Converged infrastructure is very similar to an Airbus 380 – it has a tremendous amount of capacity, but there is no way a human can control it without the help of a very sophisticated control system. And just as this type of control system is relatively new to aircraft, the control system required to manage converged infrastructure is equally new, and very different from the tools and processes used to manage legacy infrastructure. This isn't because complexity is anything new – data centers have always been extremely complex. It is because this new complexity is different – what was physical complexity is now becoming virtual complexity, and components and connections that were once in plain view for people to see are now becoming sealed in what is, in some cases quite literally, a black box.
This shift in complexity is underscoring the need to revisit several aspects of IT management, some of which need to improve and others which until now have been completely ignored:
Policy – Every IT environment has a purpose, and to fulfill that purpose, it must behave in a certain way. This includes how well it must perform, what level of availability it should offer, how it deals with security and data protection, how it manages multi-tenancy and resource sharing, and so on.
In legacy physical environments, and even early virtual environments, these criteria were not actively managed, as the environments were too static to warrant it. But in converged infrastructure, these management criteria must be formalized into proper policies, and these policies need to be clearly defined and actively controlled. Failure to do so will result in “workload anarchy”, where there is no clear authority over the operation of the environment.
In this sense, policies can be thought of as the contract between supply and demand, setting clear operational goalposts and protecting workloads from each other through tightly controlled rules of engagement.
Bookings – Virtual and cloud environments are much more dynamic than their physical predecessors, as they allow the deployment and decommissioning of workloads with greater frequency, higher speed and lower friction than ever before. When this is combined with converged infrastructure, which often provides a tremendous amount of capacity in a single unit, this pipeline of activity is concentrated. This makes managing what is coming into and leaving the environment critical to maintaining control.
Without visibility into upcoming demand, you are forced to estimate requirements and waste capacity (or worse, risk running out). An analogy is useful to demonstrate this: if physical environments are like houses and virtual environments are like apartment buildings, then clouds are like hotels, and clouds build on converged infrastructure are like very large hotels, with extremely busy lobbies. A good reservation system is essential to establishing the necessary control over bookings and available capacity, and while one wouldn’t dream of managing a hotel without a reservation system, IT is currently lacking in this area.
Analytics – The sheer amount of data created by management and monitoring tooling for these systems can overwhelm even the best infrastructure manager. Compounding this challenge is the flow of workloads into and out of them, and the operational policies required to govern them – and it is extremely difficult to understand what is required to proactively head off capacity and performance risks and understand infrastructure requirements.
Increasingly complex decisions about how much capacity is required, how to allocate resources, where to place workloads must be made more and more frequently, making it clear that the manual ways of making decisions will not survive. In fact, some speak of an “analytics revolution” being ushered along by the move to cloud and the shift to convergence. This not only provides the control system referred to in the airplane analogy, but when done properly can transform the way IT environments are managed.
Operational data, policies and bookings can be used in predictive analysis to gain insight into the future that isn’t otherwise possible. This allows IT managers to shift from a reactive mode of fire-fighting to a proactive one, in which they spend less time “just keeping up with things” and more time “getting ahead of the curve”.
Convergence provides many benefits, but in virtually every advantage there exists management challenges that must be addressed. Rather than see this as a negative, however, organizations should see these as an opportunity to revolutionize the way they manage infrastructure.
By adopting new ways of thinking and new-school control systems, humans can effectively tame the new complexity and take control over their converged infrastructure.
By Andrew Hillier, CTO and co-founder of CiRBA
Andrew has over 20 years of experience in the creation and implementation of mission-critical software for the world's largest financial institutions and utilities. A co-founder of CiRBA, he leads product strategy and defines the overall technology roadmap for the company.