While Arch Rock had great technology, the business model was never formalized — Arch Rock routelinely messaged the “Internet of Things“, a vague concept where every device is connected to the Internet.  In contrast, Silver Spring focused hard (think “crossing the chasm“) and aimed to be the network platform provider to the smart grid.  They established firm OEM partnerships and secured early adopter utility companies.  Arch Rock, on the other hand, was still wading through a variety of market verticals and never got the chance to focus on the smart grid sector.  It was a classic case of a technology in search of a solution.

Here comes Cisco to the rescue, who takes an interest in Arch Rock’s networking technology to enable the smart grid.  This is not a coincidence — Arch Rock had boasted Roland Acra (CEO) who sold Procket Networks to Cisco in 2004 after previously working at Cisco for 11 years, and Judy Estrin (external board of directors member) who formerly was CTO at Cisco.  The price of the acquisition was not announced, which is not surprising, since Cisco would be unlikely to pay much simply for technology in today’s economy.

Arch Rock spent an incredible amount of time in trying to establish 6lowpan as an international standard at the IETF (Silver Spring only peripherally participates, and that’s a recent change).  This was one of Arch Rock’s fatal flaws — adoption by customers in all of these verticals was not dependent on a new standard existing.  Instead of spending each hard-earned dollar on new business, the money was spent on standards and technology, which leads to the difference in success between Arch Rock and Silver Spring.  A business vision and customer validation led to $255 million raised by Silver Spring on good terms to build an industry; Arch Rock raised $15 million to test a technology.

Fast forward to today.  Silver Spring has a multi-hundred million dollar backlog of orders, and has been able to branch out beyond the “network card for smart meters” that they used to get a foothold.  They are addressing security and application frameworks in the smart grid, demand response programs, smart home initiatives, and even intelligence in electric vehicles.  They have a strategy that is light years ahead of both Cisco and Arch Rock when it comes to understanding the market and predicting where it will end up.  This acquisition is good news for Silver Spring — it validates their business, shows they are the market leader, and will further drive up their valuation beyond the $3b it is estimated at today!  Congratulations Silver Spring!

• Silver Spring has raised $255mm in investment from the top names like KPCB, Arch Rock raised $15mm.
• Silver Spring has approximately $800mm in order backlogs.  Arch Rock has $0.
• Silver Spring  is technology independent (WiFi, GSM, Zigbee, 6lowpan).  Arch Rock is 6lowpan only.
• GE, Itron, Landis+Gyr, and others are all making meters with Silver Spring technology.  Cisco has only been able to land Itron so far (just announced this week!) and Arch Rock didn’t have any smart grid partners.
• The brain trust behind Arch Rock, Professor David Culler, won’t stick around and instead is driving smart grid innovation at UC Berkeley’s LoCal project.

What we can tell is that Cisco wants to play in the smart grid space, but they realize they need to do a lot of catch up.  Arch Rock is just one step in a process that will undoubtably take years to establish partnerships, technology, and product.  Cisco is infamous for pre-announcing products as if they exist to drive market vision, only to find they won’t ship for another 9-12 months.

There’s a few lessons to be learned here for startups:

1. Great technology doesn’t make great companies.  You need business leadership.
2. Focus on a market vertical and make that one successful first.
3. Identify all of the partners (VARs, OEMs, influencers, etc), sign them up, and lock everyone else out.
4. Sell investors on the big vision — “this one vertical is just the beginning”.

The whole idea of freezing ice in the evening when power is cheap and then melting the ice during the day is so simple it is brilliant. It isn’t very different to when one of Sentilla’s founders, Rob Szewczyk, observed that you can compute how much energy any piece of equipment is using (from washing machines to computers) if you know how much work it is doing.

I first heard about ice energy storage about 3 years ago, when an urban legend was circulating through the IT community. A gentleman named Bill Beck had been rumored to store ice in the basement of a Credit Suisse building on Madison Avenue in New York City. First, who would even believe you if you told them you were storing ice in the basement of a Manhattan skyscraper? Second, you’d probably think Mr. Beck was simply crazy — why create vats where ice could be stored simply for cooling data centers (not even the building as a whole!)? Today, Credit Suisse calls these vats “slush tanks”. While no one will confirm the rumors, I’m inclined to believe they do exist.

The most publicized instance of ice energy storage is with the new Bank of America building at One Bryant Park, also in Manhattan. They’re not the only ones, the USGBC estimates that there are thousands of ice energy storage projects in the US. For example, Glendale, California has begun installation of a 53MW ice energy storage project. Here’s the idea: we store ice at night when energy is from wind power (renewable, and thus low carbon impact) using refrigerant that freezes water in a tank. During the day, the same refrigerant flows through the tanks, and due to the higher temperature, melts the ice (which cools the refrigerant) which is then used in the chillers. It is even more of a “win-win”: the air pumped from the ice cooling is actually colder, increasing occupant conform during the summer months.

