Sustainable IT | Ted Samson » TAG: Green benchmarks

April 22, 2008

Study: Lack of standards slowing green datacenter initiatives

TAGS: Digital Realty Trust, Green benchmarks, Green standards

Companies seeking to inject sustainable practices into their datacenter operations are hankering for clear standards to guide the way, according to results of a study conducted by Digital Realty Trust, a developer and manager of technology-related real estate. The good news is, plenty of organizations, from private companies to non-profits to the Feds, are working to develop standards.

Among the respondents to the survey, comprising IT decision-makers at large North American corporations, 82 percent lamented that there is no clear industry standard for green datacenters. The figure is up from 75 percent compared to results from a similar survey Digital Realty conducted last year.

Further, respondents indicated a clear idea as to what they want in a green datacenter standard: 94 percent said it "should outline how to achieve efficient power usage (i.e. maximizing energy delivered to IT equipment by the facility)."

Another 83 percent said that the standard also should outline "how to enhance HVAC systems to use energy more efficiently."

There certainly are a number of efforts under way to develop standards to help companies make their datacenters greener, which means less waste, a small carbon footprint, and often numerous other business benefits.

Among them, the EPA is working to develop an Energy Star rating for datacenter infrastructures and is in the process of collecting data from organizations to help with the effort. Specifically, the EPA is asking companies to gather and share a year's worth of stats on energy use and the operating characteristics of their datacenters, stand-alone facilities as well as those in offices and other types of buildings.

Jim Smith, vice president of engineering at Digital Realty Trust, also pointed to the efforts of The Green Grid. "We ...support industry-wide green datacenter initiatives by continuing to be an active member of The Green Grid, which is doing excellent work establishing standards and best practices for datacenter energy efficiency," he said.

Moreover, the company cited its own efforts in sharing information in the efforts: "We are ... committed to support industry-wide initiatives to increase datacenter efficiency," Smith said. "One of the key ways we are doing this is by sharing energy-efficiency data ... in response to customers and other end-user organizations who want data and benchmarks that educate them about the energy efficiency of competing facilities and about how their datacenter can support their corporate green strategy."

Posted by Ted Samson on April 22, 2008 02:25 PM

January 24, 2008

Dell's shiny green blade server

TAGS: Blades, Dell, Energy conservation, Green benchmarks, Green technology, HP, Hardware, IBM, Open Source Systems (OSS), Power consumption, Power supplies, Sustainable IT

The PowerEdge M-series from Dell has it all, from a hyper-efficient power supply to superior cooling and power management
From a green-tech perspective, the most intriguing server I've ever known (from a distance) was the Gemini Green Series from Open Source Solutions (OSS). Highly efficient power supplies? Check. Efficient internal cooling? Check. Easily swappable components -- from motherboards to memory to power supplies -- to eliminate the need for ripping and replacing upgrades? Check.

Alas, OSS is no more, but its legacy won't be forgotten -- not by me, anyway. The fact that it's gone the way of rainbow suspenders is no reason to dismiss the beauty of its product design either. In fact, I see glimmers of the Gemini in Dell's newly announced PowerEdge M-Series chassis and blades. Dell designed the M-Series "from the ground up using Dell Energy Smart technologies, resulting in 30 industry patents," clearly with sustainability in mind.

Playing it cool
Like the OSS Gemini (which was a 2U server, by the way), Dell's new PowerEdge blades employs 90-plus percent efficient power supplies, developed in-house. (For reasons I still can't fathom, the standard among server vendors still appears to be in the realm of 80 percent.)

Not only is Dell's power supply capable of hitting a higher energy-efficiency level; it does so more quickly than rival power supplies, according to Mike Roberts, senior product planning manager for the M-Series line. He says that most power supplies achieve their maximum level of energy efficiency only when the supply is running at 90 to 100 percent utilization -- which certainly isn't the norm. "We get to a really good efficiency level at relatively low threshold, 88 percent at only 20 percent utilization," he says.

