The machine in the photo above is HP’s newly announced Redstone server development platform.

Capable of fitting 288 servers into a 4U rack enclosure, it packs a lot of punch into a small space. The servers are System on a Chip based on Calxeda ARM processors but according to HP, future versions will include “Intel® Atom™-based processors as well as others”

These are not the kind of servers you deploy to host your blog and a couple of photos. No, these are the kinds of servers deployed by the literal shedload by hosting companies, or cloud companies to get the maximum performance for the minimum energy hit. This has very little to do with these companies developing a sudden green conscience, rather it is the rising energy costs of running server infrastructure that is the primary motivator here.

This announcement is part of a larger move by HP (called Project Moonshot), designed to advance HP’s position in the burgeoning low-energy server marketplace.

Nor is this anything very new or unique to HP. Dell have been producing microservers for over three years now. In June and July of this year (2011) they launched the 3rd generations of their AMD and Intel based PowerEdge microservers respectively.

And it’s not just Dell, Seamicro has been producing Atom-based microservers for several years now. Their latest server, the SM10000-64 contains 384 processors per system in a 10U chassis with a very low energy footprint.

And back in April of this year Facebook announced its Open Compute initiative to open-source the development of vanity free, low cost compute nodes (servers). These are based on Intel and AMD motherboards but don’t be surprised if there is a shift to Atom in Open Compute soon enough.

This move towards the use of more energy efficient server chips, along with the sharing of server resources (storage, networking, management, power and cooling) across potentially thousands of servers is a significant shift away from the traditional server architecture.

It will fundamentally change the cost of deploying and operating large cloud infrastructures. It will also drastically increase the compute resources available online but the one thing it won’t do, as we know from Jevons’ Paradox, is it won’t reduce the amount of energy used in IT. Paradoxically, it may even increase it!

Photo credit HP

Follow @TomRaftery

In the video above, Dave Wright, founder and CEO of SolidFire makes the point that what with ARM-based servers, OpenCompute, etc. there has been a lot of breakthroughs on the computing side of servers, to make them more efficient recently, but very little innovation has happened with storage systems. Predictably he’s gone after storage modernisation with his new company SolidFire offering SSD-based enterprise storage solutions.

My laptop

One of the biggest advantages of SSD’s, as storage for servers, is it is incredibly fast, so you get an immediate performance win. I first found this when I changed my laptop to one with an SSD, instead of a normal HDD. The drive is far faster, but because the SSD doesn’t generate heat, there is no requirement for a fan. This makes the laptop cooler (no laptop burn), quieter and it has a far longer battery life. Samsung affirmed this in a server situation when I talked to them earlier this year. Because SSD’s don’t require power-hungry fans to cool down the heat created by spinning drives, the reduced power requirement and heat generation is a big win in a data centre environment.

SolidFire are far from being alone in this field. Just last week FlashSoft announced that they had secured $3m in first round funding to develop Flash virtualisation software for enterprises. They have nifty software which runs on servers with hybrid storage (some SSD and some HDD). Their software identifies regularly accessed data (hot data) and caches this in SSD, while moving less frequently accessed data to spinning disks. Having regularly accessed data in a cache on SSD greatly increases the performance of the storage.

The hybrid model is one way of getting over the issue of the cost differential between HDD’s and SSD’s. SolidFire have a different approach – they don’t go for the hybrid model. Instead their all-SSD model uses a combination of data compression, de-duplication and thin client provisioning to reduce the amount of space required for storage.

A performance enhancing tactic regularly employed with HDD’s is to only use a small amount of the available space on the outside of the disk for your storage. The outside of the disk spins fastest giving you faster read/write access. However, this is hugely inefficient as most of the disk remains unused.

SolidFire do away with the need to have any HDD’s at all making your storage far more efficient. While in Flashsoft’s hybrid model, you can do away with the requirement for faster spinning SAS drives and instead go for slower, cheaper SATA drives without taking a performance hit. Both solutions reduce your energy and cooling needs.

