Infrastructure Control for Software-Defined Environments

By Andrew Hillier, CTO and Co-founder, Cirba

There is a clear trend toward making every aspect of IT infrastructure more software-defined.  It creates flexibility, agility, and enables common hardware to be used to create special-purpose configurations through clever abstractions.  Arguably this trend started with server virtualization, but it is spreading rapidly across networking, storage, and any other area that was previously “hardware defined”.  The ultimate goal for many is the “software-defined data center”, and although the definition of precisely what this is can vary, it is basically the nirvana of IT operations.

But as we all know, the software-defined data center isn’t something you can run out and buy; rather, it’s a state you reach when you adopt the right technologies and make the right changes to how you operate.  And it requires more than just bolting together virtualization, SDN (software-defined networking) and other technologies.  It requires a different way of thinking and a whole new level of control to manage all the moving parts.  In this sense the move to virtualization can be quite instructive, as the promise of virtualization was not immediately realized in many organizations by simply adopting hypervisors.  Management processes and tooling ecosystems needed to advance in lockstep with the hosting technology to achieve the true cost and operational benefits. The same is true in the software-defined world. Adopting software-defined compute, network and storage technologies is obviously important, but making them work together in a way that meets the needs of the business requires a level of operational control that is not present in most organizations.  This Software-Defined Infrastructure Control is key to aligning the capabilities of the infrastructure with the requirements of the applications, which in many ways is the true goal.  And to do this, there are several aspects of infrastructure management that need to move forward in order to enable this level of control.

Demand Management

If infrastructure is to be “fit for purpose” to meet the requirements of application demands, then there needs to be some way to describe what these requirements are.  This includes not only the CPU and memory allocation requirements of the components of an application (as one would get from a cloud request portal), but also the software requirements, compliance requirements, performance levels, storage tiers, geographical constraints, redundancy, and a raft of other considerations.  An understanding of the workload profile, including the time of day (or seasonal) patterns that dictate true resource consumption is also required.  Much of this is known in many organizations, and may have been used to procure hardware in the past, but is not used to match applications to existing infrastructure or to programmatically define what the infrastructure should be.

Capacity Management

Capacity management is focused on a “hardware defined” world, where demands are periodically assessed against server capacity, and upgrades or purchases planned accordingly.  This is woefully inadequate in a world where the infrastructure is programmable and the demands change on a daily basis.  The old “offline” model must be replaced by an “online” version that is constantly assessing supply vs. demand and making adjustments.  The new variables are where workloads can go, how much resource they should be assigned, and what the infrastructure must look like to deliver this.  The spreadsheets, trending algorithms, and utilization charts used by most capacity managers simply aren’t up to the task.

Policy

At the heart of all of this is the operational policy that governs how supply and demand are matched, aligned, and controlled.  This is the contract between supply and demand, the lease agreement of the hosting world.  But if you look around most organizations you won’t find it – all you will find is simplistic thresholds spread across operational tools, and people who know all the details and subtleties of how the environments operate but have no way to codify them.  To control a software-defined environment, or even to make a traditional environment more software-defined, these policies must be captured and used programmatically to plan and operate the environments.

Automation

Last but not least, the move toward software-defined infrastructures is invariably coupled to the move to higher levels of automation.  This is often out of necessity – the complexity of these environments is beyond what humans can deal with, and the goal is to manage the next generation of infrastructure with fewer people, not more.  Many organizations focus on automating the provisioning process, particularly in cloud environments, but the automation requirement for a software-defined operation goes beyond this.  For example, automating the routing decision of where new VMs should end up, locking in the capacity, placing VMs, allocating resources, the ongoing optimization of infrastructure and forecasting future requirements must all be done automatically.  This requires accurate, detailed models of existing and inbound demands, fine-grained control over supply, and policies that bring them together.

By focusing on these four areas, Software-Defined Infrastructure Control is an essential component of an overall software-defined strategy.  It helps manage the complexity that is introduced by more advanced compute, network and storage technologies, and it also enables existing virtual and cloud environments to operate in a more software-defined way.   Taking a different approach to infrastructure management will help organizations effectively cope with the dynamic nature of cloud-based operational models, while still allowing business critical applications to be hosted safely and efficiently.  Just describe demand, optimize supply, and automate.

Andrew Hillier, CTO and Co-founder, Cirba, Inc., +1.905.731.0090, Toll Free: +1.866.731.0090

Cirba has re-imagined infrastructure control for the software-defined era.

We’re enabling the world’s most successful  organizations to scientifically balance infrastructure supply and application demand—creating a demand-driven approach to infrastructure management that maximizes efficiency and cost-savings while reducing risk. www.cirba.com

Advertisement

Follow Us:

Sitemap | Privacy | Copyright © © 2017, WSTA®, All Rights Reserved.