Server sales and the challenges to UNIX as a platform

There’s a great article on theregister.co.uk today talking about server sales in Q4 and also the market place or demand for UNIX servers dropping, this is to be expected for a number of reasons. It made me think of a conversation that I had been having with one of UNIX colleagues talking about Itanium, SPARC and POWER platforms, he had noticed an increasing demand internally where he worked for transition from high end platforms towards more x86 Linux based solutions. “But they don’t realize the savings that they could make with one high end system running multiple platforms, the consolidation ratio, the high end resilience and scalability…” he went on and he’s wrong, but increasingly I find myself even rebooting or correcting myself; if anything it’s one of those statements where dad would remind me that everyone wants ‘club class’ high end infrastructure until the bill is presented and amnesia sets in, what’s the next one down from that, oh and the one down from that….

In the olden days we ran a system until there was investment available, until the business had sweated the assets and typically until the point that it was beyond economic repair. This applied to the desktop as much as it did the x86 server running Windows NT something or Windows 2000. To this day therefore it can be so easy to say it’s a DL380 G4, it’s only a web server it’s fine, when we should be virtualizing or upgrading it, to reduce support costs, improve energy efficiency and at the same time force application migration and coding for the newer operating systems to avoid being stuck in the here and now for the next few years.

Times have changed and its for a number of reasons:

• The x86 server has become ‘commodity’, it’s more economic to replace the server every 18 months or less to take advantage of the latest chip-sets, innovations in management, scale and energy efficiency, with the return on investment being in some cases months.
• The x86 server offering both in terms of the hardware and the operating systems that they run have improved significantly, redundant power supplies and memory, inbuilt RAID and much improved remote management and automation reduces deployment time, extends resilience opportunities and allows scaling of platforms on enterprise levels.
• Specifications have changed, as a server guy, you don’t tend to get disk space alerts for the OS drives or memory utilization errors as the memory footprints, the standard drive sizes are so much larger now and by the time you do, it tends to be reaching the point where you might find it’s cheaper to replace than it is to upgrade.
• The economics drive regular replacement of the underlying hardware, therefore in terms of capacity planning, reliability and resilience, I need a platform that lasts for a few years, not decades. The design therefore the questions about the specifications and requirements as part of the on-boarding process become more simple, will a x86 offering suit? Will it meet the requirements for the next 18 months? In some respects if the answer is yes, let’s proceed, even though the high end solution might last four years or more, because it’s x86, because we deploy on demand and automate the x86 space, building out a new environment on new hardware shouldn’t be an issue and probably will be demanded in terms of costs and supportability. Crucially as we look at tools like IBM PureSystems or HP Cloud System where I press WebSphere platform deploy and everything just happens without the trauma, the app load, the engagement and the complexity.

The UNIX platform is here to stay, but what we are finding is a marked change towards the rules of engagement, certainly personally as a sales guy, a project manager or a consultant, I would be stating if it can go on x86, then it goes on x86, the business, the vendor or the application team have to justify why we can’t use industry standard servers, opting for commodity standard offerings to drive down capex, opex and deployment costs/lead times. We can customize on the most effective high end resilient platform of your choice, it comes down to ‘the cost of doing business’ does the resilience or the application mandating that platform generate enough revenue, branding or empowerment to those paying for it to justify that extra level of investment away from a standard offering on x86 to an industrial solution based on UNIX derived solutions on high end platforms.

This site uses Akismet to reduce spam. Learn how your comment data is processed.