Then out of Japan comes news that in response to requirements for energy efficiency there (due to the earthquake earlier this year closing nuclear power plants), Nexsan have come up with new power managed storage systems with in-built MAID capable of supporting any combination of SATA/SAS/SSD drives. Because MAID allows disks to be spun down when not in use, Nexsan are claiming up to 85% savings in energy usage for its systems.

It is true certainly that SSD’s have a shorter lifetime than HDD’s but even this has been given a boost with the recent announcement from IBM that their new Phase Change Memory chips (PCM) will be faster, cheaper and longer lasting than todays SSD’s.

So while Dave, above, feels that there isn’t much innovation happening in the efficiency of storage, I would respectfully differ and say this is very exciting times to be looking into storage energy efficiency!

Follow @TomRaftery

Photo credit Tom Raftery

Samsung have an interesting section on their GreenMemory site about Solid State Drives and how they are considerably Greener than traditional HDD’s.

This rings true with me because my own laptop has an SSD instead of a HDD and I find that it out-performs my desktop for some applications, it runs cold (and quiet as it doesn’t need a fan) and the battery life is better than any laptop I have owned to date.

I caught up with them at the recent Sapphire Now conference where they kindly led me through a demo, showing exactly how much better SSD’s perform compared to HDD’s..

Here’s a transcription of the demo:

Tom Raftery: Hi, everyone! Welcome to GreenMonk TV. I’m at the Sapphire Now conference in Orlando and I’m with Steven Peng from Samsung. Steven is demoing a new way of looking at SSD’s, solid-state disks and some interesting metrics around the solid-state drives. Steven can you tell me why you are telling me that solid-state drives are Green?

Steven Peng: Yes, solid-state drive doesn’t have a moving part by nature, you can save more power of a traditional hard disk drive and you can say that’s the reason that it is Green.

Tom Raftery: Okay. So you have numbers here, you can talk to around that both in terms of cost and in terms of throughput. Can you talk us through some of those?

Steven Peng: Of course, welcome to our Green SSD versus HDD demo. And a lot of people state the SSD you get cost, it’s too high compared to current technology hard disk drive. However, if you look at a system cost, it actually is the cheapest cost in delivering the performance needed in systems.

So, I have a demo here, I have two identical systems, server systems. If you look at the CPU, the memory configuration, they are identical. However, if you look at the Green SSD system, I have four Samsung MLC SSD drive at 200 gigabytes each, so total capacity is 800 gigabyte. And I can also put two or three 7200 RPM HDD.

So the total capacity is a really comparable to the HDD system here. In this system, I have a 12 high speed 15K SAS HDD, a 300 gigabyte drive capacity each.

So, if you look at two systems, the cost of four SSD drive and 12, 30 gigabyte 15K hard disk drive, the cost is about the same. So, the system costs are same.

Tom Raftery: It is compatible, okay.

Steven Peng: Yeah. So, the interesting area of what about the performance that really tells the system cost difference, right. So the demo here is, we are also running the Benchmarking software, TPC-C and online transaction processing benchmark software and that’s software allows us to know what’s the difference in terms of performance.

If you look at the number here, we recorded in the Green SSD server system, we have 7,000 transaction per second level. If you look at the hard disk drive HDD server system, we only recorded like 1,900 transaction per second level. So, immediately it is about 3X delta.

Tom Raftery: So, it’s significantly faster in terms of performance, but also the amount of watts that are being consumed are considerably lower as well?

Steven Peng: Exactly. So the wattage of a Green SSD system consumption is about 170 watts. If you look at the HDD system, it’s 280 watts. So, you see the delta right there, 60 percent range in power saving per system.

Tom Raftery: And that power saving is coming from a couple of things, I assume, you can tell me if I’m wrong, my assumption would be that a) There is no moving parts as you said earlier but also, b) There is no heat being generated or less heat being generated by the SSD than the hard drive, would that be right?

Steven Peng: Yeah, indeed. Since SSD, the system you generate less heat you don’t need the fan to spin faster and also you can imagine in the data center, you can spend less cooling cost.

Tom Raftery: Cool, so faster performance, compatible pricing and lower operating cost in terms of power usage.

