It’s time for IT to have its own ERP and CRM, according to HP. That’s what the business technology optimization tools it’s developed are for. Today that’s the product name, but it’s such a good phrase that Tom Hogan, the senior VP and global manager of HP software (and, since he bought EDS, services), is thinking of coming up with some other name so he can keep it as a description. It’s meant to make you think of business process optimization, where you discover the way your company does everything has been wrong all along and it’s going to take an expensive stint of consultancy to fix it.

The way most companies do IT is hand to mouth, piecemeal and manually intensive. Imagine a car assembly plant that hand-wrote scripts to control the robots every time a new part had to be made. If IT departments really were the cobbler’s children they’re often compared to, they’d have been barefoot so long they’d be placed in foster care. Most IT departments can’t add as much value to the business as the technology companies tell us their technology can deliver and that’s not just the gap between hype and reality. In a survey that the Economist Intelligence Unit just carried out for HP, an “overwhelming majority of both CEOs and CIOs” believe that “technology is integral to the success of their company” and 88% of CEOs and 90% of CIOs say they “share similar visions for how technology can deliver business outcomes at their company” - which is close enough that they must be at least on the same page. So what’s the problem? As usual, money.

The 70-80% of the budget most IT departments have been spending on maintenance rather than innovation has only just gone down to 60% according to a new survey in CIO magazine. If you’re doing really, really well, you’re only spending 35% keeping the lights on and if you’re supremely ambitious you want to get that down to 20%; maybe that’s the 7-10% of companies that Tom Hogan guestimates have already got their IT automated.

After all, why should the majority of your budget go on doing the same thing over and over again so that the business can stay where it was last month? You should have the routine automated; it will stop you losing staff to sheer and utter boredom too. But once you plug everything in enough to automate it and track it  - not just in terms of transaction throughput or whether the next laptop you buy will trigger a discount if you buy it from vendor H instead of vendor I but who has a problem, who’s about to have a problem, who’s working on it right now and who has already spent how long fixing what on what annual salary -  you can also start to put a value on it.

The next time you ask for a new server, you can show how much it’s costing to chug along on the old one - and what difference you’ve made with the last budget you were allocated. You can show which SLAs the IT team has met and what that means in cold hard cash. When you get asked to do too much with too little, instead of sucking your teeth like a builder you can say exactly what it will cost to do properly and what they can get for the money they want to spend. You can even tell the business what projects are a bad idea before you plug in a single server.

It’s not rocket science; a good project has high value to the business, a small cost to implement and a low budget requirement. The IT department spends the whole year putting in systems that let the rest of the business evaluate business ideas and existing services in those terms; isn’t it about time you had that for yourself too?
-Mary

A comment from Wyse popped into my inbox the other day, criticising the government for using desktop PCs instead of thin clients which are “inherently more energy efficient” (surprise surprise).

David Angwin, director of marketing for EMEA, claimed that “thin client computers give users exactly the same applications and performance as a PC and run on as little a tenth of the electricity.” Certainly, Wyse is one of the few thin client manufacturers who can claim to support a wide range of applications; I know one financial company who had to replace the first batch of thin clients they tried with Wyse kit almost within the week because the others couldn’t cope with video clips. But is that power figure the whole story?

Earlier in the year I was talking to Barry Goodall at the Royal Borough of Kensington and Chelsea. He’s spent a lot of time and effort greening the council’s IT and although he’s a big fan of server virtualisation, he has a much less positive view of the green credentials of thin clients after he disproved the figures in a Frauenhofer Institute report on green computing. “The report said we could save million of pounds by using thin clients, so we were quite interested in this! We looked at some of the details and things leapt out at us; in particular the power consumption of PCs was markedly higher than ours - we use Dell desktops.”

He was checking his Dells anyway, because Dell was claiming upgrading to model 745s would save as much energy as changing from CRT to LCD screens. “We have an electricity monitoring gadget from Maplin which I highly recommend: don’t trust anything the manufacturers tell you! It’s very easy and you need to measure it yourself.” His measurements showed the model 745s used the same 60 Watts of power as the Dell kit he already had; Dell’s 45 Watt figure assumed energy management features that weren’t turned on by default. “Energy saving features in the BIOS count for nothing unless you enable hibernation in Windows!”

