Sentilla Blog

Hello, I'm Spence Murray, a member of the Sentilla Server Engineering team. My responsibilities include design and development of core networking and data management technologies for Sentilla Energy Manager. Having been involved in the project since its inception, I've had the pleasure of seeing it evolve into a very useful tool for both energy and network management.

First off, I'd like to congratulate our CTO, Joe Polastre, on his recent selection as a one of BusinessWeek's Best Young Tech Entrepreneurs of 2009. This award is indicative of the growing buzz Sentilla Energy Manager is generating.

Today, Sentilla announced version 3.0 of Sentilla Energy Manager.  My role at Sentilla is varied, but one of my responsibilities is creating and managing the product roadmap.  With version 3.0, I'm really excited about how much we've added into this release.  SEM 3.0 is truly revolutionary, providing a ton of features and functionality that no other vendor provides.  It is built on our Sentilla Software Platform, which is in its 4th generation, is very robust, and has served as the basis of all of our products since 2006.

On the same day, two opposing articles have been published with completely opposite points of view.  On one side, Brian Fry argues that location no longer matters and we should build data centers in the most efficient locations and supply fiber connectivity to them.  On the other side, Paris Burstyn argues that location and latency are business critical for companies resulting in Equinix's acquisition of Switch and Data.

At the center of Brian's argument is that low cost, low carbon power is good for data centers.  If you can live with your data centers a few milliseconds away from everyone else's, then your IT operations can be greener and cheaper, a win-win scenario.  But in the world where every millisecond counts, that's not the right solution for all industries.

Recently, there's been a slew of articles about how IT managers have identified that standardized access to power information over SNMP is one of the top ten problems that they face when managing energy.

We were listening. And today we announced the release of Sentilla Energy Manager for Data Centers version 2.1 which integrates all the great things in the original release like the ability to analyze your energy profile at each piece of equipment and we extended that to include third party equipment. That means that, within an hour of installation, you can get a high level overview of what's going on across the entire data center.

This blog entry first appeared as a contributed news article at ZDNet.  The version posted here includes additional pictures and links to various related topics in line with the post, to provide greater context and justification for the claims made about Google's data centers.

Sentilla is honored to win another product award.  We’re very excited that energy management has gained so much traction this year.  But it’s no surprise.  As the UK embraces the Carbon Reduction Commitment and with the publication of the European Code of Conduct for Data Centres, it has become clear that a solution is needed for managing and tracking data center energy use. The award validates the need for energy management in data centers and Sentilla is humbled to be part of this movement.

It is not always clear what the right way is to measure the success of your data center's energy strategy.  While many would promote PUE as the right way to go, it does not tell the full story.  Think about these examples:

• A data center is commissioned, but no IT equipment is installed.  The PUE is infinitely high, but it doesn't mean the data center is inefficient.  As IT equipment is added, the PUE drops but the data center's efficiency did not improve.
• As servers and storage are consolidated or virtualized, the overall power consumption of the data center decreases.  Not only is there a reduction in the IT power, but a smaller corresponding reduction in the cooling required.  While the net is a win (more power capacity and less power usage), PUE typically increases.

These aren't the only bad examples of PUE, but they illustrate the fact that it is simply a ratio and it must be understood when talking about it. 

There's been a lot of discussion recently about the difference between monitoring and management.  This week, I received an email with the title "Revolutionizing Energy Management".  Interesting, I wonder what management solution this company provides for energy.  The content of the email went on to talk about a brand new meter that provides real time power load information.  While I'm sure this company's meter is very innovative (names purposely omitted to protect the innocent), it was clear that this company did not understand what the word "management" even means.

I'd like to point out the difference between management and monitoring.  There are a LOT of tools for monitoring but much fewer for management.

Monitoring is the process of being aware of the state of a system.  It involves observing the current situation and typically necessitates a measuring device or meter.  Monitoring typically results in a large set of data, un-correlated and un-analyzed.  The data is not tied to your business objectives but is just data, it is that simple.  It is up to you, as the human, to figure out what all this data means.

Management, in contrast, is the act of getting a system to deliver a desired goal/objective.  It involves managing and allocating resources, organizing resources to execute a task, designing and re-designing systems, and optimizing a system to produce useful outcomes.  Monitoring is a key component to management; after all you need credible information to make decisions.

With a lot of the data center energy efficiency focus on facility improvements and virtualization, I've decided we need to take a step back and look instead at the applications.  As Moore's Law has increased the capabilities of servers, and disk density has doubled (roughly) every year, we're no longer as constrained as we used to be by physical resources.

I am, at the heart of it all, a computer scientist.  In grad school, I spent a lot of time working on optimizing software that runs on embedded systems -- little microcontrollers with limited resources.  At the core of these systems was power.  If you wrote your code inefficiently, not only would it fail to fit on the device, it would also burn through batteries.  And as data center operators know, a server without power is a VERY BAD thing.

So why is it then, that we don't look at how efficiently our applications are written?  There's no apples-to-apples energy comparison, and few people take energy into account when buying a software package.  My prediction is that will change.  In a few years, when you evaluate whether to buy (or renew) SAP or Oracle, you'll ask about the energy operating cost of those software packages over the life of the contract.

Figuring out how to implement the right efficiency plan in the data center is a daunting task.  I often start discussions with customers by asking "How much electricity does your data center use?" and "How much are you paying for electricity?"  While these may seem like such amazingly fundamental questions, you'd be surprised by how many people don't know the answers or respond with "Let me go and look that up."

Further confusing the issue is vendors claims about what they can achieve.  Have you heard the marketing campaign about virtualization reducing your energy bill by 80%?  Sounds compelling right?  The trouble is, you're not going to virtualize every server in your facility and shut off every chiller.  Let's say 20% of your servers can be virtualized (such as staging, dev, and test), and your servers consume 70% of your IT energy consumption, which is 45% of your facility's energy consumption.  So if we add that up, .8*.2*.7*.45 = 5%.  That means that, for your data center, 80% savings just became 5%.

This means that data centers really need to figure out what their strategy is.  And, of course, you want the biggest bang for your buck.  By measuring your baseline, applying the various strategies to it, and calculating ROI, a plan can be formulated that makes sense of the efficiency measures for your specific facility.  I've written an eWeek HowTo titled How To Achieve 40% Energy Savings In Your Data Center.