Steven Peng: Right. So, we suggest to people when you are doing your next IT refresh server system, look into SSD system and that can give you the lowest initial purchasing cost and also the ongoing operating cost saving from the power you know saving payout.

Tom Raftery: Steven, that’s been great. Thanks a million.

Steven Peng: Alright, thank you Tom.

You should follow me on Twitter here

Back in 2008 I wrote a post asking how Green is Cloud Computing?

Back then I didn’t really have an answer to the question. I mean logic would lead you to think that Cloud Computing leads to more efficiently run computers, so therefore it HAS to be Green, right? However, at the time, none of the Cloud providers were publishing their energy utilisation numbers, so no conclusions based on solid data could be drawn.

Fast forward from 2008 to now. How many Cloud providers are now publishing their energy utilisation info? Oh that’s right, none of them are.

There is a move by some Amazon shareholders calling on the company to prepare a report that will assess the impact of climate change on Amazon and make it public – but there is no guarantee that even if Amazon vote to do this, that they will include detailed energy consumption data.

One thing which might force this issue is a requirement for organisations to report emissions data – until that happens though, if the last three years is anything to go by, it doesn’t look like Cloud providers will be publishing their efficiency data any time soon.

What we need to see from Cloud providers is data in the form of watts/compute cycle so we can cross-compare their efficiency, and compare it to alternative infrastructures. This is something they have been singularly reluctant to report on to-date. One has to wonder why. Could it be that, in fact, Cloud Computing is hugely inefficient?

And even if Cloud Computing is shown to be more efficient, as Simon Wardley is fond of pointing out, Jevon’s Paradox, may mean that we end up using much more of a more efficient resource, paradoxically increasing energy consumption, which is definitely NOT Green.

So what do you think? Is Cloud Computing Green Yet? And if not, will it ever be?

You should follow me on Twitter here

Photo credit supertin

I went along to the MashupEvent Energy 2.0 – Energy goes Digital get together in London last Thursday.

It was a good event with some interesting speakers, including Usman Haque of pachube, Ajit Jaokar from FutureText and Paul Tanner (self-confessed energy nut!).

The talks were good – for me, one of the more interesting learnings was how pachube is being used to crowd-source rediation readings from hacked Geiger counters in Japan! Seriously awesome stuff, and a real case of people using the pachube platform for purposes never dreamed of when it was first created, I suspect.

When the floor was opened for questions and discussion, some interesting topics were raised. When the question was asked from the podium, one brave member of the audience confessed to being from a utility (British Gas) and she went on to raise an interesting point – she said it was hard to motivate people to to make any changes. British Gas, she said, have offered people free insulation, which would potentially save them hundreds of pounds, and they don’t take up the offer.

This is not the first time I have heard these kinds of stories. Why is that?

Toyota Prius dashboard driving info

There are a few reasons for this, as far as I can see:

1. When you are getting electricity bills like the one above, you have no idea what your actual consumption is like day-to day, minute-to-minute. I bought a Toyota Prius a number of years ago and it totally changed the way I drive because of how well the consumption information is fed back to me on the dashboard – the same is not true for the Honda Insight, as I discovered when Honda lent me one to trial, so not alone is it important to give people information on their consumption, it is also vital to present it in an easily digestible way.
2. People don’t trust their utility companies – traditionally utility companies only communicate with their customers (who they often refer to as ratepayers!) when they are sending a bill or when a bill is overdue. This form of communication is not particularly conducive to establishing a good relationship. The basis for any good relationship and thereby trust is communication. Utility companies need to radically step up their customer communications, but not in a way that is spammy. They need to engage with their customers on all channels (if my preferred method of communication is Twitter, I want them to engage with me there, if my Dad’s preference is phone – talking to a person, not an IVR, then that’s where they need to talk to him). This is going to be a hard lesson for many utilities to learn but they fail to learn it at their own peril and
3. Energy is too cheap! Possibly fixing 1 and 2 will persuade people to become more energy efficient, but I suspect, the real driver for energy efficiency will only come along when energy becomes more expensive. Only when the cost of energy really starts too impact on people’s lives, will they start to pay attention. Luckily, that’s the direction energy prices have been going for some time now!

