We noticed you're blocking ads.

Keep supporting great journalism by turning off your ad blocker.

Questions about why you are seeing this? Contact us

Font Size

- Aa +

Sun 30 Mar 2008 04:00 AM

Font Size

- Aa +

Out with the old…

The upgrade cycle is a continual concern for IT managers, but new technologies and trends may be changing the game for enterprises.

The upgrade cycle is a continual concern for IT managers, but new technologies and trends may be changing the game for enterprises.

They were state-of-the-art when you bought them - the fastest processor, the most memory, the latest operating system. But now these same machines are creaking along, several generations behind the most up-to-date technology and struggling to cope with the latest applications your enterprise is running. And the case is so early-noughties.

But deciding exactly when to replace IT equipment is not a precise science. Leaving aside the dictates of upgrading software, which almost always necessitates a simultaneous upgrade of servers, if not desktop systems, the decision to dump old machines and bring in fresher models can be somewhat complex.

On the server side, the calculation on one level can be as simple as "is this machine capable of running the software being used, and handling the current number of users?" If yes to both, then the server can stay - if no, it's time for a new system.

For desktops, other issues have always entered into the equation, not least office politics. Jealousy of colleagues' new PCs may spark demands by other workers for shiny new boxes, regardless of their actual requirements. And of course senior execs frequently require their systems to be ‘refreshed' more frequently than other, lower-ranking staff.

And when it comes to other areas of an enterprise's IT systems - such as networking, security and storage devices - the imperative often comes directly from need at a particular time, rather than a planned replacement or refresh cycle. Why discard a perfectly good switch, when it fulfils your needs as well as any potential replacement?

In fact, though, regional end users do not always consider networking hardware as a separate proposition to ‘core' items such as servers and desktops, according to Taj El-Khayat, head of the enterprise and channel group for the Middle East and Africa at Juniper Networks.

"Regional end users are definitely not making optimal choices when it comes to upgrades, to be honest. The mindset today is based on the experience they have - networking has traditionally had a much lower value than it has today, so there's a learning curve that decision-makers need to go through," he comments.

"Customers tend to have this mindset that they have to upgrade everything every two or three years - I think they're using the same trend they have for servers and client-side devices. The norm today in the Middle East is to refresh your infrastructure every two or three years, driven by expansion and the booming economy.

Our message is that there's a lot of different angles customers need to look out for when looking at high-performance networking - our recommendation is that it's not necessary to refresh your network infrastructure completely in that timeframe," El-Khayat adds.

Juniper's pedigree as a service provider specialist brings with it much longer refresh cycles for its products, and an expectation that its devices will have a longer working life than the average server. Other networking vendors also seem to expect less frequent refreshes of their equipment - good news for firms with stretched IT budgets.

In the meantime, a number of new factors have started to change the game when it comes to enterprise upgrade cycles on the server and client side. These - potentially - could change some of the fundamental concepts behind the idea of a refresh cycle, although quite what this will mean in practical terms for enterprises is hard to say.

First, virtualisation - that most pervasive of buzzwords for 2008. Spreading like wildfire across much of the western world, virtualisation has been slower to catch on to the same degree in the Middle East, which - as ever - may do regional enterprises some significant favours in the medium term.

The principle of virtualising compute resources is now well understood, but the implications for longer-term system buying are still emerging. In principle, with a fully-virtualised IT system, there is no requirement to replace servers - if additional processing power is needed, an enterprise can simply add another machine to the virtual ‘pool'.

This potentially means cheaper server purchases for enterprises - if it's merely being used as a top-up, there is not necessarily any need to buy the highest-spec server on the market. For fast-expanding firms that need to increase capacity but may not want to overhaul their entire IT systems - or invest in redundant capacity ahead of time - this may prove to be an attractive option.

By moving to a ‘rolling' upgrade policy, enterprises may be able to do away with the traditional idea of a three- or five-year refresh cycle, instead using a continually developing and evolving infrastructure which still retains the core functionality of the original design.

