Server virtualisation is the first step to achieving higher productivity from IT infrastructure. Enterprises need to get it right before they can move to the next stage. NME examines the ways in which enterprises can shift smoothly to virtualisation and the mistakes they should avoid.
The IT manager of the average enterprise in the Middle East is well aware that he only has to start uttering the word ‘virtualisation' to have vendors flocking to the door. In other words, enterprises are quite spoilt for choice when it comes to virtualisation technologies.
In theory, server virtualisation is about creating multiple servers within a certain piece of hardware to obtain maximum performance. In practice, it can be much more complicated.
However, virtualisation is a very broad term. It includes in its range a set of technologies that can change the way an enterprise functions and uses technology over time. But most people in the industry agree that the shift to these technologies has to begin at the server farm or, in other words, the datacentre.
Gautam Srivastava, VP for sales and marketing and MD of the Middle East, Africa and Pakistan for AMD says: "Within a datacentre, a server spends nearly 70% of its life doing nothing. Bringing in virtualisation, can help enterprises get a lot more from their hardware."
"Virtualisation is not just about servers. It is infrastructure, software, services, storage and PCs - it embraces all of IT. Servers are the first step. People can get their first experience of the technology by virtualising their servers and this proves to be an excellent first step in looking at all the aspects of virtualisation," says Arnaud Gardin, servers and solutions marketing manager for NEC Computers.
In theory, server virtualisation is about creating multiple servers within a certain piece of hardware to obtain maximum performance. In practice, it can be much more complicated than that and as enterprises in the Middle East move rapidly towards adopting this first step to enterprise-wide IT virtualisation, it is essential that they avoid some common mistakes while deploying and using these virtualised platforms across the servers in their datacenters.
The path to higher productivity
Virtualisation comes with many benefits, including reduced power and electricity costs, but arguably the biggest driver for it is the promised optimal performance of IT infrastructure.
The first step to higher productivity from servers is asking yourself whether you really need virtualisation. Many an enterprise tends to adopt these set of technologies while blindly following the huge hype that is created around them. However, it is essential to understand that virtualisation is not for all enterprises.
"The first thing is, not everything needs to be virtualised. It's not like next year, we need to virtualise every server there is. The competition today talks about virtualisation as a solution for everything because all they sell is virtualisation - for any ill, issue or complexity, virtualisation is the answer, but actually that's not accurate," says Zane Adams, director of server marketing at software vendor Microsoft.
"Virtualisation has become such a buzz word that enterprises just move towards it because it is the next big thing instead of giving thought to or studying how it can be of benefit to them specifically. I do understand that this is a hot topic but enterprises need to take a step back and analyse it in detail before taking a decision," says Chandan Mehta, regional product marketing manager for enterprise products at IT vendor Fujitsu Siemens.
One way to be sure that you need virtualisation is to check on the current utilisation of your hardware.
"The common mistake will be if the move to virtualisation did not include a initial professional assessment phase of current infrastructure and server load that will provide the customer with a view on what the new virtual infrastructure will look like and the challenges if any he needs to overcome," says Antoine Aguado, regional manager of Citrix Middle East.
"Enterprises have to begin with current systems and the resources they are using. They have to check the utilisation of servers and storage. There are good tools for doing this. If there is a high utilisation of servers - or anything above 55% of CPU usage - there might not be a need for virtualisation at all," says Mehta.
According to NEC's Gardin, one of the other common mistakes enterprises make when they decide on virtualisation concerns the assessment of the footprint of the applications.
"Inventory or asset management is essential before the move to virtualisation. At the preparation level, it is important to size future servers and the physical-to-virtual changes that will come along. Enterprises have to invest effort in understanding assets and building the target infrastructure," he says.
The application footprint on a typical server ranges between 15% to 20%, but it is essential for companies to understand the exact nature of their applications for an effective move to virtualisation. They also need to understand the type of usage that these applications undergo in the organisation and the availability levels that they would require.
"Many people concentrate on the memory and the CPU levels but forget that there can be a bottleneck at the input/output (I/O) level. When you are doing an application footprint analysis, it is essential to understand its I/O consumption. Companies also need to decide the kind of redundancy they require and the restoration parameters for applications before making the move," adds Gardin.
Most industry vendors advocate that companies move non-critical server applications to a virtualised environment before trying any of the critical ones. They also warn that enterprises should study application capability or suitability, along with licensing practices (which vary for each vendor) for virtualisation before taking the step.
