Supposing you were the sort of person who would consider shaking a stick at offerings of "educational" information on green IT, you might find, right now, that there is a bit more work than you can handle.

It seems like everybody's doing an informational webcast on green IT at the moment - and more often than not they're sponsored by one of the big technology companies with a vested interest in the decision-making of the people looking for information on "green IT," like HP, Sun or IBM.

The notice I got today from Internet.com was on an upcoming webcast called "Governing IT in a Green World." And it's sponsored by IBM. The upside of the sponsorship being the "FREE" price tag for attendance (to "qualified attendees.")

As far as I can tell, that "qualified attendees" bit refers to registered Internet.com users, a status that is not difficult to achieve. I signed up just a moment ago at the registration page that came up when I tried to follow the link to the webinar signup page.

The session is scheduled for December 2, 2008 and will be presented by George Spafford, principal consultant at Pepperweed Consulting.

According to the registration page, the key points of focus for the webinar will include

- Understanding the risks facing IT, such as skyrocketing energy costs, Global warming and the potential for new regulations

- Developing and implementing "green" strategies in IT, including how IT can help the organization and what approaches are relevant to IT

I get the feeling from the material I've seen that the presentation might include a rehashing of some of the more regularly-repeated "green IT" mantras. Particularly in the "understanding the risks facing IT, such as skyrocketing energy costs etc." portion of the show. But I doubt at this point that a webinar would be targeting the imaginary demographic of "IT people who have no understanding at all of the energy-related issues facing businesses today."

Besides, the presentation is "45-60 minutes" and it's very possible that there will be something worth seeing during the "developing and implementing 'green' strategies in IT" part.

IBM makes a lot of offers and announcements related to "green" data centers and IT. By now, we've all seen plenty.

But more often than not, they're worth looking into. In general, I've found that, as information issued by companies in the interest of self-promotion goes, most of the "green IT" info that has been issued in the last few months has been quite interesting.

Last week, as part of its sponsorship of ITWorldCanada's excellent "Green IT Playbook," IBM offered up a consultation on data center energy efficiency.

With a sort of cute little interface navigated by dragging a cartoon bee (because they're from the environment, I suppose?) and presented by a smooth-talking cartoon owl (from the environment and wise, I'd imagine) IBM invites visitors to submit their information by filling out a form promising "an IBM representative will be in touch."

Be forewarned: if you ask to be contacted, you're asking to "learn how IBM systems, software and services for a greener world can help." That is, don't be surprised when the IBM representative who gets in touch comes carrying a very IBM-centered sales pitch.

As I said earlier, however, I'm of the opinion that some of the most interesting information on green IT has come from vendors' efforts to promote their services.

It can't hurt to listen, right?

We received a very long press release this week from TrendPoint Systems, a company that produces a "turn-key data center energy management solution."

It's not a company we're especially familiar with at the WHIR, but with TrendPoint's fairly singular focus on data center energy efficiency and carbon monitoring, it's an organization that ought to be thoroughly on the radar of the hosting business.

Along with its TrendOne, EnerSure and EnviroCube products, which are generally hardware offerings designed to monitor environmental conditions and energy efficiency in enterprise data centers, the company appears to be actively publishing white papers on the general subject of data center efficiency, which makes it especially relevant to just about anyone operating a data center.

I know at the WHIR, we've got a pretty heavy appetite for published "green data center" information at the moment.

The TrendPoint data sheet can be downloaded in PDF form from the company's website, so I won't go into exacting detail about the plan itself.

In introducing the four-point plan, the press releases on several of the more important widely held truths regarding data center efficiency. Namely, that data centers are well on their way to becoming one of the world's largest consumers of energy (and subsequently producers of carbon emissions), and that as a result of that, we're not far from seeing some tough regulatory attention to carbon emissions worldwide.

The release includes a quote from TrendPoint CEO Bob Hunter, who offers an in-a-nutshell view of the current energy near-crisis facing data center operators.

"Data centers will soon be hit with a 'perfect storm' in terms of coal and natural gas driven utility cost increases coupled with the new carbon caps. These sites already have energy densities that are ten times greater than that of commercial office buildings, and their energy use is doubling every four years. The combination of rising energy usage coupled with significant electricity price increases and carbon caps creates a very troubling picture for data centers."

Before we get too far along, here, I'll warn you that there most likely isn't going to be a "eureka!" moment for any of you in reading the four-point plan. It's a pretty straightforward list of common-sense tactics responsible data center operators (especially in the service provider business) will most likely already have investigated, if they're not already implementing them to some significant degree.

