The looming end of support for Windows Server 2003 is causing headaches for the many companies still using it – but Veeam regional manager for Southern Africa Warren Olivier says there are ways to mitigate the risks and lessen the pain of upgrading, without endangering the constant system availability that is required by the always-on business.
Many thousands of companies around the world are affected, says Olivier: “Late last year Microsoft estimated that there were still well over 20-million instances of Windows Server 2003 running, or about 39% of the entire Windows install base. All of those servers need to be migrated to a newer platform, but that takes time – and there are risks.”
He says companies running older software they have written or modified themselves face some of the greatest challenges.
“There have been good reasons not to move until now – some applications might need to rewritten entirely to work on a new platform, and it’s usually sensible not to fix what isn’t broken. But carrying on regardless is no longer an option. As of 14 July 2015 Microsoft will be stopping all support for Windows Server 2003, which means no security updates, no applications support and the loss of compliance status.”
Migrating old software to new platforms is always risky – but, says Olivier, companies that have virtualised their servers can lower their risks substantially by using copies of their backup files as sandboxes for testing.
“The only reliable way to test is in an environment that’s identical to your production environment, but safely isolated from it,” says Olivier. “A virtual machine is effectively a single file, so a backup tool that’s optimised for virtualised environments creates copies that can easily be reproduced. You can spin up a new copy of your entire environment in just a few minutes, so software patches and updates can be easily tested and retested without compromising or putting strain on the production environment.
“The one thing you can’t afford is downtime,” he says. “Today’s always-on business doesn’t allow for comfortable weekends of downtime while systems are migrated – the switch needs to be seamless and instantaneous. And the only way to achieve that is test exhaustively, both for stability and for user acceptance, before any changes go live. We’re fortunate that virtualisation has made this testing quicker, easier and less risky than ever before.”