But 60 Watts or 45, it was still a far cry from the 120 Watts that Frauenhofer was assuming for a desktop PC. That’s what you’d expect from a top-end home machine with a high-power graphics card for gaming; business desktops are rather more frugal.

That wasn’t the only place he felt the sums didn’t add up. “Although the report said in the text that they had accounted for PCs being turned on maybe ten hours a day, terminal servers are typically running 24/7. If you tot up the number of hours people work out of the year, even though it feels like you work all the hours God sends, it’s actually about 2,200 and the figures in their tables hadn’t taken that into account. When we plugged in the correct figures they supported the opposite arguments; with the number of clients per server they assumed, it was more expensive in terms of CO2 than a typical fat client environment. Thin client can be more energy efficient but you need to be clever and turn some servers off when demand is low; you have to be monitoring the workload so you can turn some servers off overnight and come the morning, start turning them back on again - though you’re running a little bit of a risk that maybe one or two servers won’t start up and you’ll struggle a little.”

When I talked to Jon Stewart at Cisco about security trends recently, he slipped in a few network arguments (as you’d expect from a network company). “I have a feeling [that] what you’re going to end up seeing is very thin, light application suites that are endpoint based and a very rich experience using massive network build out. It’s already started to happen; definitely BT has gone down this route. You’re basically saying the end point is going to matter less at a computational level. The display and the keyboard and the system that you interact with, is the most valuable. Think about Lufthansa going to wireless on their planes, they’re trying to solve the inability to do work when you’re mobile. Everything about handset mobility, you’re trying to solve work when you’re mobile. But each time it happens, less and less computational necessity exists on the device - you’re just getting the service on the device.”

But do we care less and less about devices? Again, you’d expect Steve Ballmer to favour the PC, but he told his audience at the Partner Conference that actually, all the devices that are getting attention are fat (we just need to make them easy too). “It’s ironic, people talk a lot about whether people want thin clients. And I don’t deny people want reduced cost, and complexity of management. I think we’re all hearing that from our customers. But people don’t want to really give up the richness and capabilities of a rich client. We even see that in phones. What’s going on in phones today? Phones are actually getting richer. That’s what Windows Mobile is, that’s what the iPhone is, that’s what Symbian is, that’s what Android is: all of these things are getting richer, and Windows PCs will be the richest, most capable device that most people ever own.”

Chatting with Peter Biddle, ex of Microsoft and now at UK enterprise social networking startup Trampoline, he suggested that as usual, what matters is both the device and the network. “Think about it; when did you last do any useful work without being online?”
-Mary

With Intel’s 40th birthday on the horizon (and with it the 40th anniversary of the microprocessor), Intel’s Pat Gelsinger took a few minutes yesterday to ruminate on the past, present and future - and to take a few questions.

Beginning with a look back to the i386, and the shift from 16 to 32-bit computing, Gelsinger pointed to a time of technical and industry transition, much like today. It was the point where Compaq moved ahead of IBM, and Windows and Microsoft began to shape the software industry. We’re in the middle of another shift at the moment, what Gelsinger called the “third era of Moore’s Law”.

The first era was the age of invention, with the second concentrating on scale and manufacturing. Gelsinger calls the third era “The right hand turn”, where the industry starts to concentrate on energy efficiency. He went on to describe the industry’s success as resulting from “the power of compatibility”, where compatible software means that each generation of silicon can inherit the work of the entire industry (with just a little recompile along the way). There have been plenty of changes in Microprocessor design, purely by increasing numbers of transistors - the power controller on Intel’s Nehalem processors is bigger than Gelsinger’s first processor. There’s a sheer complexity to these machines, which Gelsinger described as “the most advanced things ever built”.

That’s the past and today, so what about tomorrow? Intel reckons on having 10 years of visibility into the future of silicon. Gelsinger described silicon as “the scaffolding for half the periodic table”. The future will be much the same, even if it’s based on silicon nanowires and spintronics. The first big change will be in just a couple of years, with the shift to 450mm wafers. The investment this requires will be huge, and Intel expects this to trigger a wave of industry consolidations - just to help pay for the new fabs.