Unsurprisingly, and understandably, the four-point plan is liberally populated with TrendPoint sales pitches. This ought not stun anyone who has ever read any white paper ever produced by any company.

Set an energy budget

Companies, says TrendPoint, should have an energy and carbon budget that can be broken down among users, departments and sites. "Colocation facilities, in particular, need to be able to provide each customer with the ability to manage their own energy and carbon usage and to provide a system to bill back customers appropriately."

Virtualize servers

The oft-repeated point - consolidating your physical resources and trimming away unused capacity is a data center energy saving tactic of the "do it right now" variety. TrendPoint's interesting add-on: "TrendPoint has seen that virtualized servers generate significantly more heat visa-vis the under-utilized machines and, therefore, need careful attention with their cooling management. Without proper cooling, virtualized servers, like all highly utilized systems, can develop 'server thermal inversions' within a data cabinet"

Equalize heat and cooling balance

TrendPoint says matching cooling resources to the needs of each individual cabinet can save 25 percent or more on their cooling energy use. And you can conserve more by balancing heat loads through grouping servers into zones within cabinets and working toward equalized heat loads.

Manage to the metrics

One of the more overt TrendPoint pitches in this section, but an interesting point - as data center equipment changes, managers need to continually monitor and manage heat and cooling. The company makes solutions that manage energy use according to The Green Grid's PUE standard.

In the press release, TrendPoint describes how a couple of its customers are seeing fast ROI from using the company's products.

The PDF is worth going through. It includes a lot of study-supplied supplemental information. And the great thing about energy efficiency is it has that potential to impact the bottom line of your data center operations. If the ideas (or even the products) discussed in the document have the potential to do that for your data center, you can definitely afford the 15-or-so minutes it will take to read.

One more note - there's a link at the end of the press release to a site with more information about some of the proposed energy regulations.

The Green Grid http://www.thegreengrid.org, the growing non-profit group aimed at advocating for energy efficiency in the data center held a conference call on Friday to discuss several announcements pertaining to the organization's technical forum taking place today and tomorrow at the Parc 55 hotel in San Francisco.

On the call were Green Grid directors John Tucillo and John Pflueger, the latter of which is also a member of the organization's technical committee.

Beginning today, (not coincidentally the one-year anniversary of the Green Grid's launch in 2007) the technical forum is titled "Get Connected: Assess Today for Efficiency Tomorrow," a sort of rallying cry to help people understand the organization's goal.

Coinciding with the technical forum are several announcements from the organization.

The Green Grid says it committed to the industry a year ago that it would continue developing deliverables that could be built into a framework that will let end users to assess their data centers in a way that will enable them to improve their efficiency. Today, the organization says it is presenting that framework, and demonstrating how it operates.

The framework itself is comprised of tools and technologies, as well as strategies and methodologies, developed through the collaboration of the Green Grid's participants.

According to Pfleuger, four specific "deliverables" are being introduced at the tech forum this week, representing the efforts of three of the organization's original working groups:

Addressing Organizational Barriers to Managing Efficiency

In many cases, says Pfleuger, organizations are set up in such a way that the paying of the energy bill and the decision-making that affects that energy bill are handled by entirely separate groups

"This paper takes a look at that situation and offers some alternatives and some guidance on how to possibly address those organizational barriers."

Baseline Efficiency Market Study

Produced by the data collection and analysis working group, which goes out into the community, and puts together the picture for the rest of the technical committee.

"The first deliverable out of this group is a fairly comprehensive study on existing and common practices in the industry today with regard to improving energy efficiency and managing for efficiency."

The data came from personal interviews with data center owners and operators, drilling down into what they're doing today and may be doing in the future.

Peer Review of Lawrence Berkley National Laboratory on High Voltage Direct Current in the Data Center

"Folks who have been following this space is very aware that one of the questions that has a lot of visibility in the community is one of power distribution in the data center. In 2006 Lawrence Berkley completed a study on direct current in the data center. And as a result of that study, a lot of people have been interested in this particular technology, and what opportunity it affords for improving efficiency in the data center."

The power task force has completed a fairly extensive peer review of the study, and will be presenting that report at the tech forum.

Five Ways to Save Server Power

The operations working group, concerned with providing guidance for things that happen after a data center is commissioned, and day-to-day data center operations.

