Legal Technology Journal

The benefits of virtualisation are many – it’s time you looked at it more closely. Some technologies are a bit like fashions – they get invented, forgotten, reinvented and sold as the ‘new black’. If you have been in IT for a few years, you could be forgiven for thinking that server virtualisation is one such technology.

As a basic concept it’s been around since the 1970s. At that time computing power was very expensive and if you were to pay out a small fortune on a shiny new mainframe you wanted to make sure that every single clock cycle was used, accounted and billed for too.

With the advent of the personal computer and relatively cheap and powerful Intel-based servers, server virtualisation was largely forgotten. After all, if you could buy a new piece of tin for each application, why not? If the application didn’t run fast enough, you could simply get a bigger, more expensive piece of tin or run the application on multiple tin boxes.

Not processed

Perhaps the most striking IT statistic you can gather concerns server utilisation. Speaking very generally, most servers in a typical organisation will usually run at about between 5% and 15% utilisation. Turn that statistic on its head and it means that your servers may be wasting up to 95% of their processing power for most of their operational lifetimes.

Apart from the common sense principle of trying to get the most value from the servers you buy, there are a few other, more concrete reasons to justify the need to find a way of squeezing every ounce out of the firms investment in its servers.

Bad for cashflow and the environment

There have been some advances with regards to power management within the latest generations of server that have helped reduce some inefficiencies, but most of that unused processing power will still be wasted as heat. This means that a business potentially loses money twice – once for purchasing a server that has too many resources for 90% of the time, and again because it wastes electricity directly and indirectly through the need to be cooled by powerful air-conditioning units.

In practical terms these losses may appear small in comparison to overall operating costs, but bear in mind that a typical 200-user law firm may have as many as 20 or 30 servers and you start to see a significant multiplier.

Saving space

Firms traditionally love to keep their servers and their data locked away in a nice air-conditioned room in their head office. This head office is likely to be prime real estate where every square foot of space costs the firm money. Obviously, the most efficient use of that space will be to fill it with fee-earning staff. So why devote more space than is absolutely necessary to the server and comms rooms? If you can cut the number of servers from 50 to 5 then you have cut the number of racks from 10 to 1. This would allow the server room to be downsized and perhaps fit more fee-earners in the newly created space.

Now, if yours is one of the growing band of firms looking to place their primary data centre in co-located facilities – perhaps with their ISP – then every inch of rack space you use will have a direct monthly or yearly cost. If you can cut the number of boxes from 50 to 5 that’s a huge saving on co-location costs.

What’s going on?

The better server virtualisation products will provide some very useful tools for finding out just how much of each resource (memory, disk, network, processor etc) each virtual server is demanding, and they will provide a full history along with nice charts of how these resources are changing with time across your entire virtual infrastructure. This information is not usually easily available using traditional server approaches, and certainly not without investing in some reporting tools. So it can be difficult to get a coherent history of all resource utilisation across every piece of tin within the firm.

Remote working

Also, because you can easily manage every aspect of a virtual server remotely there is little need to actually access the server room other than to maintain the few physical servers left. Virtual servers can have their resources allocated on the fly just by clicking a mouse; no longer does your IT team have to schedule down time for a simple upgrade. If a firm has multiple remote data centres this can translate into fewer transport and people costs.

All day every day

Here we come to perhaps the most useful use for server virtualisation. Many firms today need to maintain a 24/7 operation. Whilst this can be done using traditional, hardware servers it can add huge complexity to otherwise simple implementations. Server virtualisation offers a potentially simpler route. Some platforms are capable of moving virtual servers between pieces of hardware without the need to power down that virtual machine.

This simple ability provides some very powerful solutions to the problem of keeping all your applications running all the time. It allows running servers to be migrated from failing or ageing hardware without the need to shut down, meaning users can carry on accessing the applications on those servers oblivious to the fact that it is now running on a different tin box.

Because of this flexibility the IT department is able to fix and upgrade the physical servers during normal working hours with no down time. If a physical server does die unexpectedly, the virtual servers it was running can be instantly powered up on another physical server. No more waiting around while the hardware is fixed.

VMware in particular has taken this a stage further. Its software allows virtual machines to be moved around physical servers automatically depending on how many resources each virtual machine needs and whether it detects an impending hardware failure. For the business this means that systems will remain available and responsive no matter what time of day it is. Because this is all done at the virtual layer, its complexity does not grow as fast as more traditional methods such as clustering, where each new server actually requires two new pieces of hardware.