Gelsinger also sees Intel’s IA architecture as a key differentiator between it and the rest of the industry. As multicore systems become more and more common, and as IA scales up to teraflop terascale systems and down to milliwatts, software will be compatible between all the different versions of the architecture. There of course will be different languages and libraries (especially for parallel processing systems), but code will be portable.

The result will be what Gelsinger calls an “AE724″ world. Bill Gates’ vision was a computer on every desk and in every home, Intel’s is much more ambitious. It’s a world where everyone has access to the Internet, with computing embedded into the environment and the infrastructure - everywhere you can imagine. It’s certainly a big picture - and one that will mean a shift in the way we develop applications and in how we design networks and data centres.

We blogged about GPU-based computing last week, and Gelsinger was asked about Intel’s response to NVIDIA’s CUDA and AMD’s CTM. Describing CUDA as “an interesting footnote in the history of computing”, Gelsinger talked about GPU computing as a cool idea that required a new programming model. He felt that this would be hard to deal with compared to general purpose computing techniques, and suggested that Intel’s massively multicore Larabee would be the right answer in the long term.

It’s true the microprocessor and the software stack make a huge difference. I probably wouldn’t have dialed in to the conference call if Skype didn’t connect to US 1-800 numbers for free from anywhere in the world. Whether the future’s all Intel is another question. IA is an important architecture but there’s still space for low power alternatives like ARM, or for specialised co-processors from the likes of Toshiba, Azul, AMD and NVIDIA. General purpose silicon is just one way of working - and if you’re prepared to target a specific niche there’s still plenty of scope to make a very healthy profit with specialised silicon.

–Simon

(or “The return of the co-processor”)

The white heat of technology in the 1980s was focussed on the BBC Micro. Not only was it the heftiest 8-bit machines around, its open bus made it possible to add more processing power. With everything from music machines to Z-80s running CP/M, the BBC Micro could share its keyboard with many different CPUs.

Those days are on their way back.

Last week Toshiba announced a new range of consumer notebook PCs. Like many of Toshiba’s systems they’re designed to be media players, and in a side swipe at BluRay, they now come with an upscaling DVD drive. That’s where the coprocessor magic comes in, as Toshiba is using a derivative of the same Cell processor in Sony’s PS3 to drive its imaging software. A quad core version of the Cell sits alongside a dual core intel processor, and it’s used to handle a range of processor intensive tasks - acting as a feed to the GPU that drives the screen. Not only does it upscale DVD streams (very impressively) it also can be used to handle file transcoding (so your movies end up on your iPhone that much quicker), and also works well as a way of quickly indexing images and video.

Focused on video, Toshiba’s co-processor is also taking advantage of bundled web cams for a limited form of gesture control. Stopping a film by holding up a hand is effective, as is using a clenched fist as an in air mouse. Bill Gates’ departure reaffirmed his belief in alternative user infterfaces, and this is one approach to delivering those new ways of working.

Co-processors aren’t just for flashy graphics. Back in the 1990s I was writing mathematical simulation software, and at one point I had some electro-thermal models running on one of the MOD’s Crays. It wasn’t just any old Cray - it also had a co-processor in the shape of an additional vector processing unit. That vector co-processor made short shrift of my arrays of partial differential equations. Its direct descendent is a lot closer than an MOD research facility.

In fact, if you’ve got an NVIDIA graphics card it’s right in your PC’s GPU.

Back in January we wrote about Tesla and CUDA, and NVIDIA updated us on the next generation of the Tesla hardware earlier this week. The new G10 Tesla systems are looking very impressive, and the CUDA parallel programming language extensions are now able to work with standard multicore PCs as well as NVIDIA’s GPUs.

Memory is important when you’re using co-processors, and you need a lot if you’re signal-processing seismic data. Tesla will now support 4GB of directly attached memory per GPU, so a quad-GPU system can work with 16GB of data at a time. The numbers look good - and using Folding at Home a single Tesla 10 comes in at more than 40 times faster than a standard CPU, and more than 6 times faster than a PS3. Other demonstrations showed significant savings in space and in cost - one finance house has reduced its annual costs 9 times, replacing a 600 CPU options valuation system with a handful of front-end CPUs and 12 Tesla GPUs.

