During the evolution of Data Center Pulse we had participation from many different business sectors and companies which influenced concepts and drove new ideas. Here is a Stack version that was the last officially published.
The Stack version 2.2 which looked at market influences on data centers
See the original video that kicked off The Stack concept for Data Centers.
In this episode, Dean Nelson talks with Jeremy Rodriguez, the Co-Chair of the Data Center Pulse Technical Advisory Board (TAB), about the development of the Stack Framework. The Stack is an output of the DCP Summit in February, 2009. This video runs about 9 minutes.
Transcript from this video:
Dean: Hello and welcome to another edition of Data Center Pulse. Today we're going to be talking about the Stack, or the framework that we believe needs to be developed to address the entire data center challenge. Joining me is Jeremy Rodriguez. Jeremy is one of the founding members of Data Center Pulse. He's been here from the beginning.
Jeremy: Absolutely.
Dean: Welcome to the show.
Jeremy: Thanks Dean.
Dean: Do you want to tell us the two roles that you're filling right now?
Jeremy: Sure. I'm filling the role on the technical advisory board, running the development of the Stack framework for Data Center Pulse, so -
Dean: So you're the coach here of the actual advisory board, but then you're actually responsible for the development of the Stack, which is one of the things that would go through the board.
Jeremy: Correct.
Dean: Okay. Great, and so we're going to be talking about the different components and why this is even important to us. I would say that in February of 2009 we had the summit and Jeremy, with another member, was actually running the cloud piece.
Jeremy: The cloud summit piece and that's where the Stack all started. During the cloud technology track we realized that we needed a common frame work to understand data centers, so we can talk about them as peers and equals.
Dean: The Stack was actually brought in by one of the major financial companies that participates in Data Center Pulse.
Jeremy: That's right.
Dean: When they brought it up we all kind of had that epiphany at the same time. It was -
Jeremy: We went a-ha, and that makes a lot of sense. This describes a lot of things that we as individuals in our organizations have a challenge with.
Dean: Because all of use were approaching all these different components, so whether it was the power or whether it was the networking or whether it was the storage, whether it was the power in cooling, all of them tie together at some point. The Stack really made it clear that their associations, connections, that make sense, so we need to develop that whole picture, that framework so we can address all those different layers.
Jeremy: The first problem we think about when we look at the Stack and what we're trying to do with it, is where is it and what feeds it? Those are two things that are extremely important to the Data Center, whether you're into Data Center site selection or you just operate in the Data Center, they change your world, those two questions.
Dean: Just to kind of give you the basics, we're looking at a score in, then what you do with it inside, and a score out. Very simple. The carbon score in is one that's common across everybody. You've got power from somewhere. Whether you generate it yourself, and most people are actually getting it from a utility, you should be able to know how clean that power is. Then the decisions you make inside of that data center are the other building blocks. Right?
Jeremy: How is that data center designed and what does it do? Then you get your score out -
Dean: Two big questions.
Jeremy: Big questions, and you see these questions being asked by many other groups outside in the industry, your peers, your data center organizations in different ways. We really don't have a common way to do that yet and that's what the frame work is trying to establish.
Dean: Yeah, so if you look at it, the way you do mechanical, electrical and plumbing, how the data center is designed, right. How it's designed will tell you how efficient you're going to run physically. The metric for that one, PUE, DCIE, whatever you want, the key being that how it's put together and how much power goes to your actual systems.
Jeremy: We have your real estate metric, the PUE, you can have your server performance metric. Is the power going in actually producing a lot of useful work, to use a green grid term. Or is it doing everything you need it to do? The cooling systems, how efficient they are?
Dean: We're not trying to reinvent the wheel here if you think about it. You just said the green grid, you look at the DCIE and PUE things, that was developed by other people. What we're saying is the metric has to be there. The problem is people are coming up with hundreds of metrics. What we want to do is be able to say this block will have this metric, so PUE seems to be pretty good for the physical environment, but it's the above ones that are the most difficult. That's why the green grid has the eight measurements of useful work.
What we want is that people are going to pick the metric that makes sense for their business. Okay, so each block will have metrics associated with it to be able to have a score out, right?
Jeremy: That's absolutely correct, and when you look at your high performance data center versus a normal business data center, you'd have your baseline in and your score out, and that would be your ability to compare those two data centers together.
Dean: Right, so the power in then that score out really is the culmination of all those other metrics in the different blocks. They will roll up into one top level metric. We do need a top level metric.
