Septaz User Guide

Introduction to Septaz

Septaz is a completely automated system that makes no assumptions about your infrastructure on AWS.Septaz works seamlessly with your existing auto scaling groups, database clusters or anything else that you could be running on AWS.

How to Setup Septaz?

Integrating Septaz with your AWS account is a very simple process. Septaz can be integrated in either one of Automatic or Manual mode.

How to Setup Septaz on Database Instances ?

You can use Septaz reliably on instances that write data to disc like database instances, search instances or data processing applications.

How to Setup Septaz on Apache Hadoop, Apache Spark ?

You can use Septaz reliably on Apache storm, Apache Hadoop.

Disabling Septaz

There could be circumstances where you would want to disable Septaz such as a planned infrastructure upgrade or a migration to different region. In such scenarios follow instructions below to disable Septaz and roll back all changes.

How to Setup Septaz on Database instances ?

You can use Septaz reliably on instances that write data to disc like database instances, search instances or data processing applications.

Before following the instructions here, ensure your instances do not store data on root volume. If your instances are storing data on root volume then you will have to modify your instances such that they store data/logs in separate EBS volume. Your root volume should contain only application software.

Septaz monitors all instances that have been registered using Monitoring Rules in automated mode or instances that have been registered by scripts in manual mode. Whenever Septaz sees an opportunity to cut costs or there is going to be interruption in spot market, Septaz will replace existing instance with a new on-demand or spot instance. This replacement process behaves like a reboot.

For single instance database instances running DB instances like MySQL, this will result in a downtime. Users can convert their large single instance database instances to multiple small instances to form a multi-master with read replica cluster to achieve no downtime and better performance.

For databases like Cassandra with replication factor of more than 2, users can spread their instances on multiple AZs to ensure near 100% availability.

In most cases, downtime at instance level is less than 30 seconds.

It is most important to setup start and stop hooks correctly when using databases with Septaz.

Septaz first launches a replacement instance and then calls stop.sh on running instances before detaching data volumes and elastic ip. Once stop.sh executes successfully, Septaz detaches all required resources from existing instance and attaches them to new instance. Once this process completes, script.sh is executed to complete the replacement process.

Start.sh should be used mount all EBS volumes, start all services related to instance including starting database instance and putting the instance in production mode.

Stop.sh should be used to cleanly stop all services including database instance. After stopping of all services, no process should be accessing EBS volumes that store data. Once all services have been stopped, all EBS volumes should be unmounted. so that Septaz can attach volumes to replacement instance.