Of course with Snow Leopard around the corner, one of the obvious questions was about Apple’s support for OpenCL. It turns out that CUDA is best thought of as a personality layer on top of NVIDIA’s parallel thread execution (PTX) hardware, and it produces device-specific assembly code. There’s no reason why other GPU programming environments can’t produce the same PTX code - but CUDA will remain NVIDIA’s own route to the GPU as a processing tool, and it will be adding support for additional languages beyond C and C++ with Fortran just around the corner.

The future of the co-processor seems assured, for now at least. It’s time for software companies to start taking notice and to deliver on the promise of additional power beyond the CPU.

–Simon

Every three seconds, HP sells a printer (two of them in Europe). That makes HP responsible for a lot of the 22 pages office workers print every day, half of which end up in the bin. All that paper and ink can make it hard to think of HP as particularly green, especially when Vyomesh Yoshi, the VP of the print and imaging division, talks about wanting to see more pages on HP printers.

Naturally enough, he doesn’t think it’s what much of a contradiction. “We are in the printing business; we don’t want customers to not print. We have to make sure they use it, but also make sure they use it effectively. We want to make sure every printer they buy from HP has lower energy consumption than any other printer. ” Make sure printers turn on “instantly” and people will be happier to turn them off; use the WebJet admin software to turn printers off at the weekend and you’ll save even more energy. “Make duplex printing the default and you can save a tremendous amount,” he says.

And if you think we throw away a lot of paper in offices, 20% of newspapers are discarded, as are 40% of books and 20-30% of marketing bumph. People print too many copies that go out of date, because of the setup charges on offset printing. Naturally again, HP has a solution; customised on-demand printing for everything from wine labels to out of print books.

HP is also pushing green ideas for the data centre like running the air conditioning four degrees higher by blowing cold air directly into the blades. The air coming out the back of the blades is a lot hotter - more like a sauna - but heat behind the blade doesn’t matter so much. But if CEO Mark Hurd is right to predict that data centres will use 50% less energy soon, it’s going to take more than hot air.

The reason smart cooling works is that HP puts sensors on each rack to make sure the air is only as cold as it needs to be. Without those, says HP fellow Chandrakant Patel , a home air conditioning system is more sophisticated what’s in most data centres. The next step is to use optical interconnects and lasers to replace copper data cables - which saves the 20% of your energy that’s heating and cooling the copper. More likely now HP Labs has come up with a photodetector so sensitive it works as a solar cell.

But the carbon footprint of a data centre includes the CO2 from the concrete used to build it, and the manufacturing and transportation of everything from the blades to the carpets. Really reducing that means calculating it and Patel is working on a framework to cover technology in general. That would measure the true energy cost, down to what it takes to deal with the fertilizer runoff from the fields growing the corn that’s made into the ethanol that goes into the biodiesel that drives the backup generator. It’s a huge undertaking, but it strikes me as more likely to help than maintaining that video conferencing can solve the problem by taking cars off the roads (hint: maybe not if that means more data centres to run the video conferencing).

Microsoft makes a lot of noise. The company holds dozens of conferences, broadcasts its ambitions in every market from mobile phones to data centres to next-generation TV, goes on a buying spree, gets taken to court by everyone from Novell to the EU. HP also makes acquisitions and has ambitions in a lot of markets and employs over twice as many people as Microsoft, but it doesn’t make nearly as big a splash in the industry, for some reason. It’s not for lack of success. Microsoft boasts of the 31 million Windows Mobile phones it’s sold; HP boasts that eight out of every ten text messages are sent using HP technology (inside the mobile operators rather than in your hand).

Wherever Microsoft is, HP is there too (from mobile phones to data centres to next-generation TV); in almost all cases, selling infrastructure rather than competing software. The exception is system management and when a dealer asked Mark Hurd this week why he’d asked Steve Ballmer along to the event where HP was sharing what it wanted dealers to get excited about this year, Hurd pointed out that even there HP takes a wider view. “Microsoft is very focussed about managing Windows environments and Microsoft environments. That’s what’s important to them. And it makes sense for them to be the best in the world at that. We have to be the best in the world at managing heterogeneous environments; we have to be able to take an IBM environment, a Linux environment, a ‘insert name here’ environment and be the best. Microsoft has to optimise simplicity of management of the Microsoft environment. We don’t believe the world will ever be exclusive to Microsoft.”

