VMTurbo Questions VMware, Virtualization Management

“VMware Cloud Infrastructure Suite is really more of a marketing term. Those of you know our products deeply know that they don’t fit this well together as they need to. Some of them have multiple databases, some don’t look the same, some install differently, and what I can’t stand that is Site Recovery Manager doesn’t currently work with vCloud Director. So, what we are basically able to say is that we created and acquired companies that led to a lot of individual products that don’t work well enough together yet.”

“It has become so clear that we’re on the wrong trajectory, especially when we come to virtualization and cloud. The outcome is what you see in the last video from VMware. Yes, they talk in the first part about integration challenges, but more importantly in the other part is the whole way they think about management, which is kind of what I call the ‘last decade’ way of thinking. If you look traditionally in IT management, we have a very, what I call, ‘bottom-up’ approach. You say, OK, let’s look at what we can collect, let’s collect it, and let’s try and figure out what to do with it. We never asked ourselves at the beginning, is what does operations really need to know about, and what can I really control in order to maintain the environment in a good state.”

“And as a result, what you see in the last 20 years is a barrage of hundreds of different little tools that are focusing on collecting data, and then planning and reporting and alerting to a user. And maybe we kind of got away with it in static IT prior to virtualization, but it’s definitely not an approach that can scale and drive the economies of scale that you’re looking at in cloud environments and major financials. And the reality is, when you are working in that kind of paradigm, first of all there is the question of what I collect, and I ended up collecting thousands of unnecessary metrics and I report on those metrics, create all kinds of alerts to the user. And then I’m building a whole system on top of that to basically suppress the amount of data to the user, because the user is overwhelmed with the amount of information.”

“VMware came into the planning environment, and instead of building a management system and integrating it to their environment, they went a journey of acquiring little tools like the big guys of management. And then they bought all kinds of assets from EMC, and now they have a mishmash of little tools that collect data and do some configuration on the other end, and they have no semantic integration of creating a coherent story of how to put all those tools together. The whole management question has to be approached differently. It’s not about little tools to do monitoring of something and configuration of something, and trying to put the two together. But rather the starting point has to be the questions, what I said at the beginning. What does operations need to know about? What can I control? Then create a layer of abstraction that creates common semantics across whatever you need to do. Once you have that proper abstraction, you have a common way to integrate whatever management functions that you need to solve what operations cares about.”

“Our starting point to begin with, was not what I can collect in the environment. But rather, what are the control points that I have in the environment that enable me to maintain the environment in a healthy state. And just by taking that approach, we say, if this is what my goal is, my goal is how do I maintain the environment in a healthy state, OK, what are the analytics that drive these control actions, and given the analytics, what kind of data that I may want to collect to drive the decisions of this analytic engine. So from the beginning we looked at the very tight, closed loop between what we are monitoring, analyzing and controlling in the environment. The result is we have a platform that has a common abstraction that looks at the virtualized cloud environment in very abstracted ways. On top of that we have an analytic engine that can drive decisions, based on that abstraction. And on top of that we have a very tight, closed loop on which we are developing delivery management functions starting with controlling the environment, maintaining it in a healthy state, and yes, if you need reports and alerts about the environment, we support producing that. So basically we created an integrated platform that delivers a full suite of functions that you may need, starting from planning, deploying and controlling the environment, and keeping it in an optimal operating zone.”