When you have decided to move your application hosting to the cloud such as amazon ec2 instances, the common question is how can High Availability and Disaster Recovery can be achieved in a cloud environment.
This tutorial describes the best practice and a system architecture that you can use in your cloud deployment when you want to use HA and DR.
The proposed System Architecture is designed based on two Virtualized cloud environments, one for the Production site and the other for the Disaster Recovery site. The production site and the DR site can be in two different cloud environments. For example you can host the production site in amazon ec2 and the DR site in Windows Azure platform or any other IAAS provider in your local country.
As any application usually has a database, the Database server will be replicated in a "Master-Slave" architecture and the snapshots of the volumes will be taken at periodic intervals and saved to the Cloud Storage as backup.
A similar DB slave will be replicated in the DR site using the VPN Manager and backed up to the DR Site which is hosted externally in a different cloud infrastructure.
HIGH AVAILABILITY in more detail
In front of the Apache cluster we create a load balancer that splits up incoming requests between the two Apache nodes. Because we do not want the load balancer to become another "Single Point Of Failure", we provide high-availability for the load balancer, too. Therefore our load balancer will in fact consist out of two load balancer nodes that monitor each other using heartbeat, and if one load balancer fails, the other takes over silently, instantly.
The advantage of using a load balancer compared to using round robin DNS is that it takes care of the load on the web server nodes and tries to direct requests to the node with less load, and it also takes care of connections/sessions. If you application uses sessions, if a user is in a session on Apache node 1, he would lose that session if suddenly node 2 served the user’s requests. In addition to that, if one of the Apache nodes goes down, the load balancer realizes that and directs all incoming requests to the remaining node which would not be possible with round robin DNS.
For this setup, we need four nodes (two Apache nodes and two load balancer nodes) and five IP addresses: one for each node and one virtual IP address that will be shared by the load balancer nodes and used for incoming HTTP requests.
As for the database high availability, database replication together with heartbeat will be utilized. For example in a standard MySQL Replication the master server writes updates to its binary log files and maintains an index of those files in order to keep track of the log rotation. The binary log files serve as a record of updates to be sent to slave servers. When a slave connects to its master, it determines the last position it has read in the logs on its last successful update. The slave then receives any updates which have taken place since that time. The slave subsequently blocks and waits for the master to notify it of new updates.
When you implement such a solution, you can write automated scripts that will backup your VM images and data and application backup automatically to a cloud storage.