However, there is another paradigm that is becoming increasingly popular. The rise of blade computing offers the centralised benefits of both thin client and mainframe technology, as well as desktop functionality regardless of network availability.
The old adage that “time is money” has never been more true than it is today. With an increasing amount of business being conducted in real time, any outage – be that of equipment, power or even personnel – can have a significant impact on the bottom line. One of the key factors in any organisation is to minimise any downtime and to ensure business continuity.
In terms of equipment, this has been further complicated by the need for remote computing, whether that be the road warrior, dialling in from a hotel, or the home worker, accessing the corporate network with all of the security issues that that implies. With a geographically diverse network, making sure that that most basic of tools, the desktop PC, functions smoothly and efficiently, has become a matter of priority.
When the desktop consisted of a dumb terminal connected to a mainframe, maintenance was centralised. The dumb terminal consisted of nothing but a monitor and a keyboard, with all of the processing power centralised in the machine room. Of course, a mainframe malfunction usually led to a system-wide meltdown, with every terminal rendered useless. However, a single repair job in a single location would be sufficient to bring the network back online.
The introduction of PCs brought with it a multitude of benefits, with processing power on the desktop allowing for increasingly tailored environments available for the user. Mass storage was provided by dedicated servers, but even if the network crashed, the user still had a degree of functionality at their fingertips. However, the need to access the Internet, as well as the corporate intranet, meant that such independence became less and less useful, while 24x7 access was becoming vital.
There was a tentative move to a half-way house between the dumb terminal and the fully-featured PC in the form of the thin client; however, it soon became obvious that this offered the worst of both worlds, with a single point of failure in the server, and a lack of standalone functionality in the client. While mainframes and thin clients are still used in many corporate networks, they are almost always imbedded in a standard client-server topology.
However, there is another paradigm that is becoming increasingly popular. The rise of blade computing offers the centralised benefits of both thin client and mainframe technology, as well as desktop functionality regardless of network availability.
With blade computing, the desktop consists of a keyboard, mouse and monitor. A small router connects this to a centralised rack in which the full PC is re-envisioned as a rackmounted board containing the processor, memory, hard drive, graphics card and the operating system. This blade-rack is connected to the servers which comprise the office network.
The rack can be anywhere, allowing for both geographically diverse office environments – including homeworkers, who simply need a secure line to the blade rack – and temporary office environments. It offers complete flexibility: to increase the number of desks, the basic package is simply delivered and plugged in, with all configuration done centrally.
However, the benefits of blade computing go far beyond centralised maintenance and geographic flexibility. The equipment on the desk is effectively nothing more than a shell; all the functionality and processing power is supplied by the individual blade in the blade-rack. Companies such as ClearCube have refined the concept with the introduction of a comprehensive management application which allows real-time flexibility of desktop-to-blade assignments.
The most obvious consequence of this is the creation of a fault-tolerant network. If a particular blade should suffer a fault, the management suite enables the network administrator to switch in a different blade immediately: the user suffers only momentary downtime before their desktop is up and running with full access once more. IDC calculates that this results in a 60 per cent reduction in desktop support, a 13 per cent reduction in help desk overheads, and a 36 per cent reduction in mean time to repair.
There are other implications. Many companies today are not just spread across a single country: outsourcing to other countries is common, especially with call centres which need to offer round-the-clock availability. With blade computing, a call centre operative in London can sign off at 5pm, releasing the blade which can then be immediately reassigned to another operative in India who is just starting their shift. No processing power is wasted – it simply gets redirected, with no unused PC sitting on an unused desk. This is a version of hot-desking without the overheads of a fixed PC or the security implications of a corporate laptop.
Some of the possibilities of blade computing allow for the realisation of computing possibilities which were until recently the province of high-end clusters and arrays. In an environment in which the blade rack would be unused outside office hours, the potential computing power can be brought to bear on processor-hungry number-crunching applications such as market analysis and forecasting. In many respects, this is a evolution of the batch transaction processing that mainframes would perform outside office hours.
From this, it is only a short step to clustering. With dual boot-up, each blade can be treated as a node in a loose Beowulf cluster, and each blade rack can then be clustered to provide a resource of pure processing power available to any application within the company. Organisations then have access to both high availability or high performance through any unused racks.
All of this sounds too good to be true – or too good to be true without significant investment. But the figures speak for themselves: according to IDC, for an outlay of 44 per cent above the standard client-server model, companies employing blade computing have reported an average ROI of 413 per cent, paying back the cost of deployment in less than seven months.
The benefits are manifest, above and beyond business continuity. As well as reduced support costs and downtime, there is also the all-important question of asset security. With the blade-rack locked away in a secure environment, it is less susceptible to either accident damage – coffee over the keyboard? – or theft. ClearCube has taken this one stage further: it is possible to fine-tune the user’s access to mass storage devices such as floppies, CD/RW or flash-keys to prevent information theft. There is even an ergonomic aspect to this: without a CPU on every desk, the environment is both cooler and quieter.
Investigating the feasibility of blade-computing is a win-win scenario. It offers the centralised maintenance and management so beloved of old school network engineers, as well as desktop functionality tailored to the user. The collateral benefits – especially where the bottom line is concerned – cannot be disputed. This is definitely a technology that deserves to be given a road-test.