You should follow me on Twitter here

Photo credit Tom Raftery

To go along with the data center energy efficiency posts I have been writing in the last few weeks, I talked to RF Code earlier this week to find out what they have been up to recently. RF Code make wireless sensing devices which are proving quite popular lately in data centers.

I was speaking to Chad Riseling, RF Code’s VP of Worldwide Sales and he told me that a large portion of their 2010 growth came from their wireless environmental monitoring solutions. RF Code has wireless tags to monitor humidity, temperature, leak detection, PDU and CDU power usage (for certain vendors, as yet), (rack) door status and dry contact status.

The wireless sensors which RF Code sell, are roughly the size of a box of matches, they run off a battery which is rated to last around three years (and which starts alerting you about low battery status three months before the battery is depleted) and they have a range of mountings, including a peel and stick option, to facilitate easy deployment almost anywhere in a data center.

If you are wondering why a wireless solution is such a big deal, well think about the wireless internet network in your own home and how that has changed how you browse the net. Now you can access it anywhere in your home. Similarly, wireless sensors in a data center don’t need any extra cables to be rolled out for deployment and so can be installed quickly and relatively ubiquitously.

According to Chad, in a small 3,000 square foot data center, you could have up to 10,000 sensors being read by 3-4 readers and the data is handed off to the software stack, called Sensor Manager. Sensor manager can be used to track the data, or if companies have already invested in BMS or service management software, the data can be integrated with that.

One nice touch that RF Code have is that, in an homage to the Puppy Dog sales technique, they sell Starter Packs which contain sensors, readers and management software (enough to get you going, in other words) for as little as $2,995. If you are happy with the starter pack, you can simply buy more tags, readers, etc. to build out your solution.

Yesterday RF Code announced that they are launching a European Channel Program to grow beyond their current, predominantly US-based, market. Cool.

You should follow me on Twitter here

Photo copyright RF Code

If, like me, you have been using technology for a while now, you will be used to the constant release-by-release bloating of software. The first time I installed Excel it was version 2.2 and at the time it fit comfortable on a 1.4mb 3.5″ floppy disk – remember floppy disks?

In a pleasant bucking of that trend, SAP’s Holger Faulhaber told me in a recent call, that the latest version of their NetWeaver platform (v. 7.3) is a much leaner beast! While it is unlikely to fit on a floppy disk, it does have some significant performance wins, along with the simplified architecture and the functionality improvements you’d expect from an upgrade.

One of the main reasons for the improvements are SAP’s dropping of its two-stack approach in favour of a single Java stack for NetWeaver 7.3. This significantly reduces the amount of hardware which needs to be deployed and also because messages only need to be stored once, compared to 3-7 times previously, you get even more energy savings.

To demonstrate this, SAP carried out testing of their new NetWeaver Platform using the SAP Application Performance Standard (SAPS).

SAP defined a medium-sized landscape as being 37,500 SAPS for a NetWeaver Process Integration (PI) customer. Based on that they found that the savings potential for PI is in the region of:
- 60% less energy consumption or around 18,000 kWh/yr
- 16 tons of CO2 savings per landscape/year and
- €6.5k saving potential per landscape/year

The numbers for NetWeaver Portal 7.3 are for an SAP defined medium sized landscape of 30,000 SAPS. In that case, you see a savings potential of:
– 30% less energy consumption, 13,000 kWh/yr
– 6.5 tons od CO2 savings per landscape/yr and
– €2.6k saving potential/yr

While for NetWeaver Business Process Management 7.3 the potential savings for a 30,000 SAPS medium sized customer are:
– 57% less energy consumption, 24,000kWh/yr
– 12 tons of CO2 savings per landscape/yr and
– €6k saving potential per landscape/yr

The new software was tested on identical hardware to the previous version to rule out any efficiency gains from improved hardware according to Holger. He went on to mention that SAPs Quick Sizer [PDF] tool to help customers design their SAP landscape – has been updated for 7.3 so you don’t overspec your SAP installation.

