Balanced services from ANU

If you run a high volume internet service such as a popular web site or email service, you may find that a single server cannot meet all of your requirements. Typical problems you might encounter include:

• Poor performance during peak times
• Downtime caused by high traffic, software problems, hardware failures etc
• Configuration changes or software updates require service downtime
• Adding capacity requires downtime (installing additional memory or disks)
• High performance servers are expensive

Performance

A cluster increases performance by distributing the load of your services between multiple cluster nodes. The load on each node equals the total service load divided by the number of nodes in the cluster. If you need more capacity, simply upgrade your cluster nodes or add additional nodes.

Uptime

The load balancer which manages your cluster will automatically detect node failures and remove failed node from active service, redistributing traffic to other healthy nodes in the cluster. The transition is seamless to the end user. Once an administrator has fixed the faulty node, it is automatically added back into service and will begin handling requests again.

Change management

Maintenance such as software updates or hardware upgrades can be performed on each cluster node individually. A node can be taken offline while other nodes continue to serve your traffic. The changes can then be tested prior to adding the node back into active service, thus ensuring a smooth upgrade with zero downtime.

Capacity planning

We will provide you with detailed usage graphs for both your overall cluster and for each individual cluster node. By monitoring trends in performance you will be able to determine in advance when your next upgrade should occur, giving you time to plan accordingly.
The addition or removal of cluster nodes is seamless to the end user.