Those seemingly small differences among power supplies can actually make a dramatic difference in terms of reducing a server's power and heat waste -- and the associated costs scale impressively the larger your datacenter.

Dell also claims to have developed an innovative internal cooling system, designed to adapt to the needs of both high-end and low-end configurations. In addition to its optimized fans (as well as their underlying algorithms), the system boasts a superior airflow design, according to Roberts; Dell has taken great pains to remove impedance throughout the chassis. "The easier the air flows, the less hard the fans have to work," says Roberts.

The chassis also has three distinct cooling zones, each cooled by its own fan bank. In lower-end configurations where the chassis isn't fully loaded with blades, "the fans on the side can run really slowly because they don't have to work hard to cool their zone," says Roberts.

There's also the swappability factor (a word you will not find in Webster's, by the way). In an ideal green-tech world (mine anyway), an IT admin would be able to swap in and out all major components in his or her servers, while they're on the rack, such that machines wouldn't need to head to the shop or, more likely, to the recycling bin if a substantial upgrade is needed. It doesn't look like the major hardware vendors of the world are ready to offer that level of hardware interoperability, but here, Dell has taken a step in that direction.

Specifically, the company has developed what it dubs FlexIO switch technology for easily upgrading the machine's network connectivity up to 10Gig without replacing the base switch. For the enclosure, customers can opt for an upgradeable Dell PowerConnect M6220 Layer 2/3 Ethernet blade switch, with 1Gb ports and optional bays that can support either 10GbE or stacking ports.

Further, customers have three Cisco Ethernet switch choices, including a switch with a variety of 1Gbps, 10 Gbps, and stackable ports. Also available: a Cisco Infiniband switch. Add to that the options for two Brocade 4Gbps Fibre Channels, as well as Fibre Channel and Ethernet Pass Through options.

Additionally, with an eye on the future, Dell has designed the M1000e enclosure not to be dependent on specific server processor/chip set architecture. Further, it will be able to accommodate double-wide blades down the road.

Secret software sauce
Dell's hardware choices alone don't result in a more energy-efficient machine. Like an increasing number of vendors, Dell is turning to server management software to rein in energy waste -- a promising development that stands to lower those power bills and extend the life of hardware.

In Dell's case, that software takes the form of its Version 5.3 of its OpenManage systems management suite, released last November. Dell touts the package -- which comes at no cost with the blade package -- as "easy-to-use yet powerful management tools that help reduce the cost and complexity of managing computing resources."

Among its features is dynamic power management, which enables admins to set high- and low-power thresholds to help ensure blades operate within their defined power envelope. The norm is for servers to consume the maximum recommended amount all the time, even if they're not usually being run at full bore. Moreover, the package offers real-time reporting for enclosure and blade power consumption, and the ability to prioritize blade slots for power to provide optimal control over power resources.

So, for example, if an admin were to allot 3,000 watts to a given chassis, the system would distribute power evenly among all the blades. However, if the chassis wasn't pulling enough power, for whatever reason, it could be set to prioritize which blades would be throttled down first.

Your benchmark or mine?
As Dell tells it, the various greenovations it's injected in these babies reap superior power efficiency (that is, performance per watt) over rival blade offerings from HP and IBM. According to a Dell-sponsored study by Principled Technologies, "the PowerEdge M-Series consumes up to 19 percent less power and achieves up to 25 percent better performance per watt than the HP BladeSystem c-Class. Compared to the IBM BladeCenter H, the M-Series consumes 12 percent less energy and achieves up to 28 percent better performance per watt."

Were you to apply a cost-per-kilowatt amount to those figures, as Dell has, you'd save $2,600 annually per year over HP's blade competitor and $1,500 per rack per year over IBM's.

I wouldn't discount Principled Technologies figures outright, just because the study was sponsored by Dell. It's entirely conceivable that, given the power-efficient measures Dell has taken, its blades consistently delivered better per-watt performance -- but only in that test, which used the SPECjbb2005 benchmark.