"This deliverable is essentially a best-practices sheet for saving server power."

The deliverables discussed in the call, in addition to being presented at the event, will include extensive white papers. While they don't currently appear to be available at the organization's website, they'll presumably be found here.

It was a long call, and this is becoming quite a long post. But rest assured there was more to discuss. I'll follow up tomorrow with some more interesting stuff.

The Green Grid Issues New Guides We’ve discussed The Green Grid before. They’re the “consortium of information technology companies and professionals seeking to improve energy efficiency in data centers around the globe.”

There isn’t really a voice of authority just yet in the green IT issue, which tends to be a problem among companies touting their “green” services, or more importantly among the customers attempting to discern the value of one “green” claim in comparison to another.

The Green Grid may not quite have it’s eye on that voice-of-authority position, but it is certainly setting itself up at the center of some consensus-based thinking about the environmental considerations of hardware and data center building.

And while involvement in the organization is a paid-for privilege, the products of that thinking aren’t always kept under lock and key.

 

     

Last week, The Green Grid posted three new white papers on its site, available for free download to anyone who wants to look at them. Their rather lengthy and descriptive titles are as follows:

“The Green Grid Data Center Power Efficiency Metrics: PUE and DciE”

“Existing Metrics, Guidelines and Programs Affecting Data Center and IT Energy Efficiency”

“Qualitative Analysis of Power Distribution Configurations for Data Centers”

And they can all be downloaded from this page.

There isn’t a ton of freely available hard-numbers analysis or nuts-and-bolts instruction on measuring or improving data center efficiency out there. And that alone should make the white papers worth perusing.

Whether the standards and metrics that the green grid proposes here are destined for widespread recognition and adoption isn’t a sure thing. But metrics and practices such as these have a built-in value that exists independent of any official adoption or endorsement.

On Wednesday, Google was assigned the patent for something called the “Modular Data Center,” a distinction for which it applied in 2003, and which was given the final approval this week by the US Patent Office.

 

Specifically, the idea describes a data center built within an intermodal shipping container - the sort of container handled by multiple carriers and shipped in multiple modes.

 

The obvious application there being the sort of prototype mobile data centers that have been on the road the past few years, such as Sun’s Project Blackbox or, say, APC’s InfrastruXure Express.

 

The patent itself describes using rack or shelf mount computing systems, mounted and configured to be shipped and operated within the container (A lot of that is, from what I can tell, language taken directly from the patent, just barely paraphrased).

 

So it would seem that Google managed to patent the (very general sounding) idea behind some of the more interesting data center construction projects in recent years. However, the point here is not “what about Project Blackbox?” As many of the writers covering Sun’s Project Blackbox tour were quick to point out, the mobile data center is not so much a breakthrough as a refinement. HP and IBM have offered things like this for a while – to customers who were willing to pay.

 

The idea, according to this Register article, seems to be that Google is unlikely to set out immediately to impose its patent rights on companies like Sun, given its past complaints about the injustice of the patent system. But it’s not clear just what Google does intend to do, or whether the company has modular data centers either in production or operation.

 

According to The Register, Sun is aware of the patent issued Wednesday, but is holding off on making any sort of comment about it until the company is better acquainted with just how it stands to be impacted.

This week, the US Environmental Protection Agency released an alarming, possibly alarmist, report on power consumption by data centers. A very thorough retelling of the material found in the report is available in this Computerworld article.

 

Here is a somewhat more concise summary:

 

- Data centers and server equipment currently consume approximately 61 billion kilowatt hours in the US - approximately 1.5 percent of the nation’s power usage.

 

- That amount doubled from 2000 to 2006, and is expected to double again by 2011 at current rates of growth.

 

- At that rate, the US would need to build 10 new electric power plants during that period in order to meet the energy demand.

 

The environmental impact of IT technology is a topic that comes up quite frequently here at the WHIR. And it’s something I’ve personally pondered quite a bit. While it’s well out of my realm to offer any insight into the accuracy of the EPA’s numbers, I think it’s a fairly reasonable assumption to assume that the largest growth took place during the last two years, as technology phenomena like BitTorrent and YouTube made the distribution of video material commonplace, and led to the consumption of considerably more computing and network resources.

 

It’s probably not outrageous to imagine that those technologies aren’t the richest or most resource-hungry applications that we’ll see in the immediate future. I couldn’t really speculate as to whether the prediction for the future is conservative. But I can say with a fair degree of certainty that it is relevant.