Furthermore, because of the ability to snapshot a virtual machine whilst running, backups of an entire virtual server can be made in a few seconds. Okay, the actual backup of the snapshot will take a little longer, but you have successfully reduced the backup window that would affect the performance of the virtual server.

There are benefits for disaster recovery (DR) too. If you run a virtual environment, it is possible to mirror that virtual environment in a remote DR data centre with a minimum of hardware investment, and potentially the DR data centre could be up and running within minutes of the primary one failing. No need to rush hardware across the country, or to hastily build existing hardware with the latest images, or to maintain a complete mirror image of all your physical servers in both locations. Obviously, DR is a bit like insurance, and the more you pay the better you are covered. But with the reduced hardware and space costs, and with the inherent ease of management that it would bring, a virtual infrastructure has clear advantages for any firm and can provide a more robust solution if carefully planned.

And there’s more…

There are a whole host of other benefits such: as speed of new server provision – it takes minutes and not weeks; standard builds – servers can be built from templates within a few minutes of a mouse click; legacy servers – no longer do you have to maintain legacy systems on old hardware that’s hard to find parts for; test and development – you can take a snapshot of a running virtual server before applying the latest patches or updates, so if the patch breaks the server it can be rolled back to the snapshot within a minute or two without the need to go hunting for a backup; and standard drivers – all virtual servers will have the same drivers regardless of the underlying hardware that is running it, simplifying troubleshooting and easing their management.

An all-out winner

It is difficult not to evangelise about the technical benefits of server virtualisation. However, it does have some very clear benefits for firms currently battling with problems with the size, power requirements and manageability of their data centres.

Server virtualisation also has benefits for any firm that has yet to experience these problems. It allows for great flexibility in provisioning new legacy applications and maintaining them. It provides a platform that allows the firm’s IT to respond very quickly to the firm’s needs, and to grow or shrink as quickly as the firm decides. It also allows new offices to be set up very quickly, and servers for those offices to be pre-configured without the need for technical staff to be shipped to the new location other than to take delivery of the hardware.

However, it is not a one-size-fits-all technology. How it’s implemented, the exact size and shape of the project and how many servers you can cram onto one box depends very much upon the individual needs of the firm and its perception of comfort levels. Some vendors tout ratios of 40 or 50 to 1. Whilst this is possible, take these sorts of figures with a pinch of salt. But it is surprising just how high that ratio can be – 10 or 20 to 1 is not so unrealistic.

Selling the system

Okay, now you know the advantages, but how do you sell virtualisation to the decision-makers? Talking about a technology to some of them can be like something out of a comedy sketch. No matter how much sense the solution makes, you are greeted with an expressionless look – perhaps with slightly glazed eyes – and you half expect to hear the words ‘Am I bovvered?’ It’s a skill they have learnt after years of technical people trying to baffle them with science to get their pet project approved with little regard to how it will improve the firm, the lives of its staff and ultimately its clients.

Server virtualisation is not like rolling out a new PMS, DMS or CRM system as it will, if correctly implemented, go completely unnoticed by the firm. So how does the firm perceive and measure the benefits if it doesn’t notice any change? Perhaps this is why only 30% of organisations in the UK have looked at or implemented some kind of server virtualisation project. I suspect the percentage of law firms is far smaller. When Winckworth Sherwood began its virtualisation project last year, there did not appear to be another firm in the UK that had virtualised more than a handful of their servers.

Because it’s a technology largely hidden from the firm, often the most direct way of demonstrating real-world benefits of going virtual is to do a cost justification exercise and to compare the costs of maintaining the current status quo and of going virtual.

Server virtualisation doesn’t need to be a ‘big bang’ project. VMware training highlights the need to carefully test and benchmark your servers within their virtual environment before going live. Indeed, most implementations appear to have been carried through carefully, integrating the systems with other projects. Any project that demands a new set of servers, data centre or a server refresh should at least weigh up the benefits of implementing a virtual server infrastructure. It’s not a universal solution and it may not fit 100%, but it will more than likely cover the remaining 95% of the server infrastructure.

Finally, in the longer term firms will need to embrace server virtualisation. Microsoft in particular has said that it will embed virtualisation technologies in the next versions of its operating systems. With the improvements being made in enterprise virtualisation products, and with virtualisation tools from both Microsoft and VMware made freely available for download, can your firm afford to ignore it? 

Sam Luxford-Watts is head of IT at Winckworth Sherwood.