]]>
There's four main components to Sentilla Energy Manager: data acquisition, Sentilla Inference Engine, Sentilla Analysis Engine, and sophisticated presentation via dashboards and reports. We've taken an IT enterprise management approach to power: energy is a key asset in the data center and you should be managing it just like your other assets, such as servers, storage, networking, and applications. By managing it as an asset, its not just about cost reduction. I can't tell you how many people I've talked to that have power provisioning and capacity problems and stay up at night worrying that the data center might go down. Capacity planning is a huge concern as the growth of IT services and the cloud skyrockets. And how about performance: let's get the most out of the equipment we have. Most data centers utilize just a few percent of their computing capacity but spend the money to run those systems non-stop.Let me start with the Sentilla Inference Engine, which is new in this release and a major step forward for the industry when tracking power. Lots of data center managers have lamented to me that they'd like to track the power profile of their applications, but installing power meters on every server is simply infeasible -- the disruption is too much and the cost is too high. What the Sentilla Inference Engine does is compute the energy consumption, continuously in real time, of every asset in the data center -- metered or unmetered. The result is full coverage of everything in a data center, including UPSs, CRACs, power distribution units, and every asset, server, storage, network gear, or piece of equipment. We're the first and only company in the industry that can provide this level of granularity. And we do it in real time.
Of course, we take metered data from the BMSs, UPS, Modbus and SNMP devices, a wide range of PDUs, HP iLO, Sun ALOM, and other server basebands, etc. But that usually doesn't extend down to the servers, and the servers are what IT managers actually care about. So we construct the "power tree" of the data center using the one line diagram and track the power from the switchgear down to the assets. We attach meters to their location in the hierarchy and we balance the tree and compute the consumption of unmetered branches, like water flowing through a series of smaller and smaller pipes. We also approach the problem from the bottom up: the Sentilla Inference Engine, through integration with your existing system management tools like Tivoli, Ganglia, and Nagios, computes the power consumption of unmetered servers, storage, and networking. With information about how much work is done, Sentilla knows how much power is consumed -- with extremely high accuracy. We have built an extensive set of profiles and models of equipment and validated our models with real data and SPEC power benchmarks. It seems like a no-brainer: you already have information about the applications that you're managing to ensure SLAs and keep them available. Why not use that data to figure out the power consumption too? We've filed patents on multiple aspects of the inference engine, and we intend to protect Sentilla's innovative solution to data center energy management.With the Sentilla Inference Engine, energy data from facilities and IT is consolidated to a single view. Facilities can better provision power, and IT can better track performance, efficiency, and risk. It is a win-win for both sides. Sentilla has introduced new dashboards based on feedback from the largest financial, telecom, and manufacturing companies. You can now track key metrics right on the front page, watch your favorite (or less favorite) assets, see alerts, and uncover key trends. Everything you'd expect from an enterprise management solution. Integration is key with Sentilla Energy Manager; we're here to complement your system management tools, integrate with them, and make your overall management environment richer, more comprehensive, and unified.
We've beefed up the Sentilla Analysis Engine, which takes all of the raw data from the meters and Sentilla Inference Engine, and crunches it to provide actionable intelligence. Basically we're turning data (billions of data points) into content. We provide warnings and alerts based on thresholds and capacity limits, and you can track these capacities through the power tree. One of the new features in this release that's been incredibly useful so far is the ability to identify idle servers. The Sentilla Analysis Engine studies the consumption and utilization of every asset and alerts you when equipment is idle for prolonged periods of time. We've already seen reductions of 20% or more from using Sentilla's idle detection feature. And the operating cost reduction of removing idle servers is huge: the elimination of costs in operational management, software licensing, switch fabric, etc can be up to 10 times more than the power cost savings (source: IDC).
Finally, a whole new flexible reporting system is introduced in SEM 3.0. Chargebacks are becoming a reality, as facilities want to offload the power bill from their books. And what better way to improve efficiency than to make each line of business responsible for their use? We've added a number of report templates, including PUE, asset utilization, carbon impact, alerts and recommendations, and chargeback. You can customize them, and they're built using the Jasper Reports platform (from the inventor of Crystal Reports), giving you full flexibility to develop reports that fit your business needs.At Sentilla, we're really excited about how we're revolutionizing the way that data centers manage energy. Stop by and say hi at Data Center World in Nashville, and attend my educational session (not product pitch) on key energy management metrics at 4:30pm on Wednesday March 10th (session FG06).
Sentilla is rolling out new website content and white papers, so keep an eye on this space. If you want to learn more about SEM or have questions, drop me a line, post a comment, or email sales@sentilla.com. ]]>