Listen in as Jon Inaba, Director of Power Management Solutions at Raritan, discusses data center power utilization, power efficiency, as well as space, power, and cooling optimization.
Click Here to learn more about Raritan's data center power management solutions.
Data center power usage has been all over the headlines lately. Between February 15th and February 28th, there were over 15 headlines in the news about it. Jon will get deeper into the issue and discuss the real challenges data centers have and the innovative solutions they're finding to meet them.
Welcome back to Raritan Radio everyone! Today our guest is Jon Inaba, Director of Power Management Solutions at Raritan. Welcome Jon, how are you?
So, your role Jon at Raritan is to promote and support Raritan's enterprise power management solutions. These enable customers to maximize their IT power utilization, power efficiency to optimize their space, power and cooling resources. Naturally, I looked around at good deal on Google news and I just got very curious about how many articles there were that discussed power consumption in data center. There has been over 30 articles since February 15th discussing power consumption. Obviously, this kind of screwed me and has made this a very hot topic and you're right in the forefront of it to offering this kind of solutions, you must be running into this a lot, right?
Yeah, really this discussion is still ongoing for in a material level really for about the last five years. Five years ago, people really understood that there was a strong connection between very poor utilization for design and for practices and rapid power consumption.
So, how long do you think it will be before data centers are so energy efficient, that the news media has no story to publish with regards to that, because that is the goal obviously.
Oh, I think a lot of the large consumers and the very sophisticated end-users are really making some really good choices. Unfortunately, there is always the risk mitigation and the design or over design or under utilization which are really leveraged against the fear factor.
Can you explain that fear factor for me?
Sure, I -- you know as far as I'm -- basically it's sort of the old baggage you know you never get fired for buying IBM or you never get in trouble for buying products __02:25__at your doing why. And what happens is that people over built the under utilized based on the assumption and generates an erroneous assumption that they're providing better up time and better service. And so, it's typically where you see 40% utilization of a circuit reading down to 25% utilization of the circuit because that they feel that, that's providing them less risk than actually utilizing the circuit completely and having the appropriate redundancy in place.
And as part of that they're expecting or they need to -- they feel they need to prepare for an on-slot of traffic or demand, peeked them in, are they over voting for that or they just simply -- they feel better having more toys in the toy box.
It's more the latter than the former. Its -- mostly in the utilization thing and that's -- that's been more of a big trend than a later trend.
Okay. So I understand that many data centers are focused on metering these days. Is outlet metering really necessarily and how it's being applied in data center operations?
Yeah, absolutely. Outlet metering at this point is really -- I believe essential. Because at the end of the day, what you are dealing with in providing power to a server that is turning a -- basically it's just delivering a computational cycle. And so the ability to equate a -- whether to watt or part of the watt to a computational cycle is critical. And so it's at a device level that you are making decisions whether it's a utilization of that specific device, the resiliency of that specific device or the health of that specific device. It is really down to the device level that you are now looking at the systems and looking at the architecture to make sure that you have not only the redundancy, but of the service delivery that you need.
So what is top of mind in their decisions about metering choices? How do they go about this? You and I have talked about the awareness and the importance of meterings out there. Where is it most relevant, is it in new data centers, is it in retrofit situations? How does this two...
How did --
Almost across the board, almost across the board. What happens is that people are faced with the reality that they simply have to make more economical decisions about power consumption because they are out of capacity. Whether that capacity is, as the result of the utility not being be able to deliver them more power or that they simply don't have the critical infrastructure to support more power. In a new build were they're extending, obviously they wanna make the most out of the real estate and get the most things for their butt out of every square foot. And so, it really comes down to anybody who is operating a data center, is going to look at efficiency and look at ways of optimizing the powers then.
Now, from a new data center obviously I can build with the most latest and greatest technologies and my -- I don't run in to so many roadblocks. Can you talk a minute about some of the roadblocks that retrofit situations run in to?
It runs organic. Literally, it goes from where there in the middle of -- perhaps a UPS unit that would undersize begin with or that it's -- it is just simply not utilized because of support configuration. And so really from a retrofit perspective, the obstacles are almost universal and most of the time, it's more of an issue of not wanting to change out of current practices. It's the things that people are familiar with and are comfortable with. And so often times you will see a conflict between two business groups, one group does things on their way, the other one does another way and its not necessarily that either way is wrong but they're not, they're not speaking the same language and they don't, generally they don't have a common set of goals.
Right, right. And it's probably some of the "if it ain't broke don't fix it" mentality playing in their as well, right?
Yes, that's huge with data centers, right. If I make an adjustment and it results in an outage, that means I'm losing my job.
That does not went well. So, and you know -- I'm sure the listeners know at some point we are gonna segue way to Raritan's solution for this and I'm doing a bid early here, but obviously you guys have some solutions that make us an easier decision for a data center manager to get into metering and get into better ways of power management and monitoring, correct?