“One of the main learning for SAP from this exercise is that I/O is expensive. For current tasks, it is nearly 3 times more expensive than utilising the CPU”, according to Holger. “Because of this we are now telling developers not write to locks, don’t write to the db – dropping locks increases performance” he continued. It will be interesting to see if SAP can move developers towards writing more energy efficient code.

This is a big potential win for SAP customers. Now they can gain significant performance and energy gains with a simple software upgrade, as opposed to having to buy any new hardware.It’d be great to see more companies adopting this type of approach to software development.

You should follow me on Twitter here

Photo credit Marjan Krebelj

If you have been following this blog, you’ll know I have been profiling Data Center efficiency companies over the last few weeks. This week I take a look at Sentilla.

I talked to Sentilla’s CTO and co-founder, Joe Polastre, the other day and Joe told me that Sentilla came out of Berkeley where they had been looking at data analytics problems around large, potentially incomplete or inaccurate, streaming datasets. The challenge was how to turn that into a complete picture of what’s going on so people could make business decisions.

Sentilla takes an industrial manufacturing approach to Data Centers – in manufacturing you have power going in one side, and products and (often) waste heat coming out the other. In the same way in data centers you have power going in one side and coming out the other side you have the product (compute cycles) and waste heat. To optimise your data center you need to get the maximum data/compute (product) output with the minimum power in and the least waste heat generated. Sentilla thinks of data centers, as data factories!

Unlike most of the data center people I have talked to, Sentilla don’t talk so much about energy savings. Instead they emphasise maximising performance – getting the most out of your existing data centers, your existing storage, your existing servers, your existing power supply. By far the greatest saving from deploying Sentilla, Joe claimed, is not from the energy savings. That pales in comparison to the capital deferment savings gained from being able to delay the building of extra data center facilities by however many years, he said.

So how does Sentilla help?

Well Sentilla analyses the energy profile of every asset in the data center, whether metered or not, and makes recommendations to improve the planning and management of data center operations. I highlighted the “whether metered or not” bit because this is an important differentiator for Sentilla – they have developed and patented what they call “virtual meters”. These are algorithms which look at the work that a device is doing, and based on models which Sentilla have built up, and measurements they have done, as well as some benchmarks which are out there, Sentilla computes how much power is being used by that equipment.

The reason this is so important is because the most inefficient equipment in the data center is not the new stuff (which is likely to already be metered) but the legacy devices. These are the ones which need to be most carefully managed, and the ones where the greatest performance gains for the data center can be made. And because Sentilla can pull usage information from management databases like Tivoli, it means the Sentilla doesn’t need to poll every piece of equipment in the data center (with the increased network traffic and data that would generate).

Also, because Sentilla has its virtual meters, it is a software-only product and can therefore be rolled out very quickly.

The other nice feature Sentilla has is that it can identify the energy utilisation of virtualised servers. This is important because with the increasing ease of deployment of virtual servers, under-utilised VM’s and VM clutter are starting to become issues for data centers. VM clutter isn’t just an issue for energy reasons – there are also implications for software licensing, maintenance and SLA requirements.

I asked Joe about whether Sentilla is a SaaS product and he said that while they have a SaaS version of the product, so far most of Sentilla’s clients prefer to keep their data in-house and they haven’t gone for the SaaS option.

Finally I asked about pricing and Joe said that Sentilla is priced on a subscription basis and, apparently, it is priced such that for any modest sized data center, for every $1 you put into Sentilla, you get $2 back. Or put another way, Joe said, deploying Sentilla will generally mean that you reclaim around 18-20% of your power capacity.

Disclosure: Sentilla are a client (but this post is not part of their client engagement)

You should follow me on Twitter here

Photo credit The Planet

If you’ve been following this blog in the last couple of weeks you’ll have noticed that I have profiled a couple of data centre energy management companies – well, today it is the turn of Power Assure.

The last time I talked to Power Assure was two years ago and they were still very early stage. At that time I talked to co-founder and CTO, Clemens Pfeiffer, this time I spoke with Power Assure’s President and CEO, Brad Wurtz.