Ballmer wasn’t offended by that and used the broader view line himself, emphasising all the places from printers to blade servers where the two companies collaborate. He’s also banking on HP to put some style and sparkle back in a PC marketplace that can look lacklustre compared to Apple products that look good even when they can’t compete on features. “We’ve got a lot of work we’re doing on the future of the PC and what that looks like; driving down price, driving up features, driving more excitement. Certainly neither Microsoft or HP likes the shots we’ve been taking with Apple’s adverts and the blah blah blah… On the consumer side there’s so much opportunity today, we can add value to business productivity ; we’re stepping back to remind ourselves what we can do.”

Adding features matters more than driving the price down to get businesses to keep buying new PCs every three years rather than pushing older machines to last a decade and that’s where HP Labs comes in.
HP cares about research, but it’s a means to an end – solving problems by creating products and services – rather than pure knowledge. For pure knowledge you stay in academe; although Microsoft’s Bill Buxton points out that he left academic research for commercial when he was asked to write business plans rather than papers. Mark Hurd is totting up his R&D dollars but it’s not the cost he complains about; “We spend 4.2 billion in R&D to get the best products and services and then only go after half of the market.”

Rather than squeezing the research budget specifically, he’s leveraging it and putting more emphasis on the D than the R. HP Labs looks five to ten years ahead, but it also collaborates with engineers to create products. Microsoft Research uses a mix of technology transfer and researchers who move across to product development groups to shepherd their project into the commercial world. The really important thing is that they can always go back to research afterwards; their job is guaranteed to be there.

HP takes a different approach. Phil McKinney is the CTO of the personal systems group – everything from iPaqs to the Blackbird gaming system that’s selling to developers to the 2710p tablet PC we both use to the shiny and cute new 2133 Min-Note UMPC (which manages to achieve Apple levels of desirability despite the Via C7-M processor – which might make it even more like an Apple product). But he also runs the Innovation Programme Office and I don’t think it’s named IPO by accident; it’s certainly about taking things public.
The way it works is that a team from the IPO works side by side with the researchers (quite literally; they sit at the next desk). For 12-18 months the two teams work together; the researchers carry on researching, the designers build products and gradually the researchers do less and less and the designers do more and more. Then one day the designers have learned everything the researchers have found out and they spend six months running that into the final product.

There are 28 products in the pipeline with the IPO, coming out two a year – which means starting with 1,800 pipedreams that get whittled down to 200 ‘workable’ ideas. Blackbird was the first, cherry-picking existing HP technologies like blade cooling and push-fit hard drives. The new DreamColor screens are the second. These are LCD screens with colour accurate enough to satisfy DreamWorks and there’s a 30” screen on the way. And there’s a team in HP Labs right now, sitting next to the data centre than rendered Shrek 2, working on the next project. Odds are, it will be something that Microsoft will be interested in…

How do you know you’ve been hacked? You may have a suspicion that someone’s inside your network, but if your log files don’t show anything, don’t assume that your systems are secure. The bad guys know all about standard compouter forensic techniques and have toolkits full of techniques and programs to cover up their traces. The computer security team at Verizon are finding that anti-forensics are used in more than 2/3 of intrusions.

One of the most common techniques is data wiping, used to reduce the evidence available to security analysts. Used in only 18% of cases in 1998, things are very different today, with data wiping used in 80% of cases. The popularity of data wiping can be seen by the sheer number of tools available on black file sites - with more available than all the other types of anti-forensic tools combined.

Luckily for us data wiping is not perfect, and even the best tools leave some files behind - especially when files have been locked or are still in use. It’s a good idea to think outside the box - often literally. Perhaps a backup has traces of the bad guy at work, or there may be traces of his tools and actions on a clustered storage array somewhere else in your data centre. And of course there’s the old forensic stand-by: running memory. A memory dump can show traces of running programs in old page files.

The next most popular technique is data corruption, closely followed by data injection. The aim here is to hide from your logging tools - or even make your log files unreliable. One technique is very simple, with intruders resetting system clocks to create a whole new log that can be deleted when they leave. If there unexpected holes in log files, there’s a distinct possibility that someone is changing your system clock. More complex techniques use tools to corrupt log files to cover up attacks, or to edit out an attackers actions.