Additionally, enterprises should not blindly reuse old platforms or hardware for virtualisation when making the move. Application availability might not be provided by existing hardware and to derive effective value from an investment in virtualisation companies might need to put in initial capital in new hardware, as well as new software.
The management aspect
The move to virtualisation can bring about a lot of changes to a company's datacentre, and to handle these changes effectively and to profit from them, the firm might find itself in need of new organisational and management products and practices.
"The most important step is how do I manage after I virtualise? Because think of it from this perspective - once I run multiple virtual machines on a server, in essence I'm running multiple servers on that box. I need to start managing that differently as an IT administrator. Then the tools that we use need to be the same for the physical server as well as the virtual and shouldn't be a complex, different learning experience. It goes from assessing and deploying, to managing because TCO reduction and efficiency comes from being able to manage once you do virtualise," says Microsoft's Adams.
"As systems infrastructures are becoming more complex with virtual systems, they require management solutions that understand and manage the relationships between infrastructure components. A good management software solution must automate server provisioning based on policy, enable policy-driven configuration auditing for identifying changes in server configuration before it can impact service, and balance system resources (physical and virtual) to optimise datacentre assets and reduce unnecessary spending caused by under utilisation," says Antoine AlIbry, regional technology specialist for technology sales at CA.
"The datacentre must be empowered with a solution that employs visualisation, automation, self-management and resource-sharing technologies to optimise IT resources and help ensure the performance and availability of business-critical processes. It needs a central platform to monitor and manage complex enterprise environments, enabling companies to implement and build the IT infrastructure that suits their needs," he continues.
Citrix's Aguado agrees with him stating: "Global enterprises which move to server virtualisation will need to put in place automated provisioning and virtual infrastructure workflow manager."
While the shift to virtualisation can seem like a complex task, enterprises can, and sometimes should, use external consultants to make the path easier for them.
"This depends to an extent on whether the enterprise has staff with the necessary virtualisation related skills. If they have sufficient resources then they might not need a consultant. If they do not, then they would be well advised to get a consultant who has got the relevant experience," says Mehta from Fujitsu-Siemens.
The relative success of a virtualisation project also depends, to a large extent, on the vendor in question.
The application footprint on a typical server ranges between 15% to 20%, but it is essential for companies to understand the exact nature of their applications for an effective move to virtualisation.
"Customers have to pick a vendor who is both an expert in hardware and software. They should be cautious about this fact and make sure, at the beginning, that they have expertise on both sides of the fence," says Gardin.
Enterprises should also put in the effort to check out vendor references thoroughly, visit proof-of-concept centres globally and run small pilots within the organisation to ensure that the solution is exactly what they require. They should also ensure that the vendor has adequate service and support options in the country, region or globally - depending on the girth of the company and the breadth of the virtualisation project.
In the long term, customers will need to develop an enduring partnership with the vendor to ensure that their virtualisation investment continues to perform smoothly and gives continuous benefits. They would also need to adopt best practices, train their internal staff and get in the right tools to extract the most from the virtual platform.
"Many customers pull back from full implementation because of the lack of operational maturity. In other words, they do not have fully documented, time-tested processes on how to upgrade, change and maintain their IT infrastructure. Companies need to focus and make sure these processes are in place before moving to virtualisation," says Basil Ayass, MENA product manager at Sun Microsystems.
"A lot of customers do not invest time in and do not value training enough. A virtual infrastructure simplifies your IT environment yet it requires different skill sets. Not more complex, not harder - just different skill sets. We really recommend training; a lot of customers go into virtualisation without the required training," he adds.
Far from the end
Server virtualisation is essentially the first step to obtaining higher productivity from the entire IT infrastructure of an organisation. Once an enterprise has a functioning virtualised-server environment, or in parallel with its deployment, it will need to pay attention to other, crucial areas like storage.
"One of the bigger challenges for some customers in virtualisation is not having any sort of storage in place. In order to get a full feature set, there needs to be some centralised storage in use at the firm," says Thomas Huber, EMEA manager systems engineer, eastern region at VMWare.
Virtualisation in storage technologies becomes a necessity for enterprises that want to take virtualisation beyond just servers, but that is another story altogether.
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.
Subscribe to Arabian Business' newsletter to receive the latest breaking news and business stories in Dubai,the UAE and the GCC straight to your inbox.