Given that the concept of infrastructure, particularly in the cloud, now spreads a great deal further than just a few racks of servers, disk drives and network switches in a datacentre, a recent announcement about desktop upgrades by market analysts, Gartner, raises as many questions as it answers.
The company has issued advice to businesses about setting and managing timetables for the migration of desktop PCs that currently run Microsoft’s Windows XP to Windows 7. It notes that some 80% of users bypassed the opportunity to migrate their desktop PCs to Microsoft Vista, but suggests that they should contemplate a move to Win 7, not least because Microsoft plans to curtail support for XP in 2014. It suggests, therefore, that businesses start now with serious planning to undertake the migration, either as a one-time, rip-and-replace exercise, or as part of the rolling process of attrition, with a latest start date of next year.
At face value this is sensible advice, not least because unlike Vista, Windows 7 has received broadly favourable reviews and is showing all the right signs of being robust and reliable in operation. It is, however, advice that seems to be based on the assumption that users have little practical alternative to upgrading their desktop environments when in practice other alternatives do exist, and arguably proffer solutions that are competitive in terms of both Capex and Opex, as well as offering better security and management capabilities.
This is also where cloud infrastructures come into play because the alternatives are all variations on the thin client, which is increasingly being seen as a good client environment as part of a cloud-based infrastructure. For many years thin clients have been derided as just a cheap, slow and inflexible desktop tool – aimed solely at the serried ranks of single-task, data-entry operators often found in large enterprises. But now thin clients are better defined as DaaS (Desktop as a Service). And because the service is provided to the desktops at the server level the ability for the business to manage the utilisation profiles, the security, the performance and resource allocation and a host of other factors is greatly improved.
In addition, the operational costs can be much lower for a number of reasons. The energy consumption per desktop is much lower for a start. All the leading thin client vendors now offer close to zero energy clients, with some of them being genuinely zero power drain when the desktop system is not in direct operation. With PCs they are either on and consuming power – even in standby mode – or they are off and not useable. With all the applications, services and data held on the servers, this also reduces the resources that need to be dedicated to end user maintenance and support, where IT support staff are often required to visit end user locations to resolve problems.
There are a number of financial issues that need to be considered as well. For example, the hardware costs of upgrading to PCs capable of running Windows 7 in an enterprise will be high, and such an upgrade will be a necessary part of the plan. But a typical medium-sized business of, say, 1,000 seats is going to cost at least £300,000 to equip – probably more when the high-end `road warrior’ laptop upgrades are included.
Yes, dedicated thin-client hardware can cost nearly as much if that route is adopted, but the systems themselves then have a much longer operational life than a PC, as they are not required to run the operating systems or applications, or managed data storage. They upload input to the system and download the results. Indeed, potentially redundant PCs can themselves be redeployed as thin clients rather than be scrapped – with all the possible costs of compliance with the Waste Electrical and Electronic Equipment Directive. Old PCs can be used until they die, which will greatly reduce the Capex budget demands.
This approach will probably increase the server component of the Capex budget, as hardware will be required to run the virtual desktops. But as many businesses are still in the throes of consolidating servers and datacentres as a precursor to any move to private clouds, the additional cost may not be that much. This will most likely be reduced further by the fact that the bigger the workload on the servers, the more powerful the servers that can be specified – and with the way that commoditisation of the technology continues to occur, the cost/performance ratio of individual servers gets better the more powerful they are. More powerful servers means more virtual servers running on each physical box, so the cost of virtual desktop instances can be kept low.
The licence cost of PC operating systems and applications will still be a significant component, and the site licences will be broadly similar whether they are virtual or real desktops. It is fair to say that this will be another area of cost in any upgrade, not least because Windows 7 will usually lead to another upgrade cycle to the latest versions of applications. This in turn will add installation and both user and support staff training costs, plus the potential for creating operational problems through inconsistencies and inter-operability miss-matches during any staged roll out of the upgrade.
A move to a virtualised desktop/thin-client model does open up the possibility of side-stepping some of these issues, perhaps with a move to open-source alternatives to the main applications suites. It also opens up the opportunity for businesses to move into the cloud, using managed services to provide the management expertise and shared resources to drive thin clients on user desktops.
And what about those `road warriors’ that need high-powered laptops? Well that market may well continue, but most of the thin-client architecture vendors also provide dongles that create a business-only partition within the laptop which can then VPN through to the server – and that is where all business-related operations take place, ensuring much greater security, management, operational compliance and the like. So it may well be possible for many of them to work effectively using the new, small, cheap Netbook laptops and a dongle.
Indeed, the same approach can be used with standard desktop PCs, so that office-based users can have access to the Internet outside of the enterprise firewall where they can do what they will, and a dongled, business-only environment managed and run only on the servers, with neither environment interacting.
As a final thought, there will also normally be an inherently higher level of security, particularly from external virus and hacking attacks, with a server-based virtual desktop model than with a traditional environment based on networked PCs, where every system represents a threat that must be managed – not least because of the potential for inappropriate behaviour and actions by individual users.


















































































Post new Comment