The Invisible Hand of the Data Center: Analogy or Algorithm?
by Shmuel Kliger May 20th, 2013
I had the pleasure of presenting to the members at the recent London VMUG. My talk was entitled “Software-Defined Control – Transforming IT Operations, the Invisible Hand of the Data Center.”
In today’s Software-Defined “X” world, it’s easy to get blinded by the hype and marketing around the term Software-Defined. So I wanted to convey to the members of the London VMUG how VMTurbo Software-Defined Control (SDC) addresses the biggest challenge in operating virtual data center and cloud deployments – the Intelligent Workload Management (IWM) problem. By providing a software control plane that represents your environment as a marketplace of buyers and sellers that trade commodities that happen to be compute resources, you can control your environment in the desired state in which application performance is assured while the environment is utilized as efficiently as possible.
During my presentation one of the members tweeted that he liked my marketplace analogy. This wasn’t the first time I heard this from people with whom I speak. But it isn’t an analogy; It is our patented technology and it solves the IWM problem. Let me spend a bit of time here explaining what I mean.
As we continue to scale out virtualization and cloud deployments, enable workload mobility across heterogeneous data centers and hybrid clouds, and introduce more and more moving parts to our converged fabric and storage with the Software-Defined world, the IWM problem becomes increasingly critical. The question of “What Workload to run Where and When to maximize your ROI from whatever compute capacity you may have,” or, in other words, the IWM problem of “how do I assure application performance while utilizing the environment as efficiently as possible” is the hardest and most important challenge we face in trying to unleash the power of the Software-Defined World. This problem is important because if you don’t solve it: 1) you impact the service you deliver and reduce your revenue, 2) you increase your capital expenses by spending too much money on compute capacity, and 3) you increase your operational expenses by continuously fire-fighting and troubleshooting. This problem is complex because hundreds of thousands of continuously fluctuating metrics and constraints, across the entire environment and IT stack, must be continuously analyzed. No human, regardless of how many reports, trends or alerts he looks at can solve this problem. It must be solved by software. But how?
As previously mentioned, VMTurbo SDC represents the virtualized data center as a marketplace of buyers and sellers who trade compute resources. All entities in the environment are buyers and sellers. The Hosts, Data Stores, VMs, Applications, Data Centers, Virtual Data Centers, Zones, Regions, Filers, etc., are all buyers and sellers of compute resources. For example, the hosts sell memory, CPU, IO, Network, etc. The Data Stores sell IOPS, Latency, etc. The VMs buys these and sell vMemory, vCPU, vStorage, etc. Furthermore, all constraints—configuration and business—are also commodities traded in the market. The sellers continuously price their resources as a function of their utilization. The buyers, at the same time, continuously shop for the best deal they can get for the resources they consume. The “invisible hand of the market” drives the environment to an equilibrium where the workload demand is best satisfied by the supplied capacity at every layer of the supply chain, i.e., at every layer of the IT stack. The equilibrium is the desired state to control your environment such that application performance is assured while the environment is utilized as efficiently as possible. In solving the problem, VMTurbo SDC drives a broad set of actions to control the environment to the desired state, including workload initiation and termination, workload placement and configuration (initial and continuous), compute and storage provisioning and de-commissioning, compute and storage configuration, and more. With VMTurbo SDC, you control your environment in the desired state now and in the future, because whether you are trying to control the current workload, about to on-board new workload, or planning how the environment will look in a day, week, month or year from now given any changes and fluctuations you may face, VMTurbo solves for you the problem of how to control your environment in the desired state.
Nearly two years ago, Marc Andreessen wrote about software eating the world. And, looking at today’s software-defined world, it really is all about software. Perhaps I have to concede that the marketing hype may be right. The Software-Defined world is here and requires Software-Define Control. We apply the “invisible hand of the market” algorithm to do exactly this–it isn’t just an analogy!