| Case study: How high availability can support business continuity
Since the conception of the first commercially available Unix clustering product which was developed by two British technical specialists in the very early 1990’s, there has been a misconception that high availability has been beyond the reach of most moderately financed IT Departments. Cost and complexity have always been elements of any planning and technology overviews with a huge army of vendor consultants always ready to paint a picture of ‘worst case scenarios’ But there is an easier way of ensuring not only high availability but business continuity too. In 2001 a popular and growing North East England university purchased a comprehensive application to meet the demand of an increasing number of students. But in what way could it solve two initial problems? The loss of a critical service by means of technology failure and the complete loss of the data centre from a fire. The early discussions involved multiple teams from application development, project owners, IT staff, the estates office, and hardware vendors. You can picture the scene - complete chaos. With no clear solution on the table, the development team gave the task to the IT dept with a remit of: ‘The application must be live to the end user 24x7’. Trials of the new system were undertaken and studies showed that the system was at its peak on Sunday afternoons. The task was to ensure that a new seven figure student record system with all the components of access to mail, filestore, and an e-commerce suite which allowed credit card bill paying and purchase, ran without a glitch. As always in any high profile public IT project, men in dark suites and sharp pencils were observing from afar. One solution was to employ the services of the chosen hardware vendor, and have a six-man team ‘camp out’ for four weeks and deploy the ‘standard’ cluster solution. The plan turned out to be very effective for high availability but with a huge price tag and a six week deployment. The University didn’t have the budget or the time scale to implement before the start of term. So this plan was written off. Un-wittingly the University with the aid of their specialist high availability provider put a plan in place that would answer not two, but four problems. The plan was to place a server and storage technology 500 metres apart in two separate buildings with the usual multiple heartbeats communicating between the sites. In the event of a hardware problem, such as a CPU failure, the application and all users would be diverted to the secondary site. High availability had been achieved. Fire or flood was an issue the department was keen to avoid and with the secondary site having the exact same architecture in place, the IT team would be able to commence work in the event of site A destruction with no loss of data, service or somewhere for them to physically work. Business continuity had been achieved. The other two important factors would be the ability to use the secondary site for test and development. New releases to the application underwent formal test before going live and rather than have a piece of hardware redundant for ‘just in case,’ the department used the technology to its full potential. The final factor was significant and today remains a powerful demonstration as to why their chosen high availability solution was indeed the right choice not only for disaster recovery, but for business continuity reasons too. The university would be able to manually bring down and failover services to undertake vital patch up-grades, hardware revisions / service and install new ‘bolt ons’ to the application in normal working hours. This ensured the ability to provide scheduled maintenance periods without expense, loss or reduced service by the university’s own system admin team. This also kept the bean counters happy and to some extent staff, as long periods of out of hours and weekend overtime was no longer necessary. Increased functionality without any comprise in technology saved this university a great deal of time, effort and expense. In fact the solution was cheaper by a factor of x4 to its nearest rival. So cost and complexity can be significant elements in any high availability and business continuity provision. The best advice is to shop around, and just because the solution carries a big brand name doesn’t always mean it’s the best fit. Accredited specialists most often produce the most interesting deployments. Check that your high availability provider does not have ‘mandatory’ hidden provision’s which can only be carried out by their own service team if you wish to make any changes to your configuration. This particular technology that the University deployed today is largely un-changed from its initial conception. Allowing the more proficient IT technical expert to deploy the entry level solutions such as web serving and e-mail services has contributed to significant cost savings. With the charge of Linux now common place in almost all area of IT it’s satisfying to know that an increasing number of very low cost high availability solutions have emerged. The challenge as always will be to keep IT high availability and business continuity simple. Andrew Ostcliffe is US Country
Manager High Availability.com Ltd
•Date:
21st January 2005 •Region: World •Type:
Article •Topic: High
availability |