 

Part of the impetus for the EPA report seemed to be the proposing of solutions to the escalating energy consumption situation. The US could escape a future energy crisis, says the EPA, if data center operators and IT managers consider this a reason, and an opportunity, to reinvent their architectures around energy efficient technologies.

 

And the EPA has it’s own agenda (protecting the environment) to promote. According to the Computerworld article:

 

“There are efforts under way by industry groups and the EPA to develop a metric to allow IT managers to compare power use against workloads for these lower-end servers. The EPA is also interested in seeing development of a standard that would allow IT managers to measure the total power consumption against the power consumed inside a data center, giving data center managers a benchmark to use.”

 

The EPA report has seen quite a bit of coverage up to this point, and the beginnings of what I can only assume will be a somewhat widespread reaction.

 

The Green Grid, an environmental .org-type organization becoming fairly well known to the Web hosting faithful, apparently held a webcast Tuesday to outline its own intentions as far as making data centers more energy efficient.

 

I didn’t see the webcast, nor can I find a place to view it after the fact. But I am interested in the Green Grid’s progress. It will be interesting to see if its members produce a truly unified vision, or if there is a potential for self-interest to get in the way of progress.

 

UPDATE: The PR folks at the Green Grid sent me a link to an archived version of the webcast. If you'd like to watch it, you can do so here.

As you may already be aware, Hostway suffered a series of hardware failures that turned a planned Monday migration of servers into a serious downtime issue this week. A large number of dedicated server customers acquired in the company’s purchase of Affinity Internet were offline well in excess of the planned 12 to 15 hours – some of them for days, some still offline as of right now.

 

Hostway’s vice president of marketing and business development, John Enright, understandably a busy man at the moment, took a few minutes to talk to me today and offered some insight into the situation.

 

Enright says the move brought 4,000 servers – representing about 3,000 individual customers – from Miami to Tampa.

 

“Our goal in moving the customers from one facility to another was to improve the quality of service in the long run due to the fact that our Tampa facility has better connectivity and more capacity.”

 

Of course, the irony of a protracted outage in that case was not lost on Enright, who seemed genuinely contrite and took pains to impress upon me that Hostway’s resources, including its data center and executive staff, are all focused on getting servers back online at the moment.

 

“The majority of customers and the majority of servers were actually back online within that time frame. However, for reasons that we have yet to determine, there were an abnormally high amount of hardware failures among the servers that we migrated. I don’t have a definitive number, but it was approximately 500, give or take 100 servers, that actually experienced the hardware failure.”

 

He says the company anticipated some hardware failure, a predictable result of moving any server, or just turning it off and back on. Hostway had spare parts waiting at the Tampa facility. But the number of hardware failures exceeded the company’s most pessimistic estimates.

 

“We’ve been working around the clock since then to get all of those customers back up and running. We have additional staff in place in our Tampa data center, and in our call center to respond to customer inquiries. At this point we’re down to about 50 servers that are still affected by a hardware outage of some form or another. We’re diligently and feverishly working through those to bring every customer back online.”

 

Enright says Hostway is compensating affected customers on an individual basis with credits, and reviewing its next steps. But the company is understandably focused on getting things back online at the moment.

 

“The priority here is to make sure every customer is back up online and running. That’s 100 percent of our focus. And as soon as we get to that state where every customer is back up, we’re going to review what our next steps are.”

 

Hostway is obviously hoping that state arrives sooner than later, but Enright couldn’t offer a firm out-of-the-woods ETA.

 

“As we get down to the last few customers, the hardware problems are getting a little bit more complex. In some cases it’s more than one component that has failed. At the rate that we’re going, we do expect everybody to be online shortly.”

 

Customers have complained of being unable to contact the company when the outage first occurred, which is usually the case with a major outage. It is unlikely that Hostway has the capacity to handle 500 customer calls at once. But Enright says the company is reaching out to each customer as it is able, and providing them with the status of their servers.

 

“Right after the migration took place, communication was a challenge. And we did have an enormous amount of phone calls coming into the call center. We added additional staff to both the call center and the data center immediately. And since then, service levels have improved. They’re still nowhere near what we would consider acceptable in an ordinary case, but customers are able to reach us of an average wait time of 20 minutes, give or take.

 

“We acknowledge that communication was a challenge in the first part of the outage, but we’re doing everything we can to rectify that by reaching out to each and every customer.”