Why even bother with this “complex” process? The idea is fairly simple (only complicated by the utility companies): if I reduce the total “demand” of the building, then I reduce its energy bill. Commercial and industrial electricity users are charged by both how much they consume (consumption rate) and the maximum demand (demand charging). So if I use 1MW of electricity on average (8.76 million kWh) but my peak demand is 2MW, I get charged for the 8.76 million kWh plus a monthly demand charge for the 2MW (typically charged per kW demand). Why? The electricity company has to provision that amount of power distribution in their substations, so they’re charging for the cost of maintaining the distribution of that much power.

As a result, if you can smooth out your consumption so that there’s less peaks, you can reduce your electricity bill. This is exactly what Bank of America and Credit Suisse are doing. Use more energy at night when the consumption rate is cheaper and demand is lower, then use less electricity during the day when demand is typically higher and consumption rates are higher. Seems brilliant, but what I love about it is that it is simply brilliant.

Printers alone are estimated to be 12% of the IT cost in 2012 (8% in 2007).  The total carbon impact of IT and Telcom equipment will be more than 1.5% of global emissions by 2020, 57% of which is attributable to PCs and office equipment.  While there’s fancy software out there that will power down these systems for you, the reality is that you can do most of this yourself too!  Just set up a policy that everyone turns things off when they go home, and power it back up when they get in to work.  This weekend is a good opportunity to introduce this concept to your company.

To do my own part, I have installed a TED monitoring device at my home in San Francisco.  I used this data to decide what to turn off and its impact when I go out of town.  I found that I can unplug my printer, turn off my thermostat, and power down everything around the TV except for the DVR.  I even unplug the microwave and toaster.  The impact has been about a 50% lower energy use during periods that I’m not home.

There’s a side note here about PG&E’s smart meter installation.  They’ve decided that Silver Spring’s technology is not able to handle the density and elevation changes in San Francisco, as told by a PG&E representative in a recent SVLG meeting.  But there’s other complications too, San Francisco City Attorney has also asked PG&E to halt smart meter installs in San Francisco.  The good news is that PG&E hasn’t started in San Francisco or the very rural outlying areas.  Thus the need to install my own meter.

Back to energy efficiency — the same idea holds for the office as it does for my home.  People tend to work 8ish hours per day (unless they’re in a startup!) and not on weekends.  By powering down for the weekend, we can make a big impact.

What’s more important, is that EO 13514 has specific language about data centers.  Federal organizations must “implement best management practices for energy-efficient management of servers and Federal data centers.”  The goal is to bring the Federal government in line with commercial best practices.  The government has thus set some very generic “Guiding Principals”, and then it is up to each of the organizations to meet their stated GHG goals for Class 1, Class 2, and Class 3 emissions. Let’s take a look at what some of the government organizations have done to comply with EO 13514.

The IRS has taken, in my opinion, the most significant stand on improving energy efficiency in data centers.  They commissioned an audit late in 2009 from the Treasure Inspector General for Tax Administration, and the results of that audit including IRS rebuttal were made available in May.  While the audit is a good first step, the results and recommendations are not earth-shattering, nor is the analysis behind the recommendations.  The good news is the IRS is trying to find out exactly where their weaknesses are (primarily focused on airflow management, a red herring to the real problem of underutilized equipment and inefficient/old applications), the bad news is that the IRS and auditor differ in opinion, and the timeline to rollout efficiency upgrades is really extended (to 2012 and beyond). Most “best practices” can be implemented in 6 months or less in a moderately-sized data center.

The #1 recommendation was to implement “policies and procedures to evaluate and determine which best practices to implement.”  Basically it says that, now that we’ve identified a number of areas for improvement, someone has to run the analysis to decide which to implement.  The #2 recommendation to the IRS?  Implement an energy monitoring and management system, including sub-metering down to the equipment to get an accurate read on which is using the most power.  Guess what?  Sentilla Energy Manager with Virtual Meters can do that for you without having to buy, install, or commission a single meter!

While the Department of Defense has not disclosed specific data center plans, the DoD has set an agressive target of 34% reduction of GHGs by 2020.  The DoD consumes 80% of the electricity used by the federal government.  The buildings and land operated by the DoD add up to an area the size of Pennsylvania. The Defense Information Systems Agency operates the 14 DoD data centers (consolidated from 50).

In other news, the NSA is planning a $1.6 billion / 1,000,000 square foot / 70MW data center in Utahto consolidate operations and support future growth.  It can neither be confirmed nor denied that the NSA data center in Fort Meade, Maryland is responsible for power grid brown outs. Keep in mind that Federal organization can opt-out of EO 13514 in the interest of “national security”. I’m sure that NSA is going to do exactly that.

The federal government also has a huge data center consolidation plan underway, being led by the OMB.  The plan, discussed at CIO.gov, requires all Federal data centers to take a full inventory of their systems and then identify each system with a code between 0 and 5 that determines the system’s eligibility for consolidation.  Many have pointed out that this is an impossible task, and I tend to agree.  To expect an inventory of hundreds of thousands of servers to be completed in a reasonable time period without pre-budgeted tools is a nightmare.  While the OMB will provide a list of tools that may be used, they do not provide funding or assistance.