Problem is, that benchmark wasn't really developed to measure power efficiency. Organizations and analysts are still grappling as to what method is best for accomplishing that, as evidenced by a recent study comparing the power efficiency of AMD and Intel's respective quad-core chips.

The point is, benchmarks are slowly emerging, such as SPEC's recently announced SPECpower_ssj2008. However, that benchmark isn't suited for a blade environment. In short, work here remains to be done.

Hopefully the InfoWorld Test Center will have a chance to test Dell's claims against offerings from HP and IBM. Whatever the outcome of that test might be (again, if it happens), I still extend kudos to Dell for further raising the green bar more among hardware vendors.

One last thought: IBM and HP may claim that they're not worried about Dell's latest foray into the world of blades. In fact, a rep from one of those vendors sent me a rather dismissive note prior to the official Dell announcement: "We've been hearing rumors over here that Dell finally plans to announce their copycat blade on Monday. Odd that they'd choose a national holiday to announce, but that's another story ;-)."

Of course, when you go out of your way to eagerly point something out and declare you're not worried about it -- you're probably at least a leetle bit worried about it. Conceivably, Dell has surprised the competition with the level of innovation it's brought to the table here, enough to generate some concern.

Posted by Ted Samson on January 24, 2008 03:00 AM

January 16, 2008

AMD yields more efficiency, Intel more throughput in quad-core CPU showdown

TAGS: AMD, Green benchmarks, Intel, Power consumption, Processors

In a recent gauntlet of tests comparing AMD's quad-core Opteron processor (Barcelona) to Intel's quad-core Xeon (a.k.a. Tigerton), the Xeon delivered up to 14 percent more throughput, but the Opteron used up to 41 percent less energy.

The series of tests were conducted by Neal Nelson and Associates, an independent consulting firm. If you've been following my blog for a while, that name may sound familiar: Nelson has run similar power and performance tests over the past year, one in July and one in August.

Nelson conducted this series of tests using similarly configured quad-core Xeon and Opteron servers using 1GB memory modules at 4GB, 8GB, and 16GB main memory sizes. He also used one- and two-socket configurations at speeds of both 2.0GHz and 2.33GHz.

Cutting to the chase, Nelson determined the following in his tests:

-- When the sizes of the database working sets were small enough to fit in the servers’ kernel disk buffer cache, such that there was virtually no physical disk I/O, the Xeon-based servers delivered up to 14 percent higher throughput than the Opteron-based servers.

-- When the sizes of the database working sets were too large to fit in the kernel disk buffer cache, which forced substantial physical disk I/O, the Xeon-based servers delivered up to 3 percent higher throughput than the Opteron-based servers

-- When the servers were subjected to various identical levels of transaction arrival rates, the Opteron-based servers consumed up to 32 percent less power than the Xeon-based servers.

-- When the systems were idle and waiting for transactions to process, the Opteron-based servers consumed up to 41 percent less power than the Xeon-based servers. ("The power consumption at idle is particularly significant since studies have shown that many servers are powered on, but idle, 80 percent of the time," Nelson notes.)

"By themselves, the Intel processor chips may use less power, but all current Intel Xeon servers require the use of fully-buffered memory modules [FB-DIMM]. These FB-memory modules appear to consume more power than the DDR-II memory modules used by the AMD-based servers. The result is that in many cases an Opteron-based server actually uses less total power than a Xeon-based server," says Nelson in a written statement.

A better green benchmark?
Nelson reached his conclusions by employing what's he's dubbed Neal Nelson's Power Efficiency Benchmark. The benchmark works as follows: Nelson simulates users from 32 separate computers submitting individual transactions to similarly configured servers running Apache2 Web server, the MySQL relational database, and Novell's Suse Linux Enterprise Server O. He measures the throughput and power usage of the systems in increments of 50 users, from 100 to 500, over half-hour sessions. "The benchmark has a complex multi-user load with a large memory footprint, a high volume of context switches, significant network traffic, and substantial amounts of physical disk I/O," according to Nelson.