Jeremy: About your data center -
Dean: I'm sick and tired over everybody saying, "Well I have a tier one and a tier four, and whatever else," and I can't compare. This is not apples to apples. The bottom line is, you're going to do a certain work inside of that room. You're going to use a certain amount of power, so if I've got a tier four data center I've made a decision for my company that is more important to put physical infrastructure together to make sure I don't have any failures. That doesn't mean you should get points for it. That means that the other side of it isn't as resilient. The applications, et cetera.
If the applications are more resilient you can build lower tier data centers. The scoring will be worked out. The key is that we're going to measure it and it's going to be consistent across all those different types of work and companies and functions.
Jeremy: What you just mentioned is your application availability is taking place in your physical infrastructure availability.
Dean: Exactly.
Jeremy: That's a struggle right now in the industry at large where your line [inaudible 00:05:17] on software so your things that were in the past done by physical hardware.
Dean: If you think about it, the holy grail here really is that we can all have a common method of actually approaching our data center, the components in it, actually addressing them and then measuring them and reporting out for it. Everybody's trying to get to that, "Well how efficient is my data center."
The framework, this Stack, the Data Center Stack, is enabling us to actually talk the same language.
Jeremy: Right.
Dean: Ultimately what we want is the governments around the world to be able to tap in and see what the data centers have, how efficient they are and set goals against what efficiencies we should be all meeting. We want to get ahead of that wave and actually help them define how to measure. Right. How to address the data center.
Jeremy: We want to help them define that as a group of end users within Data Center Pulse, very important to say, the framework is designed by end users, we talked about it in the Summit, to bring it back to how it was created. As end users we want to use this framework to talk to the government and say, "I want to be judged on this framework because I understand it as a company, as an individual."
Dean: I've actually accepted it as something I want to implement, and so -
Jeremy: The reverse being that here is a set of criteria we're going to measure you on, and as an end user you might say, "You know what these don't apply to me."
Dean: Right, so we've already got a number of companies that have signed up to endorse the development of the Stack itself. You can go to the website that is listed here and see those companies, and if you're interested in this of course, our end users, we want you to jump in, work with Jeremy on trying to come up with the refinement around the actual Stack itself.
Jeremy: Yeah, we're looking for critical feedback. We want to hear your opinion. We want you to help us develop the Stack. It's important to get involved. Important to contribute. You can be a part of it and when the government comes and asks you how your data center's doing you can answer the question rather than run around and try to figure out something new every time.
Dean: It's healthy competition that comes out of this as well. I really want my peers to go back and say, "My data center's more efficient than yours." Because that will drive me to go make mine more efficient than theirs.
Jeremy: It will cause you to ask the question, "How is yours more efficient than when you talked to me about it."
Dean: The other part of it is that my execs up the chair will be able to see a score. How well do I rate against the industry in my data center implementations. Because they can look at the costs all day, but then they can say, "Well that's great, how efficient are we?" Well we're efficient.
Okay now we can rank and say how efficient we are against our peers and against our goals that the government will actually be putting out I'm sure. What we're looking at now is again, our end users please join into the discussions on LinkedIn. Jump on the datacenterpulse.org. Look at the actual threads in there. Give feedback into Jeremy
Industry folks, we'll be putting this back out. Watch the website for the dates of when it's going to be released. The paper itself that you can give input into there, but our ultimate goal in this effort is what?
Jeremy: Is to get a framework that the end user community agrees upon that's been reviewed by the industry, release out for public use. Everybody's checked the boxes, that I agree this is the one we need to do by the beginning of next year.
Dean: January 1, 2010, we would like our end user base to start adopting and implementing this model. Okay, great, well thank you very much Jeremy for joining the show. Follow this link. You can see the actual Stack development there. All the information Jeremy will be posting up with his team. We're also looking for people to jump into sub teams.
Jeremy: Absolutely. If you're an area expert or you want to contribute let us know. We'll find a way for you to contribute and help.
Dean: Follow this, send an email to this alias below, and you'll get right to the people that actually can help you join it. All right. Great. Thank you very much Jeremy for joining me on the show.
Jeremy: Thanks Dean.
Dean: Appreciate it, and I'm sure you got a lot of work ahead of you.
Jeremy: Absolutely.
Dean: Got to get rid of that day job so you can keep doing this. Thank you again for watching another edition of the Data Center Pulse and stay tune for more episodes.
Original Stack (draft proposal in 2009)