One case Verizon worked on was a retail customer that was seeing unexpected charges on its credit card system. Nothing was found in the logs, but the Verizon forensic team was sure that something was happening, so they began to monitor the system.

A few days later a tripwire was triggered, and they were able to watch (and screen capture) someone from the credit processing vendor coming in to the network on a trusted connection. The attacker first changed the system clock to hide their actions, and then using the debug mode in the credit card software to steal transaction data. The security team watched the attacker tidy up after themselves, deleting the debug files. Finally the attacker reset the system clock and edited the system logs to replace their external IP address with an internal one. They’d only made one mistake, which was how the security forensics team was convinced that there was an attacker.

What was it?

The internal IP address they were using wasn’t actually assigned to anything.

It’s clues like that that you need to look out for when assessing a system to see if it’s been compromised. You know what makes your network tick, what addresses are in use, and what your system logs should look like. Vigilance is the only way you’re going to be secure.

In the immortal words of Hill Street Blues: Be careful out there.

– Simon

Sometimes weeks have a theme. This last week’s was most definitely virtualisation.

Cannes in February is a refreshing change from an English winter, and it’s where HP were announcing their latest storage virtualisation platform - along with servers that come with VMware’s ESX 3i hypervisor built-in. Getting the right physical infrastructure for your virtualised server farm is becoming increasingly important, along with beign able to deploy your virtualised images quickly.

That’s where ESX 3i comes in handy - it’s a 32MB hypervisor that doesn’t need an OS. Boot your server, connect to the hypervisor from one of VMware’s handy management tools, and you’re ready to configure the hard disk and deploy all the virtual machine images you need (and if you’re using BEA’s Liquid VM thin Java servers that can be a lot!). Now that most of the major industry standard server vendors support ESX 3i, with hardware shipping from IBN, Dell and Fujitsu as well as HP, it’s going to be easy to quickly add new compute resources to a virtual infrastructure. All you’ll need to do is order the appropriate server from your usual vendor, shove it in the rack (or the blade host), and a few mouse clicks later you’ve got a server.

Microsoft is going to have to do a lot to compete with this. Its Hyper-V VM is still several months away - and it will still need its own partition to run and manage the rest of the virtual machines running on your server hardware. Sun’s xVM will have similar issues, as will the open-source Xen hypervisor.

HP’s launch was in its Sophia Antipolis offices, as VMworld Europe was just down the road. With more than 4500 attendees, it was definitely the place to be if you were running a virtual infrastructure. BT’s plans to roll out a service oriented virtual network with global load balancing was a benchmark for the maturity of virtual infrastructures, and a fascinating look at how businesses can encourage the move to virtualisation. BT’s decision to make physical server implementations subject to a rigorous review process and hefty chargebacks is intended to make this an economic decision - with virtualisation the clear winner on ease of deployment and lower costs.

At Vmworld Europe the thin client was one of the elephants in the room. While the server products got the stage time, client virtualization got a set of ropy demonstrations which were, to say the least, confusing. Conflating VMware’s impressive VDI virtualised desktop tools with managed desktop virtual machines, CEO Diane Greene demonstrated how virtual machines could be deployed to desktop PCs, and how thin client applications could be used offline and on the road. Given that presentation it would be easy to confuse two very different ways of managing virtualised desktop environments.

Microsoft made a lot more of client virtualisation at its 2008 server wave launch last Wednesday. That’s not surprising, especially when you consider that its big server success story of the moment is its relatively recent acquisition SoftGrid. Delivering applications over the network is a powerful way of controlling user desktops, and reducing your support costs. SoftGrid’s impressive sales figures are even more impressive, when you realise it’s only available through Microsoft’s volume licensing programme.

Perhaps the ideal infrastructure is a hybrid. VMware virtual servers hosting enterprise applications, with SoftGrid -wrapped applications streaming from the server network on to desktop PCs. The PCs themselves might be thin clients fed by Citrix’s tools running on an array of desktop blades somewhere in your data centre. It’s all a blast from the past - the mainframe is back.

This time, however, it’s an ever-growing array of industry standard servers hosting a virtual infrastructure, while applications are delivered to not green screens, but thin client devices with HD quality LCD panels. It’s a brave new enterprise IT world out there.

