2009: The year of the ‘Enterprise Netbook’ (and other cloud clients)

There’s exciting times ahead in 2009 for the enterprise – the maturity of cloud computing services like Google Apps and [Sales]Force.com is finally enabling them to reach to the holy grail of personal computing: reliable, cheap, connected clients, and the OEMs are finally ready to supply the (now rapidly growing) demand.

It’s not surprising that this shift is starting with the netbook (‘a light-weight, low-cost, energy-efficient, highly portable laptop‘), but expect similar devices (Update:Nettops‘) to start marching into the heartland of the desktop (including the enterprise desktop) before long too. I expect PCs and laptops as we know them today to go the way of the dodo within one or two 3-year refresh cycles (maybe 3 for the laggards), and that we’ll be able to buy these devices by the kilo from chinese manufacturers like Lenovo before long.

And why wouldn’t this be the case? On offer are smaller, lighter, “sufficiently capable” devices with few or no moving parts (and thus significantly longer lifetimes – 10 year MTBFs are not unheard of), which boast extended battery lives (8+ hours – enough for a full working day), all at a fraction of the price; order of magnitude cost reductions provide significant incentive to migrate.

The performance race is now well and truly over and chip manufacturers are scrambling to change tack so as to deliver adequate computing power as efficiently and economically as possible. The megahertz myth turned the whole thing into a farce anyway as it has long since been virtually impossible to compare chips by specification, short of conducting application-specific benchmarks; what’s more important; more cores? more cache? more cycles? Or less cost, less energy use, longer battery life and longer lifetimes?

By offloading a lot of the extra work to the server side (typically large, next generation datacenters loaded with commodity PC hardware) the thin client is finally becoming a mass market reality. You’d be forgiven for drawing parallels to the mainframe era, but we had good reason at the time to move to client-server (and good reason now to return).

Here’s what the specs for what I call an ‘Enterprise Netbook’ ought to look like by the second half of this year:

  • 12-14″ display
  • ~2GHz ‘green’ processor
  • 1-2Gb RAM (browsers tend to be heavy on the RAM which is now relatively cheap)
  • 4-40Gb Solid State Disk (SSD)
  • USB-only (‘legacy free’) connectivity
  • Wireless networking (various combinations of WiFi, 3G, etc.)
  • 6-8+ hour battery life
  • 3-5 year warranties

The key attribute missing from today’s generation of consumer devices like the eeePC, MSI Wind and equivalents from the incumbents (Dell, HP, etc.) is a suitable screen, thanks to a 10 inch display ‘glass ceiling’. This is no accident as on one hand Intel want it that way (and are said to be coercing OEMs to ‘keep a lid on it’ at 10.2 inches) and on the other they’re busy bashing the formfactor as being “fine for an hour“. Meanwhile AMD are ignoring it altogether and NVIDIA and VIA gave up on their collaboration last year (although both NVIDIA and VIA are now represented independently). Surely it should have come as no surprise that it was not (yet) going to be possible to forge a market for a third Internet device between laptops and mobiles when there is already some contention between the two.

All is not lost though as Intel’s Atom processor is starting to make its way into enterprise friendly 12-14″ devices (in my opinion Apple have found the sweet spot at 13.3″ with the MacBook line) and is already in production in consumer devices like Dell’s Inspiron Mini 12. This comes at the expense of Core 2 Duo sales as evidenced by last quarter’s appalling results (90% down), but it’s arguably better to cannibalise a Core 2 sale to earn the margin of a Celeron than lose it altogether to a competitor.

So where will these sales go if not to Intel? I touched on this last week in the context of an Apple netbook (which is more a matter of ‘when’ than ‘if’) and I’ve also talked about a ‘cloud client‘ style device based on VIA Eden that I prototyped at Australian Online Solutions over 5 years ago (well before the services to support it existed). With mind-blowing performance being a non-issue (or even a battery-sucking legacy) in this type of device, the door opens up to a myriad chip vendors – almost all of which stem from the ARM family. These RISC chips are (in many ways) better, faster and cheaper than their CISC equivalents (eg Intel’s x86), especially in mobile settings where as recently as 2007 they held a staggering 98% market share.

Bearing in mind that there’s something like 3 mobile devices for every personal computer it’s no wonder then why Apple recently bought PA Semi and Google’s Android is on the platform too. Essentially all Google and Apple need to do to get in the Netbook game is scale up the size of their iPhone and Android designs. Update: Qualcomm are another player to keep an eye on in this space – they’re already up to their necks in Android and just purchased mobile graphics tech from AMD to deliver “more advanced products that redefine next-generation mobile user experiences”. They all have little to fear from Windows Mobile in the mean time (which is due to catch up later this year) nor Windows 7 (which while a breath of fresh air after Vista will be at a significant performance handicap), though they may well face stiff competition from a vibrant ecosystem of Netbook providers.

Of all the vendors, Microsoft will have the biggest challenge ahead (Update: Netbooks become the bane of Microsoft) in justifying the existence (and cost) of a full local operating system and application stack (plus the extra hardware required to support it). My rule of thumb is that commodity software (Windows, Office, etc.) should never cost more than half of the cost of the hardware (especially when it can be had for free), which is fine on a $2,000 PC but less so on a $200 netbook when even Windows XP Embedded costs $90. There is now also the spectre of new action from the EU relating to the bundling of Internet Explorer (last week found to be in violation of the antitrust laws) which could potentially see the single most important component shipped with competitors’ software or removed altogether. Let’s hope something like Midori will come along and save the day because healthy competition is good for all of us and Microsoft still have a lot to offer.

What all this means for enterprise is that there are both dramatic cost savings and a new way of working just around the corner, provided you’re ready with the infrastructure to support the new devices (e.g. cloud computing, application virtualisation, integration, security). We’re working with large enterprises (as we have been for over a decade now) on raising awareness, developing strategy, proving and piloting concepts, governing deployments and managing change and are as excited as they are about the future of cloud computing.