In addition to sharing his conclusions for his test, Nelson has made an effort to differentiate his power-efficiency benchmark from the one recently unveiled by SPEC. "The SPECPower test has a single-client machine feeding batches of 1,000 transactions to a small number of Java-based application programs," says Nelson. "[It also] has a small memory footprint, a low volume of context switches, simple network traffic, and it performs no physical disk I/O. The SPEC test was created by a committee of computer vendor employees, and SPEC offers no guarantee that their numbers will correlate to a customer's real-world experiences."

Nelson's test results can be viewed in their entirety on his Web site.

Related articles:
AMD launches Barcelona
Intel releases quad-core Tigerton
Study: AMD more power-efficient than Intel
In AMD-Intel square-off, memory proves key
SPEC seeds future green-server benchmarks

Ted Samson is a senior analyst at InfoWorld and author of the Sustainable IT blog. Subscribe to his free weekly Green Tech newsletter.

Posted by Ted Samson on January 16, 2008 11:50 AM

December 13, 2007

SPEC seeds future green-server benchmarks

TAGS: Green benchmarks

Backed by industry heavyweights, SPECpower_ssj2008 will be a building block for future power-performance standards

Over the past year or so, there's been plenty of head-scratching as to how to meaningfully measure a server's power performance: that is, how efficiently it uses energy to do its work. This kind of metric is important as datacenter operators struggle to keep energy costs down and free up floor space -- without sacrificing service quality.

Plenty of folks have invested resources and brainpower in the task, from independent analysts such as Neal Nelson and Associates and InfoWorld's chief technologist Tom Yager to large-scale organizations such as The Green Grid and even the EPA.

Thus, I was rather intrigued to learn this week that SPEC (Standard Performance Evaluation Corporation) has announced what it deems "the first industry-standard benchmark that measures power consumption in relation to performance for server-class computers." It's called SPECpower_ssj2008, a name that doesn't so much roll off the tongue as ooze -- but what's in a name, anyway?

Driving toward meaningful metrics
Before digging into the nitty-gritty of SPECpower_ssj2008, I want to provide some context as to why a server power-performance benchmark has proven elusive. As I said, plenty of smart people have been trying to devise one, and at first blush, it may seem like a deceptively simple task.

I like to compare it to the MPG measurement used to assess vehicles' fuel efficiency. You simply divide the number of miles you've traveled by the number of gallons used, and voilá, you have a meaningful measurement with which you can easily compare vehicular fuel efficiency. A high MPG, such as that you might get from a hybrid sedan, is deemed good. The low MPG you might get from an SUV is bad. Easy.

But wait: Perhaps it's not quite so cut and dry. Is it really meaningful to compare the gas mileage of a hybrid to that of an SUV if you don't factor in how the respective vehicles are being used? If you're comparing the two when the application is carrying three passengers and light baggage down Highway 5 from San Francisco to Los Angeles, the hybrid wins, hands down.

But what if the task at hand is schlepping five passengers and their camping gear along some rough terrain toward an off-the-beaten-path destination? There, the hybrid sedan can't really compete; it's not built to. What the SUV lacks in fuel efficiency, it compensates for with superior muscle and all-terrain features. Score one for the big machine.

Thus, it really makes the most sense to aim for more apples-to-apples comparisons, matching up like vehicles based on their form factor and the tasks for which they're being used.

In the world of servers, comparing power-performance is even more complicated. Servers vary in terms of the types of applications they run, their form factors, the number of processors they have, the speed of the processors, the amount and type of memory, and storage -- as well as how much heat they produce. Cooling, after all, ain't free. In fact, cooling a server can cost as much as running one.