–Simon 

Mary and I are a small business, and we rely on our workhorse of a server. Currently running Microsoft’s Small Business Server, it sits in the corner of the office and looks after our mail and our files. Sadly, though, it’s getting close to retirement. A solitary Athlon isn’t really up to today’s workloads, and many of the features we want from Exchange are only in the latest release - which is 64-bit only.

Our new server has arrived, and it’s surprising what a few years have done to the SMB server market. We’d had to build our original server ourselves, but this time we’ve ended up with a dual-core Xeon system from HP. I’d been delaying purchasing a new server as I was expecting to pay through the nose for my hardware, but when I started looking at current prices I found I couldn’t have been more wrong.

That HP server? With 512MB of RAM and a 160GB hard drive, it cost us less than £180. I was able to buy 2GB of RAM for around £50, and two 500GB hard drives for under £100 all-in. That’s around £320 for a machine that’ll meet most of our business needs for the next four or five years. Fititing it all together was easy enough, and the server was powered up and ready to go in under an hour. Once it’s all configured I’ll be adding an eSATA card for speedy backups.

Breaking it down a bit, it all means that our server will cost us around £80 a year, before software and maintenance costs. Sure, it’s not a 1U rack mounted system, or a set of blades, but it’s a sturdy well engineered box with a processor that’s hefty enough for most needs.

Hardware is the easy bit. Software is a lot harder. I’ve now got to plan just how I’ll migrate files and mailboxes to the new machine - and how I’ll handle decommissioning the old box. If anyone’s got any good tips, I’ll be glad to hear them!

 –Simon

The barracuda is the wolf of the sea, a slim silver dart that hunts in deadly packs. It’s perhaps not surprising that NVIDIA has taken part of its name for its GPU-based supercomputing tools.

On a recent trip to the US, Mary and I met up with some of the folk behind CUDA at NVIDIA’s Sunnyvale headquarters. It was a fascinating conversation - if only because I used to write scientific computing software, and something like CUDA would have sped up my work massively. When a problem takes days to solve, something using something like CUDA to accelerate processing makes a lot of sense.

Prior to CUDA, NVIDIA had tried to use GPUs for compute, but had run into architectural problems. Things changed with their series 8 GPU, which was very different to anything they’d built before, being designed for compute as well as graphics. That’s lead to some tradeoffs - there’s silicon on the GPUs that’s unused when it’s used as an accelerator (and vice versa). However NVIDIA makes so many chips, there’s not really any financial issue, it all comes out of the economies of scale.

CUDA is more than just a set of chips - it’s a language framework for working with GPUs, that can andle both sequential and parallel code together. Developers don’t need to learn anything you, and the framework gives programmers explicit - and simple - interfaces for running parallel code on NVIDIAs GPUs. There is a long term goal of providing tools for automating parallelism, but at this point you still need to work out what code can be parallelised yourself. The result is code that’s very simple with much less code, as CUDA handles repetitive calculations for you.

Simplicity comes from the hardware as well, as it manages threads for you. All you need to do is define the tasks the GPU will handle, and manage their interactions. The GPU then runs the calculations over the data, with groups of processors on different functions at the same time. As RAM is directly attached to the GPU there’s no need to use the PC’s own memory for caching data.

The numbers coming out of CUDA are impressive. Working with the VMD/NAMD molecular dynamics tools researchers at the University of Illinois have seen a 240X speed-up in the VMD ion placement tool, and an 8 to 12X speed up in NAMD. With an eye on greener computing, they’re also finding that CUDA gives them 1W/Gflop!

If you want this sort of power for your applications (and it’s remarkably suitable for large financial applications) you can by NVIDIA’s Tesla systems. There are work station versions, along with deskside offload processors. However the version we were most impressed with comes as a 1U rack mount unit, containing 4 GPUs. Connected to a PC or a server via 5 Gbps PCI-Express connections this is the way to give your data centre applications a significant speed up, with significantly lower power requirements.

While Tesla may not yet meet NVIDIA’s aim of providing a Teraflop in a 1U unit, it certainly speeds things up. Oxford University researchers have used it to get a 149X speed up LIBOR risk analysis for an 89X improvement on performance/Watt. That’s a good deal in anyone’s book - especially if you’re working with today’s fractious financial markets.

Add one to my list for the IT Santa!

–Simon