SYDNEY – The city of West Torrens, which is home to Adelaide Airport, was prepared enough for emergencies that had a disaster recovery plan. However, it would have taken up to three weeks after a major incident before the administration would have been fully operational.
But recently it implemented a server virtualization and redundancy strategy to protect its data from unexpected disasters such as air crashes and floods.
Now, recovery is a drag-and-drop exercise in which entire virtual machines can be backed up, and system restoration can happen in minutes.
The city, the second oldest council in South Australia, covers an area of around 36 sq km and has a population of 50,500.
Its manager of information services, Chris James, said the need to upgrade the disaster plan was driven home when fire devastated a neighboring local government office. At the time, James said West Torrens was relying on uninterruptible power supply technology that allowed systems administrators to conduct a controlled shutdown sequence if power failed.
He said data was backed up nightly to disks and tapes, but resuming normal operations could have taken up to three weeks.
“We began formulating a technological and policy-based disaster recovery plan to ensure we could remain operational in the face of unforeseen events,” James said. “This was also an opportunity to update our information systems strategy, utilizing current technology to optimize the IT server architecture.” Recognizing that a virtualized environment would help meet disaster recovery objectives, the council engaged Technical Architecture Solutions (TAS) to implement the server plan.
TAS formulated a two-stage plan that included virtualizing a number of servers then replicating these to a second physical site to provide full disaster recovery capabilities. In all, 24 of the 26 physical servers were virtualized and consolidated on to four physical servers. Within a month, the council and TAS converted 15 physical servers into virtual machines, running on top of the VMware ESX Server hypervisor on four IBM X3650 servers, each with two dual-core 3GHz CPUs and 16G bytes of RAM. PlateSpin PowerRecon was used during capacity planning, while VMware Converter and PlateSpin PowerConvert facilitated the virtualization of council’s application environment by converting existing servers into virtual server images.
James said the quick wins proved the project was viable and led to the remaining servers being virtualized.
He said virtualizing the data center environment freed up 24 physical servers, some of which have been reused to construct the disaster recovery capability. “To ensure continuous availability, virtual machines can be mirrored daily between the primary and backup data centers over a dedicated fiber-optic connection,” James explained.
“This configuration significantly improves business continuity capabilities: in the event of a primary server outage, the council’s application environment can be switched easily to the redundant site or to any other third-party facility capable of receiving the virtual machine files.
“Rather than build disaster recovery over our old server environment, it made sense to virtualize and then develop disaster recovery over our new virtual center. “It’s a lot smaller footprint; deploying new servers is now a drag-and-drop exercise, entire virtual machines can be backed up and restored in minutes.” Even in normal operation, James said the virtualized. environment has paid significant dividends.
He said payroll, Web services and Intranet systems are running faster and changing out equipment is significantly quicker. “New servers can be brought online in minutes when needed rather than days and having fewer leases will slash server leasing costs by 80 per cent over the next two years,” James said.
“We’ve designed a lot of robustness, reliability and fallback into the solution. Our targeted tasks have been achieved successfully, with network administration staff working incredibly well and productively.” In the long-term, the council’s ability to move its virtual servers easily from system to system may lead to an even more comprehensive disaster recovery environment.
“The City of West Torrens is considering setting up a physically distant secondary site where similarly configured servers would provide additional protection from disaster,” James said.
“A shared services model, in which the City of West Torrens and neighboring councils would jointly develop a central disaster recovery environment physically distant from their offices, would further reduce costs and help spread risk further.”