So for those who have not gone down the virtualisation route let me summarise the benefits from a business and technical perspective and whilst we are there let’s take a look at virtualisation in the context of cloud adoption and the benefits it brings in a cloud environment. The need to embrace virtualisation normally comes hand in hand with growing hybrid environments and having some of your IT infrastructure in the cloud and some on premise. In fact virtualisation is often seen as a stepping stone into cloud.
Why virtualise?
Let’s start by looking at the benefits that virtualisation brings to the business. Virtualisation can increase IT agility, flexibility, and scalability while creating significant cost savings. Workloads get deployed faster, performance and availability increases and operations become automated. This results in an IT environment that is simpler to manage and less costly to own and operate. Other key business benefits of virtualisation include the ability for the organisation to reduce capital and operating costs, to more effectively minimise or even eliminate downtime. IT can provision applications and resources faster, enable business continuity and disaster recovery, and finally virtualisation helps to simplify data centre management.
The key technical benefits of virtualisation are many and include:
· Encapsulation – the virtual machine is described as a small number of files, typically its virtual disk drive and a resource description file contains its virtual hardware requirements
· Virtual machines separate the operating system from the physical hardware, so they are no longer tightly locked to particular hardware through device drivers. This makes moving a VM from a Dell server, say, to an HP or Lenovo server much more straightforward.
· Live Migration – a virtual machine can be moved from one physical host to another with no downtime for the operating system. This is very useful for both load balancing, but also for maintenance or upgrading of hosts.
Best practice tools and approaches
Tools are available to virtualise current physical servers, such as VMware Converter and Platespin Powerconvert. However, often it is better to build a virtual machine from new and reload the application and data. As organisations choose to upgrade their operating systems and applications, they are normally doing this from a fresh build of Windows or Linux as a virtual machine. And in terms of best practices for virtualising servers, all Intel x86 servers are now candidates for virtualisation.
The main challenges around virtualisation are normally about the risk associated with the migration process, especially downtime, and around the licensing policies of the software being run. For this reason, the main servers still running on physical hardware tend to be large transactional databases such as SQL server clusters and Oracle.
To get the greatest benefits out of virtualisation it should be in a shared storage infrastructure. In the early days, this meant fibre-channel SAN or network-attached (NFS) storage. Recently, hyper-converged infrastructures have appeared on the scene from companies such as VMware (VSAN), Nutanix and Simplivity. This takes industry-standard x86 servers with large amounts of CPU and RAM, but it uses internal disks rather than a SAN, together with high-speed networking. Technologies such as flash and solid-state disks, as well as compression and deduplication have made these storage systems cheaper and faster for virtualised workloads. Software ensures that all the data is fully redundant across the cluster, with virtual machines load balanced at the same time.
Virtualising servers and storage in preparation for creating a private cloud
Virtualisation is a key requirement for the cloud, whether migrating or creating from new. Once virtualised, virtual machines can be migrated to cloud providers via export/import mechanisms such as the Open Virtualisation Format (OVF). In some cases the format of the virtual machine will need to be changed depending on the source and target format. For example, VMware to MS Azure will require conversion to Hyper-V, while Amazon AWS would require conversion to Xen.
Rather than migrating and converting, it is often better to replicate the data over a period of time using solutions such as Zerto, Veeam or DoubleTake. This will result in a much shorter downtime required to switch over from running on-premises to a cloud provider. As an established Cloud Service Provider, iland helps customers to migrate their physical and virtual servers to the public or private cloud.
If this is all starting to sound too good to be true let me reassure you, there are very few pitfalls. Sure, in the early days there was a slight performance overhead associated with virtualisation when compared to the raw physical server, but these days server and storage technology have all but removed that.
Future developments
Virtualisation technology has matured over the past few years. The new developments are really around automation and the final virtualisation of networking which together allow for what is often termed the Software Defined Data Centre. In this world everything from servers, storage and networking can be defined and controlled through software. The utopia is where virtual machines can be moved and managed using common networks wherever they happen to be.
Virtualisation offers a host of benefits and very few challenges as the market matures. That said without right-sizing VMs and failing to track and anticipate resource needs, businesses could face issues such as VM sprawl and over consumption, so it does need to be tightly managed. My advice is that businesses should plan for the long-term to make sure that they have enough resources on hand to meet future business demands.
