Single Point of Failure

[kisienya]

Hi,

Many thanks for this very resourceful product.

How do I avoid the problem of proxyql being a single point of failure, do we have high availability?

Cheers

[renecannao]

Hi,
thank you.
ProxySQL itself doesn't have a built-in HA solution, but it is very easy to architecture its deployment in order to avoid SPOF.
Few months back Percona published some articles where it is highlighted how to avoid single points of failure:
https://www.percona.com/blog/2016/09/16/consul-proxysql-mysql-ha/
https://www.percona.com/blog/2016/10/10/consul-architecture/

It needs to be noted that ProxySQL should be considered as an agent, and therefore its deployment shouldn't make it a single point of failure. What it means is that all your traffic shouldn't pass through a single proxy: you can have several of proxies (I know of deployments with hundreds of proxies) all running at the same time, and the application can connect to any of them.
You could deploy a ProxySQL instance for each application server (and in this way you already avoided a SPOF) or create a pool of proxies the application will connect to.

The only challenge in this case is ensure that the proxies have the same configuration.
Also in this case ProxySQL doesn't have any built-in solution for that, and this is intentional: I preferred to make it easy to reconfigure, so that anybody can manage it with the tools they are already using. In fact, it is so easy to reconfigure ProxySQL at runtime that you could use a variety of approaches, like using a configuration management tool (Puppet, Chef, Ansible, Salt, etc) or a service discovery tools (Consul, Etcd, zookeeper) to automatically reconfigure ProxySQL if needed.

[kisienya]

Thanks for this reply