The spin that Power Assure put on their energy management software is that, not only do they offer their Dynamic Power Management solution which provides realtime monitoring and analytics of power consumption across multiple sites, but their Dynamic Power Optimization application automatically reduces power consumption.

How does it do that?

Well, according to Brad, clients put an appliance in each of the data centres they are interested in optimising (Power Assure’s target customer base are large organisations with multiple data centres – government, financial services, healthcare, insurance, telco’s, etc.). The appliance uses the management network to gather data – data may come from devices (servers, PDU’s, UPS’s, chillers, etc.) directly, or more frequently, it gathers data directly from multiple existing databases (i.e. a Tivoli db, a BMS, an existing power monitoring system, and/or inventory system) and performs Data Centre analytics on those data.

The optimisation module links into existing system management software to measures and track energy demand on a per applications basis in realtime. It then calculates the amount of compute capacity required to meet the service level agreements of that application and adds a little bit of headroom. From the compute it knows the number of servers needed, so it communicates with the load balancer (or hypervisor, depending on the data centre’s infrastructure) and adjusts the size of the server pool to meet the required demand.

Servers removed from the pool can be either power capped or put in sleep mode. As demand increases the servers can be brought fully online and the load balancer re-balanced so the enlarged pool can meet the new level of demand. This is the opposite of the smart grid demand response concept – this is supply-side management – matching your energy consumption (supply to the demand for compute resources).

A partnership with Intel means that future versions will be able to turn off and on individual components or cores to more precisely control power usage.

The software is agentless and interestingly, given the customer profile Brad outlined (pharmas, financial institutions, governments, etc.), customers log in to view and manage their power consumption data because it is SaaS delivered.

The two case studies on their site make for interesting reading and show reductions in power consumption from 56% – 68% which are not to be sneezed at.

The one client referred to in the call is NASA and Power Assure are involved in a data centre consolidation program with them. Based on the work they have done with Power Assure, Brad informed me that NASA now expects to be able to consolidate their current 75 Data Centres significantly. That’ll make a fascinating case study!

You should follow me on Twitter here

Photo credit cbowns

I had a nice catch-up call with Mike Holt the other day – Mike is Semitech Semiconductor’s VP Sales and Marketing and we had previously talked last August.

Mike was bringing me up to speed on what has been going on with Semitech in the last five months. Semitech make semiconductor chips specially designed for power line communications (PLC). Power line communications is essentially the use of electrical cables to transmit data.

Power line communications is receiving a lot of interest these days because of the current buzz (bad pun, sorry) around Smart Grids. Every smart meter is, by definition, connected to an electricity distribution network – if this network can be used to send and receive information, it saves having to roll out a separate infrastructure for your smart grid communications.

During our call, Mike made me aware of a number of bits of Semitech news (they’ve expanded their sales team, they had new product announcements and they closed a new funding round) but by far the more interesting part of the chat, for me, was when Mike started to tell me of the new markets that are expressing an interest in PLC.

According to Mike, there is now a growing interest in PLC coming from the urban lighting sector to help with maintenance and control of lighting. Mike gave me the example of Munich. Apparently there are five people employed in the city whose job it is to drive around the city at night spotting burnt out lights – with a PLC solution, you can receive an automatic notification as soon as your sodium lamps burn out and you get early indications that an LED is about to burn out allowing you to switch it out before it goes dark.

LED’s by themselves save large amounts of money (this is very important when urban lighting makes up to 40% of a cities energy budget) but, as we reported last year, LED’s when they are connected to an intelligent control system can save up to 87% on energy costs!

Other new markets for power line communications Mike mentioned were for remote management of solar panels on remote solar farms and seemingly there is a growing interest in PLC in the home automation field – which makes a lot of sense too.

Power line communications is not without its challenges – electrical networks can be ‘noisy’ and be prone to attenuation but as companies like Semitech become better at dealing with those issues, it is going to be fascinating to see just where PLC takes us.

Photo credit Ruud Hein