Saw this story from yesterday, referring to the contents of the latest IDC report on virtualization and multicore servers.

According to the article (I haven't seen the IDC report itself), the new material contradicts certain earlier reports that had been enthusiastic in forecasting sales and shipments of x86 server equipment.

More interestingly, the new report, "Virtualization and Multicore Innovations Disrupt the Worldwide Server Market," discusses the impact that that virtualization technology is likely to have on the market for X86 servers. The report says multicore and virtualization will cost the x86 market more than 4.5 million shipments and $2.4 billion in customer spending between 2006 and 2010.

IDC's report goes on to say that server virtualization will lead to increases in other areas of the server business, as IT managers relying on virtualization look toward more powerful equipment that can be more reliably divided up in terms of processing power, storage and other features.

Analysts, these days, seem to generally be on the lookout for something optimistic (or glaringly obvious) to say. And it can sometimes be difficult to parse it all, particularly with a contrasting opinion available in many cases (the story in the link includes contrary remarks from a Forrester Research analyst).

Common sense might reasonably lead somebody to believe that virtualization technology, which can enable a single server, in a wide variety of settings, to perform the work of several servers, might have an impact on the overall sales numbers in terms of server hardware. And that impact would, logically, be in the direction of fewer physical servers needed, fewer servers required.

But then, I don't pretend that my basic analyzing capabilities make me some sort of analyst.

One similar cause-and-effect outcome I'll be interested in seeing results from is the impact that some of the new emerging grid computing technology (particularly in the more Web hosting specific environments) might have on server sales.

We've discussed in the past how older servers - equipment that might not be up to the typical dedicated-server task - can be useful in constructing grids for delivering utility-based computing. Prolonging the usefulness of older machines will have an obvious impact on the server life cycle, which will no doubt have an impact on server sales.

As grid computing becomes more widespread in the hosting business, it'll be interesting to see how that impacts server sales, and how hardware makers adapt.

TAGS: virtualization, server

I announced last week that I was heading off to attend 3tera's Grid University. And I'll admit, I slacked off a bit. But I spent some time reviewing some of the introductory courses (archives of past sessions can be viewed online).

The courses are an unabashed advertisement for 3tera's AppLogic application platform. No surprise there. The whole project is intended to get people like you and I interested in 3tera's products, and so far, I'd say it's doing a decent job.

Class: Introduction to Grid and Utility Computing

This one's about as basic as you'd expect. It's sort of the long-form AppLogic sales pitch. A nine-minute explanation of the model 3tera uses to enable service providers to work with software as a service and Web 2.0 companies by providing them with use of a scalable, shared grid of resources.

I've explained the basic AppLogic outline in the past, and I'm probably better off not doing it again here. So I'll skip right to more of the informational sessions.

Class: AppLogic Applications

The demonstration discusses how to design, provision and back up grid applications ranging from simple control panels such as cPanel (specifically cPanel, in the case of this demonstration) to complex e-commerce implementations.

During the demonstration, the instructor builds a two-tier SugarCRM Web application (out of simple, AppLogic-packaged "catalog parts") with an input gateway, a load balanced Web tier, a data tier with databases and virtual NAS storage devices.

The process of building a simple (or complex, for that matter) applogic architecture is very similar to the process of creating a diagram of what one might build in the physical data center. Aesthetically, that's exactly the process.

An interesting security note - the only parts of an applogic architecture that has a publicly accessible IP is the user-facing gateway. Everything else is invisible (on account of its not existing in the physical world), and very difficult to attack as a result.

At about midway of this particular presentation, the instruction veers off into things like using command-line controls to assign resources to application components. Admittedly, things are starting to get a little bit over my head at this point, as my specialty is not the building and administering of application architectures.

So it's at about this point that I began to reconsider whether I wanted to remain in attendance here at Grid U. It didn't take long to figure out that I'm not going to stick around. But to be perfectly honest, the parties were lame.

Just don't tell my folks.

I think the speed with which classes did me in actually speak to the value of the content - that is, it's not a lot of PR or marketing talk. The sessions dive right into the technical details of operating the platform, which is probably what you're looking for, and exactly what would be lost in the translation if I were to repeat it here.

I don't think there's anything I can synopsize or communicate any better the sessions themselves can, except to assure you that, depending on your relationship to the architecture behind Web applications, AppLogic is pretty fascinating, either as a possible addition to your operations, or as a curiosity.

TAGS: 3tera, grid university