To make matters even worse, EO 13514 requires that all government agencies disclose their GHG goals and have their progress tracked publicly.  But there is no central place that currently exists (shame on you OMB!), and CIO.gov has failed to track the IT compliance progress of government organizations.  Agency reportcards are in different places and some cannot even be found online.

The Federal government has a good ways to go before they are in compliance with EO 13514, and they certainly have their work cut out to make their data centers more energy efficient.  The best solution is to deploy automation and management tools — use the same systems and best practices that commercial organizations have adopted in order to identify, strategize, and act on the highest return energy efficiency projects.

Anne is an entrepreneur herself, hailing from the great white north of Canada. Her podcasts interview entrepreneurs to describe how crazy people like me got their start. Anne now works for SunRun, a startup the brings solar to the home, after co-founding a solar company in Canada. Give it a listen, her episodes are great and she has a keen business eye for cleantech innovations.

Stream the audio or download the mp3 to listen later.

Visit Anne online: visit her webpage, the YII webpage, or her podcasts.

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.

When we look at the data center, there’s no shortage of examples of monitoring and management tools.  To illustrate the high level description of these terms, let’s look at systems management versus systems monitoring.  Since Wikipedia is the source of all true and credible information in the universe, I’ll reference them for this example:

A System Monitor is a process within a distributed system for collecting and storing state data.

Systems Management refers to enterprise-wide administration of distributed systems including computer systems

There’s a huge established market for commercial and open-source tools for both system monitoring and systems management.  And if you have a small data center, you can probably get away with open source system monitoring tools and accessing a few different systems to perform planning and analysis tasks.  If you run a 10MW data center with 100,000sqft and 20,000 servers, system monitoring tools are just one piece of the puzzle to create a comprehensive plan.  System monitoring tools keep your operations team abreast of key risks, but they’re not enterprise-wide and there’s too many to manage (intentional word choice).

While you may use system monitoring tools like IBM Director, HP Insight, APC InfraStruXure, Nagios, and Hyperic, you also use systems management tools like IBM/Tivoli, BMC, and JCI Metasys.  The problem with the modern data center is that every hardware vendor now provides a monitoring tool, and if you have a typical data center, that means there’s not a vendor that you don’t like (or buy from).  Now there’s 37 monitoring tools to manage, and a colossal time sink for your team.  Introduce the uber-monitoring tools, that monitor the monitors.  It is the first step to management, but they’re not management tools.  Management tools like Tivoli or BMC help you strategize about deploying services, evaluate their performance and efficiency, and then automate deployment and maintenance of these applications.

Then why do vendors talk about management when their products really provide monitoring functionality?  There’s two answers to this question.  First, they’re not trying to fool you, and there’s just a misunderstanding about the difference between monitoring and management.  Fair enough, I misuse tons of words every day.  (Don’t worry, IT isn’t the only industry that misuses monitoring and management.) Second, these vendors are trying to promote their product using a buzzword that implies it provides more value in order to fetch a higher price tag (marketers for centuries have used “words that sell”).  This is the more likely outcome, since management impacts business processes (which means big impact on financial returns) whereas monitoring ensures that you’re notified when things go wrong (also very important that the “trains run on time”).  The difference, however, is that keeping the trains on time is generally not valued as highly as creating corporate and business strategy that drives improved efficiency, process changes, and bottom line measurable impacts.  Think of it this way: It is harder to quantify the risk of an outage (and the probability that the outage will even occur) than it is to quantify an alternative way of doing business or running applications.

When you’re looking to improve the business value of your data center through energy management tools, here’s a short checklist of things you should look for:

• “Enterprise-wide” scope: Consolidating information from all of the monitoring tools, so you have a central repository and a single pane of glass.  Not just facility data, not just IT data, but information about all of the data center’s operation.  After all, the data center is there to run applications; how can you evaluate their performance without facility and IT data in one place for analysis?
• Performance reporting, so you know which users, applications, and systems are to blame for the greatest inefficiency
• Configuration management, so that everything is accounted for (often done with integration to a CMDB or asset database)
• Planning for future needs, capacity, or resources
• Historical data tracking to validate goals and project future resource needs to continue meeting goals
• Analyzing allocation and actual use of resources to evaluate efficiency
• Ability to implement new business strategies/processes
• Automating data center strategies (where applicable) to drive optimization (and free your time to work on other projects)

Don’t be fooled by monitoring vendors that talk about “management”.  Let me leave you with an observation: In every case that I can remember, the vendor that sells hardware provides monitoring tools, whereas software companies have always been the provider of management tools. Drill down to find out what analytics, reporting, business impact, and process improvements these tools actually provide.

This blog entry first appeared as as a blog post at Sentilla.