Correct. And what we've done is we release our product back in late 2007. And so we have a deep track record of being really understanding the environment, understanding what data points are important that accuracy and granular, your critical, but also access to that data in a structured and extensible format is very important. And so, the struggle is really that you're looking at a larger amount of new data and how do you use that data so that you can all start making intelligent choices and intelligent changes.
I followed data center journal pretty closely and recently, they had an article posting to remind us of the basics in terms of power management. Things like regular equipment updates, chimneys over the racks, infrastructure monitoring for things like hotspots. And that last one, you know so many people are talking about hotspots. Can we talk about that more and focus on, is that the right focus? Are they focusing on the right things if they are worried about, things like hotspots?
You know, it's right because I saw that article and certainly that point gets up in almost all conversation to have a way prevent hotspots. It's funny because well one, hotspots are generally very episodic, there are generally minor in issues. If you have a major crack outage, there is no secret everyone is getting alarmed from the facility's IT. And so, generally what happens is a hotspot today are very precise points or area of your data center because it some sort of the architectural change or power consumption sort of huge, right? Right, but that's generally rare, what's happening in some a -- really from the extensionally perspective, what's happening far more often and which has far more critical issue. And it is the fact that they're over consuming power overcooling the entire data center, right? That's cold spots that we should be thinking about. If we mitigate the cold spots, and you are properly distributing air, you are running a data center environment.
Okay. And so what we're talking about it's really -- go ahead, I'm sorry.
What you are talking about -- risk prevention is important. But being able to identify and optimize a data center and fine tune it, that's what you're controlling for. And in the positive externality of doing that, help you very quickly detect the hotspot and mitigate the exact root cause of that hotspot as oppose you walking in the data center swinging doors open, that's bad. That's generally the possible next problem.
So to get in, and we've discussed this in the other shows with other folks at Raritan. To get this kind of thing that you are talking about, which I kind of hear is more of a holistic view of the data center. But what you really probably might need if you are doing this is more granularity of instrumentation at the rack is that right? And throughout the data center.
Yes. That's a really, really good direction that most people are going in. They understand that it's now a necessity, it's no longer an option to be able to say "well, I'm metering at the circuit, that's good enough". Because problems are happening at the rack level, right? It's there happening within individual items of the rack that are basically contributing to the potential what's capturing to others.
Yeah, how does Raritan's consultative approach play into this? When you, like you're on your way to a meeting today, we got you on your cell phone on the way to New York. How does your consultative approach play into this? How do you approach the two different kinds of companies for instance, a new data center or retrofit?
Really what -- what we really trying to do is understand their goals for energy efficiency and optimization and understand the groups that are involved. And then help bridge the gap in communication or provide the scaffolding of effective communication between those two groups.
Okay so you are on both sides, you are facilitating the communication in the upfront stages of the process and as the decisions are being made, right?
Correct, that's what we really trying to help that conversation along so that we can refine really what the needs are and the critical patch in accomplishing those goals.
Okay. And then you've worked your way into a relationship where you're a trusted advisor and you're there for them as they go through the process of the build add or the change and helping them with the exchange management things like that, right?
Correct. We work really closely with our end-users and really view it as a collaboration. Because our end-users become a lot of the inputs on future product development, right? They are the ones that are saying, "this works well, it would work great if you did this". And we incorporated a lot of those, a lot of those data points into our future development.
Well that's awesome. And what else would you do tell me about what you're hearing out there, the concerns right now in the data center with regards to power management.
The concerns right now are costs, because you've just entered a position where -- previously "dam power strip" was a $300 to $400. And now when you really get into the highly sophisticated, highly-metered products, it is actually more expensive. And so, there's a budget awareness issue that pops up and they say, "Okay, how do we cost-justify this new investments", "how do we build it in to other groups", and most importantly really is "how do we guarantee to worth delivering their value for this investment in a very short period of time".
You guys have the obviously the background and experience working with some huge companies to help them do that. I mean there's a case study I guess what you call the case study on a website, raritan.com, for eBay in which I know you guys has had some huge success helping them cut power consumption in data centers or manage it more effectively.
Sorry Jon. So Jon, we've been very efficient I think with how much time we're using on the show it's not gonna take too much data to download for people, worth 15 minutes now. You have any wrap up thoughts that you wanna share with the listening group.
No, I think we're good. If you have any other questions, feel free to give me a call.
Alright, they can track you down Jon by going to raritan.com and watch you in there. On the left side, to check out products and they can find a way to power management and learn great deal more about the solutions Raritan offers. Again, that's raritan.com. Jon Inaba thanks for your time, and tune in next time folks another episode of Raritan Radio. Take care.
Thank you so much.
Sorry we couldn't complete your registration. Please try again.
Please enter your email to finish creating your account.
Receive a personalized list of podcasts based on your preferences.