Meet the benchmark
All of that brings us back to SPEC's benchmark, which was developed with assistance from big-name tech companies such as AMD, Dell, Fujitsu Siemens, HP, Intel, IBM, and Sun. The fact that so many companies -- all of which have a clear interest in seeing a benchmark that puts their respective wares in the best light -- are behind SPECpower_ssj2008 gives it all the more weight.

So let's get to the meat of the benchmark. SPECpower_ssj2008 measures server power consumption at different performance levels -- from 100 percent to idle -- in 10 percent segments over a set period of time. This graduated workload is important: It recognizes the fact that processing loads and power consumption on servers vary substantially over the course of days or weeks. Tests by Neal Nelson certainly bore that out. For example, he found that in idle mode, a server running an AMD chip used far less energy than did the machine running an Intel chip.

In order to calculate the power-performance metric, the benchmark measures and adds together the transaction throughputs at the various performance-level segments, then divides the resulting figure by the sum of the average power -- that is, the wattage -- consumed at each segment. The more work a system does at a given CPU utilization, and the less power it uses, the higher it scores on the benchmark. Thus, in essence, an SUV can be compared to a hybrid.

For the benchmark workload, SPEC selected server-side Java, representing Java business apps. According to the organization, "the workload is scalable, multi-threaded, portable across a wide range of operating environments, and economical to run."

Moreover, SPEC says the Java-based benchmark "exercises CPUs, caches, memory hierarchy, and the scalability of shared memory processors, as well as implementations of the Java Virtual Machine (JVM), JIT (just in time) compiler, garbage collection, threads, and some aspects of the operating system."

A spec in the right direction
The benchmark is certainly a good start as datacenter operators struggle to make sense of their machines' energy efficiency, as well the power-performance claims that hardware vendors boast.

In fact, SPEC is reviewing SPECpower_ssj2008 benchmark results submitted by vendors,then posting them for public consumption on the SPEC site. Thus far, HP's Proliant DL160 G5 tops the heap with a score of 698, followed by Dell's PowerEdge 2950 III, which scored a 682.

Although a good start, SPECpower_ssj2008 is just a first step toward measuring servers' energy efficiency. "SPECpower will work with other SPEC benchmarking groups to help them adopt the methodology used in this first benchmark. The intention is that there will be a wide range of SPEC benchmarks that incorporate power measurement in a consistent, repeatable way," writes Greg Darnell, vice chair of the SPECpower committee.

"The methodology can also be used by other benchmark developers interested in measuring power, in the hope that there will be a common set of practices in this new area of benchmarking," Darnell adds.

Additionally, SPEC will be looking at workloads and applications other than server-side Java, though "no definitive decisions have yet been made, however, on those workloads or applications," according to Darnell.

Indeed, running other types of workloads will certainly yield different results -- the way a hybrid's MPG on the highway will differ from its MPG on city streets.

IBM's Elisabeth Stahl, manager of performance marketing for the IBM Systems and Technology Group, expressed similar sentiments about the SPECpower_ssj2008 benchmark. (Big Blue was one of the companies that participated in its development.) "We believe this benchmark is a good first step in helping people to understand the relationship between systems performance and energy use," says Stahl. "We look forward to continued work to create benchmarks that broaden the spectrum of environments represented by the benchmark and to ensure that the data shown is representative of the many computing environments that exist."

The minimum equipment for SPEC-compliant testing is two networked computers, plus a power analyzer and a temperature sensor. One computer is the system under test; the other is the controller system where power, performance, and temperature are captured for reporting. A typical test run for SPECpower_ssj2008 takes about 70 minutes using default settings.

SPECpower_ssj2008 is available immediately from SPEC for $1,600. For more information, go to SPEC's Web site.

Related links:
In AMD-Intel square-off, memory proves key
In search of energy benchmarks
EPA seeks input on Energy Star for servers
IT heavyweights plot The Green Grid

Ted Samson is a senior analyst at InfoWorld and writer of the Sustainable IT blog. Subscribe to his free weekly Green Tech newsletter.

Posted by Ted Samson on December 13, 2007 03:00 AM