Of course, this model also brings with it new issues, such as the potential for server sprawl - especially in cases where larger numbers of low-powered machines are used. While this may be cost-effective on a month-by-month basis, enterprises may end up spending more in the longer run - not to mention the additional power, cooling and space requirements that large numbers of servers will have.
The answer to this may be to adopt a compromise position - plan server additions ahead of time, and retire older systems in favour of newer, more powerful servers. Again, virtualisation makes swapping out systems much simpler than in the past, allowing enterprises to manage a limited number of servers while increasing their collective compute power over time.

However, this leads neatly on to yet another issue - the dreaded spectre of vendor lock-in. Tier-one suppliers, especially HP and IBM, are currently pushing blade-based solutions as a simple and space-effective way to provision additional servers, but currently all the blade solutions
on the market use vendor-specific chassis and sometimes even management systems.

And although there is still the option to use rack-mount servers, uncertainties remain on the ability of servers from multiple vendors to work together in one virtualised environment. The accelerated development of virtualisation technology means this is unlikely to be a problem in the medium term, but is still a consideration - and will have an impact on how enterprises plan their future upgrades.

One of the potential ironies of virtualisation - a technology that eliminates hardware concerns from day-to-day operations - is that in the immediate future, many companies tempted by virtualisation may also be looking at a wholesale upgrade of their server systems.

While virtualisation systems are able to make use of older systems, the latest generation of CPUs and servers incorporate virtualisation-specific tweaks, making them more efficient at running virtual machines.

With more of the same set to arrive in forthcoming generations of server technology, there is a potential temptation to play ‘wait and see' - although as virtualisation is apparently here to stay, this process is likely to be never-ending.

Potentially more interesting is the impact virtualisation will have on client devices, especially in larger enterprises and those with significant customer-facing operations, such as banks or call centre operators.

By giving enterprises the possibility of creating hundreds or thousands of virtual machines, virtualisation has put thin clients on the corporate IT map again - and this time they might even catch on.

From an upgrade point of view, thin clients are very good news for firms - after an enterprise buys the initial stock of thin client terminals (usually slightly cheaper than regular desktops), it will be able to operate them for a much longer period of time than equivalent desktops. Then, if a user requires additional resources, this is simply a matter of back-end provisioning.

While this is an apparently ideal solution for enterprise IT departments (if it works), users often feel less happy - familiar with the usual round of regular desktop upgrades, they reject the small thin client terminals as "not proper computers".

There is no doubt that, for appropriate users, thin clients solve a large number of issues associated with upgrading - in addition to their oft-cited benefits on security and downtime.

But even allowing for a large proportion of users being able to move onto thin clients, the majority of enterprises will still have significant numbers of employees - especially consultants, mobile sales staff, specialists and technical support staff - that require either a dedicated desktop or, more commonly in recent years, a laptop.

For these more conventional modes of PC use, the upgrade cycle is key to keeping long-term costs down, specifically in terms of maintenance.

A 2006 study by Wipro ("Optimizing PC Refresh Cycles to Maximize Business Value"), which was sponsored by Intel, suggests that the cost of keeping both desktops and laptops working and secure increases dramatically as the machines age.

"Firms on a three-year lifecycle invest an additional US$67 per desktop PC each year on hardware, deployment and decommissioning costs compared to firms on a four-year cycle, but they manage to save $127 per PC each year in support costs. The result is a net savings of $60 per desktop PC per year - a $1.05 million annual saving for a fleet of 17,519 desktop PCs," says the study.

Wipro's researchers have derived these figures from data including the average cost of machines, as well as support and software costs. While Wipro maintains its 2004-study recommendation of replacing desktop systems every three years, the latest research reveals that, for desktop systems, refresh cycles may actually need to accelerate slightly.

"Help desk call volume spikes dramatically at three and a half years into a PC's lifecycle and continues to climb. This increase is much higher than previously seen, further emphasising the need to follow refresh cycle best practices," comments the study.

Arabian Business: why we're going behind a paywall

For all the latest tech news from the UAE and Gulf countries, follow us on Twitter and Linkedin, like us on Facebook and subscribe to our YouTube page, which is updated daily.