Server Clusters (Performance Patrol™)
What is a cluster?
Why use clustering?
Clustering is valuable for two reasons: First, it gives your Web site or application additional resilience by spanning it across two or more physical machines, such that if any one fails your service stays live. This is called auto-failover.
We normally experience roughly a 1% per year failure rate of physical servers, which results in up to an hour's downtime while the disks are swapped into a hot-standby chassis. Clustering allows you to avoid that small chance of downtime. Since clustering is now available for ®Miniserver VM's, that gives you a very low-cost way of mitigating the chance of hardware failure.
Second, some services need that one whole servers' resources, therefore the load needs to be spread across multiple machines. This is called load-balancing. This can be especially helpful if your service may have large peaks in activity, for instance a Web site that gets promoted in waves (eg. via mail shots or TV advertising) might need a lot of stand-by resource to cope with those. A load-balanced cluster would be ideal.
What we offer for clustering
Every server cluster we setup for our clients is tailored to their requirements, and we would always engage in a dialogue with them to determine their requirements. As mentioned, a cluster is just an arrangement of normal servers and as part of it, if you rent more than two servers with us we will configure a private VLAN to provide secure gigabit communication between your servers.
If you think you need a server cluster please contact us to discuss your requirements. We are happy to offer a free sales consultation to help you work out what it is that you need, and would be delighted to provide a detailed quote.
Performance Patrol™ load balancing & auto-failover
Using our extensive experience in server clustering we also offer a complete load-balancing and auto-failover solution called Performance Patrol™. This system allows the load to be shared across multiple front-end machines in a cluster and automatically manages the demand and re-routes traffic in the event of a failure.
About Performance Patrol™
As more and more mission-critical applications move on the Internet, providing highly available services has become a requirement for many of our customers. The main advantage of a dedicated server cluster is that it has hardware and software redundancy.
Because the cluster system consists of a number of independent nodes, and each node runs a copy of operating system and application software. High availability can be achieved by detecting node or daemon failures and reconfiguring the system appropriately, so that the workload can be taken over by the remaining nodes in the cluster. This system will actively monitor all servers in your cluster to ensure they are up and responding to requests, if any of them "drop out" then the solution will immediately detect this and failover any requests to the other server(s).
This solution can also scale and grow as requirements dictate and front end or database servers can be plugged into the cluster very easily.
Using a combination of LVS and Heartbeat software running on our twinned Performance Patrol™ dedicated hardware Memset® can now offer a load balancing and failover solution for all our customers at very competitive prices.
With the additional monitoring and work involved in the ongoing maintenance of Performance Patrol™, you will need to have our Fully Managed Support level for this option when buying a server. Performance Patrol™ itself costs $59.00/mo per server.
Example of a cluster
Take an e-commerce site, for example; there may be two Dell single-Quad core front-end web servers, set up to be load balanced and failed over, with two backend database servers set up in a master/slave configuration for both redundancy and performance. All these servers would be set up on their own private gigabit vLAN offering great security and performance.
This configuration allows the load to be spread across the machines, and for each server to focus on a specific task. Adding more servers into the cluster allows the capacity of the server group to be seamlessly scaled, and allows for a fail-over capacity in the event of one machine developing a fault.