers can
|
||
|
|
listen for membership changes and when a web server comes online or goes
|
||
|
|
offline, they can update their node list.
|
||
|
|
|
||
|
|
#### Clustering Memcached or Redis
|
||
|
|
|
||
|
|
Servers such as Memcached or Redis can be easily clustered by creating
|
||
|
|
a Serf cluster for these nodes. When membership changes, you can update
|
||
|
|
something like [twemproxy](https://github.com/twitter/twemproxy) or your
|
||
|
|
own application's list of available servers.
|
||
|
|
|
||
|
|
#### Triggering Deploys
|
||
|
|
|
||
|
|
Serf can send custom events to a Serf cluster. If you cluster your web
|
||
|
|
applications into a single cluster, you can use Serf's event system to
|
||
|
|
trigger things such as deploys. Just call `serf event deploy` and have
|
||
|
|
event handlers installed on all the nodes and the entire cluster will
|
||
|
|
receive this message within seconds and begin deploying.
|
||
|
|
|
||
|
|
#### Updating DNS Records
|
||
|
|
|
||
|
|
Keeping your internal DNS servers updated can be a finicky process.
|
||
|
|
By using Serf, the DNS server can know within seconds when nodes join,
|
||
|
|
leave, or fail, and can update records appropriately. No more stale DNS
|
||
|
|
records, or waiting for a Chef or Puppet run to clear out the records
|
||
|
|
within X minutes. With Serf, the records can be updated nearly instantly.
|
||
|
|
|
||
|
|
#### A Building Block for Service Discovery
|
||
|
|
|
||
|
|
One of the most difficult parts of service discovery is simply knowing
|
||
|
|
what nodes are online, at what addresses, and for what purpose. An effective
|
||
|
|
service discovery layer can easily be built on top of Serf's seamless
|
||
|
|
membership system. Serf handles all the problems of keeping an up-to-date
|
||
|
|
node list along with some information about those nodes. The service
|
||
|
|
discovery layer then only needs to answer basic questions above that.
|