Virtualisation has a number of features which can be used to help avoid disruption and improve availability. However, if a system is not designed correctly, virtualisation can become a single point of failure and reduce resilience.
3.1 Design the virtualisation system to have redundancy
Systems that use virtualisation should be designed with redundancy in mind. A denial of service attack on a virtualisation platform can impact all the virtualised machines running on it. This could be achieved, for example, by compromising a poorly protected virtual machine and using that to launch an attack on the underlying platform.
Design for an acceptable level of disruption, using multiple layers of redundancy, if required. Measures which can help with this include multiple instances of hardware, network paths, storage arrays, splitting the system across different locations, online and offline backups.
3.2 Design to be flexible, scalable, and highly available
Many virtualisation platforms have features which enable them to be flexible, scalable, and highly available. Your design should take advantage of these features, helping you to produce stable and resilient systems.
Use of these features should be applied throughout the virtualisation stack. This will also enable ongoing maintenance, such as applying security patches to the underlying virtualisation platform and infrastructure, without impacting the virtualised systems running on them.
3.3 Ensure the management infrastructure is highly available
If a virtualisation platform is disrupted, the management infrastructure needs to be available for administrators to fix the issue. The management infrastructure should be separate from the operational infrastructure, so that an attack on operations cannot affect management activities.
