Wednesday, 24 August 2011

Two-Node Clusters

Two-Node Clusters

Small businesses could be easily affected by any small changes in their business setting. Because of their small operation, they have to make sure that their cash flow is consistent to ensure they would be able to continue operating in the near future. A single downtime in their operation could mean substantial losses that they might be forced to close down. They need to have stability to ensure survival in the rugged business setting.

This is the challenge that online small businesses face. They have to make sure their website is up and running to ensure customers are satisfied as they would be serviced on time. They are still trying to build a reputation so a single downtime could be crushing a blow to their operations.

But keeping a website for online small business is often difficult since some opted to use 3rd party providers to help them keep their online applications running. This is risky since there is a single point of failure for their operations. Although a back-up plan with another 3rd party provider is feasible, getting the website back on track is too long and another provider is just too costly.

Two-Node Cluster as Alternative

As technology evolve, setting up clusters for small businesses has got a lot more affordable. Businesses today could consider implementing two-node clusters to secure their website or online application. This type of clustering would ensure continuous business operations without having to spend too much on hardware.

As the name suggests, the two-node clusters consists of only two-nodes in HA (High Availability) clustering configuration. In gist, the first node will be the primary node while the second node would be the back-up. In case the first node experiences trouble in maintaining functions or data requests, the back-up node will take over. This type of HA clustering takes on an Active/Passive format.

Aside from Active/Passive format for HA Clustering in two-node clusters, an Active/Active format could also be implemented. The functions or will be distributed on the nodes and whenever one of the nodes fails to operate, the other node will naturally take over to become the sole node for the workload.

Hardware Requirements

The hardware components to fully implement two-node clusters are minimal but each takes a very important role to ensure efficiency of the cluster. As the name suggest, it will only have two nodes with optimized processing speeds. Its only difference with larger nodes is its ability to function as a storage device which is added to assist the shared storage. The two nodes connect to the shared storage which functions as a server.

The components would also need to have interconnection hardware to ensure that each would communicate for stability. Administrators could choose between serial and Ethernet crossover cabling. Although Ethernet could still do its job to connect two-nodes, having a TCP/IP configuration is not really necessary since there are only two nodes connecting. For that reason, serial crossover cable makes it easier as it implements direct connectivity.

An optional hardware could be installed with interconnection protocol. A shared disk could be part of interconnection wherein it will serve as an active monitoring tool so that it could alert the administrator if one of the nodes is not working.

An often missed addition to the hardware setting to ensure stability of the nodes is the UPS (Uninterruptible Power Supply). The UPS will make sure that each component will be up and running even though there is no available power or while the company switches to an alternative source of power. But one UPS will never be able to handle the power requirements of the three major components in the cluster. Each major component should have a UPS so that the operation will continue.

Controlling Clusters

To properly control clusters, a physical or manual and logical control should be in place. The physical control will give the administrators the ability to manually stop the operations. The logical control on the other hand, will give administrators the ability to configure the nodes to have the right reaction in case one of the nodes will not work as expected. Although the manual control might interrupt the operations, the presence of manual control will help developers stop by force when there is a need to do so.

No comments:

Post a Comment