The future of server virtualisation
- 22 February, 2018 10:56
Server virtualisation has changed. It has rapidly evolved and is now a mature market. More than 80 percent of x86 server operating system (OS) instances have already been virtualised. If you’re an infrastructure and operations (I&O) leader, now’s the time to explore new approaches to virtualisation if you want to gain further advantage moving forward.
The maturity of the market opens potential opportunities for your organisation to deploy alternative virtualisation solutions for different uses or when infrastructure refresh decisions prompt reconsideration of existing virtualisation.
Hypervisor-based virtualisation of x86 server infrastructure is the most widely used form of compute virtualisation within enterprise environments today. This is largely due its ability to improve server consolidation, agility and availability.
Looking ahead, Gartner is seeing growth in both OS virtualisation (Linux and Windows containers) and VM-integrated containers, especially for new and cloud-optimised workloads.
Even within these categories, however, multiple factors must be considered when looking at market trends and future directions in compute virtualisation. What’s the cost and effort to change? What other projects and initiatives would have to be delayed to swap-out an incumbent, and for what benefit? Many will claim that if it “ain’t broke, don’t fix it.” A decade’s investment in tools and processes shouldn’t be dismissed just because the technology (or vendor) isn’t as cool as it once was.
Other market trends are becoming more prominent too. For example, look at the increasing role and influence of developers and open source. Developments include mainlining of namespaces and groups within the Linux kernel; developer adoption of application packaging using docker; DevOps; and the increasing dominance of virtualisation projects such Kernel-Based Virtual Machine (KVM) and oVirt within the open source community.
The future looks cloudy
Adoption of public cloud/as-a-service offerings is also growing, leveraging open source, hypervisor-based virtualisation delivered via public cloud. The most notable example is infrastructure-as-a-service (IaaS). For most of its history, AWS has relied on Xen, although that is changing now.
Then there’s the changing nature of the future data centre. This includes software-defined infrastructure (including data centre, storage and network) and hyperconvergence. Most organisations are looking for a hybrid, multi-cloud environment; however, there are multiple ways to achieve that. For most, the jury is still out as to just how far the journey to public cloud will go. Newer offerings such as serverless computing will only amplify this in the future.
Finally, there’s the increasing appeal of integrated solutions. The virtualisation license for these may be embedded, for example, in a hyperconverged infrastructure solution (Nutanix) or OS (Microsoft Windows and Linux). Somewhat similarly, some vendors provide preferential software licensing arrangements when using their virtualisation technology, such as Oracle.
Take advantage of revirtualisation opportunities
Given the high level of adoption, the market for x86 server virtualisation infrastructure is relatively stable, with most enterprise investments being made for the medium-long term. However, this longer-term view shouldn’t be allowed to generate complacency.
Multiple factors can and should be reviewed when considering how to (re) approach this market. These include innovations in product functionality, rising maturity of open source technologies, increasing adoption of cloud computing and broader strategic plays by infrastructure mega vendors — not least of which is Microsoft.
It’s important for your organisation to develop experience in newer forms of compute virtualisation by taking advantage of revirtualisation opportunities. Explore the suitability of new providers and different approaches to ensure optimum platform choice.
In particular, rising developer interest in containers warrants a proactive approach from I&O teams. Where suited, use of VM-integrated containers may suit a more conservative approach to supporting containers in production environments.
Determine the right approach
In determining the best path forward, aim to continue leveraging x86 server virtualisation infrastructure for the majority of your workloads – 80 percent or more would be a suitable target. But also embrace an approach that manages two separate but coherent styles – one focused on predictability; the other on exploration. This can be done by optimising hypervisor use to suit differing portability and availability uses, while evaluating the suitability of VM-integrated containers and/or OS virtualisation for agility and productivity.
Determine strategic fit and overlap of virtualized x86 server infrastructure with adjacent technology by collaborating with your network, cloud and storage teams. Gain consensus on the role of hyperconvergence, public cloud, software-defined infrastructure and next generation virtualised infrastructure. Recognise that various forms of compute virtualisation will be necessary.
Virtualisation is dead; long live virtualisation!
Michael Warrilow is a research vice president at Gartner and chair of the Gartner Infrastructure, Operations Management & Data Centre Summit, 30 April-1 May 2018. His focus is on the evolution of modern infrastructure software, including software defined, virtualisation, containers and cloud infrastructure.