Very much in the shadows of all the press coverage and hysteria attendant on emerging cloud architectures and customer-facing systems of engagement are the nitty-gritty operational details that lurk like monsters in the swamp of legacy infrastructure, and some of them have teeth. And sometimes these teeth can really take a bite out of the posterior of an unprepared organisation.
One of those toothy animals that I&O groups are increasingly encountering in their landscapes is the problem of what to do with Windows Server 2003 (WS2003). It turns out there are still approximately 9 million WS2003 systems running today, with another 2+ million instances running as VM guests. Overall, around 11 million OS images and a ton of hardware that will need replacing and upgrading. And increasing numbers of organisations have finally begun to take seriously the fact that Microsoft is really going to end support and updates as of July 2015.
Based on the conversations I have been having with our clients, the typical I&O group that is now scrambling to come up with a plan has not been willfully negligent, nor are they stupid. Usually WS2003 servers are legacy servers, quietly running some mature piece of code, often in satellite locations or in the shops of acquired companies. The workloads are a mix of ISV and bespoke code, but it is often a LOB-specific application, with the run-of-the-mill collaboration, infrastructure servers and, etc. having long since migrated to newer platforms. A surprising number of clients have told me that they have identified the servers, but not always the applications or the business owners – often a complex task for an old resource in a large company.
So what are the options and best practices for organisations facing the W2003K “monster”? Conversations with clients and vendors providing migration assistance point to several good practices and options:
- Staying with WS2003 is probably not an option - WS2003 servers will represent a significant liability and regulatory exposure after the end of Microsoft support, even if they still run correctly. After the end of official support, any custom support contracts for continued support will be prohibitively expensive. One service provider mentioned a figure of “around $200K annually” for “a handful of servers”.
- Seek executive sponsorship –. Organisations that do not have sufficient funding or where their efforts are being blocked or ignored by other stakeholders need to escalate the situation. You may get beaten about the head and shoulders for not surfacing it sooner, but this is preferable to the consequences of not raising it until something happens.
- Aggressively inventory your assets – Armed with your executive sponsor’s clout, you need to inventory all the servers and applications. Once you have the inventory you need to understand which ones can easily move, which ones need some level of “modernisation”, and which ones need to be scrapped and replaced with a more contemporary version or alternative.
Be prepared to make some tough decisions and possibly upset some people – In many cases, long-cherished software may need to be replaced forcing users to migrate to another alternative, and internal process will need to change, always uncomfortable for some.
One nearly universal truth about WS2003 migration is that if you have more than a handful of servers and you are still thinking about it now, you will probably need external assistance in the form of professional services to help understand your options and to actually implement the program. Right now, there is no shortage of vendors willing to assist you in this migration, with, in addition to Microsoft itself, includes all of the major systems vendors. For the system vendors, the WS2003 migration is a major potential windfall for server revenues, and some third-party forecasts indicate that this migration could add several percentage points to 2015 global server revenues.
So, where to start? A diligence of potential service partners is way beyond the scope of this document, but if you are committed to (or at least not displeased) with your primary server vendor that is an excellent place to start shopping. All of the major x86 server vendors have structured WS2003 migration programs, and coordination between the analysis and implementation should be easier if the work is done by the vendor who will be also selling you the additional inventory you will need to complete the migration.
You can obviously also engage an independent consultancy to preserve a better negotiating position, but in reality you can take the list of proposed systems from any vendor and competitively bid it if you wish. But in most cases I think that most users will tend to stay with their current incumbent vendors after a little posturing.
In any event, the most important thing is to get started if you haven’t, and to make sure you are meeting your milestones if you have.
Posted by Richard Fichera