However, Joe advised some caution, per transaction costs can kill you (this is ironic given that the concept of “Free” – so popular with the Web 2.0 world – is based on infinitely low transaction costs). Remember, with cloud computing, you can’t touch the hardware or see it, so monitoring is more important, he mentioned Nagios and Monitis which are great packages for doing just that.

Another  big issue with Cloud Computing is the data problem – it takes  a while to get data up, even if you have lots of bandwidth. That means it takes a while to get your data out in a hurry. You also have to deal with high latency, compared to working with servers on site.

The dynamics of programing in the cloud is transaction based, so you need to be thinking dollars. Poor programming costs you real money, so unit tests need to include a cost value. Security also needs to be at the forefront of what you do. Joe described the “The Britney Spears Problem “- you can’t expose your end user data to risk. Client side encryption will be needed in the long-term, but is a usability challenge right now.

In summary cloud computing is a great adjunct, but don’t get rid of your main servers just yet. Be ready to deploy to any other grid at speed, so that you have a recovery plan. Also, watch those transaction costs.

Here is a talk on Cloud Computing by Larry Ellison (of Oracle – who are set to acquire Sun):

Related Posts

• Why the Cloud Need Standards
• A Cloud Computing Tour – London CloudCamp
• Rackspace Heads for the Clouds

The definition of ‘cloud’ computing and services is undergoing distortion on a daily basis, as vendors pile on the bandwagon, eager to ‘cloud-ify’ their wares. That’s just one of the prices of fame. Stepping away from that set of issues for a while, the critical reasons that cloud computing needs good standards can get a little lost if you get stuck down in the operation weeds. To quote myself, from the post on RedMonk, seeing as IT Knowledge exchange picked up on it:

It is clear that without standards of one kind or another (de-facto or from a recognised body), there won’t be a market, and without a market, the cloud is unlikely to thrive. The competition isn’t as much between cloud providers, as it is between cloud providers and internal IT organizations. Cloud providers need to keep that firmly in mind.

Let me draw some parallels with the early days of the broadband DSL market in the 90s. For many years the market for DSL was caught up in the battles between different hardware+software vendors, each out to prove that their version was better than the others. Eventually the market settled on a standard (initially ADSL). Equipment prices fell. On the face of it, not a good thing for the vendors, but actually revenues soared, as customers started to feel safe in making purchases, knowing that they were no longer reliant on a single vendor. If they needed to change providers they could. More importantly, tendering became a competitive process – something that enables a market to come alive.

The broadband market didn’t learn from this lesson for long. Soon the broadband providers, rather than the equipment vendors, were at each other’s throats, fighting for subscribers. It was the wrong battle. They fought over the 1% of users who had figured out what DSL was. Those early adopters had figured out that it would given them their downloads faster, and the providers ignored the 99% of the market who had no idea what “digital subscriber line” technology might do for them, or why on earth they might want it.

Back to the cloud. The current variations in storage architectures, database technologies and support of network features all contribute to blocking migration between services and inhibiting the market. To be fair, some of these things will end up being market wide innovations, or competitive differentiators, but for now most of them just stop development efforts being portable, and mean that there is no ‘back up’ option in the event that a provider fails. Sure, IBM are muscling in on the cloud action with their “cloud validation service“, and other kite mark efforts will attempt to validate individual providers as good fellows, but that won’t build a sustainable market. Today’s hero is tomorrow’s three day outage, and if you can’t move your application you are probably yesterday’s business.

It is unlikely that a single body will be able to standardize all of the different aspects involved in a full cloud service. Even if they could, standards bodies always struggle to keep up in rapidly developing space – I should know, I’ve suffered in enough of them. It may be that we settle on a lowest common denominator approach for now, and a core set of functionality emerges across providers. Alternatively, development tools might get smarter at providing a “write once, run anywhere” solution.

Regardless of how we get there, the cloud needs some form of standardization, so that a market can emerge and thrive. From my days working with VCs, I remember a phrase “if you haven’t got any competitors, you haven’t got a market, and you have a problem.” Right now no-one is tackling their biggest competitor in the space: Do nothing.

Related Posts

• Start ups – Get out of my Cloud
• A Cloud Computing Tour – London CloudCamp
• Rackspace Heads for the Clouds

The fact that a few hundred people came out on a cold wet London night to discuss Cloud Computing is a testimony to the growing fascination with all things cloud-like. One way or another, it is a phenomenon that will disrupt the status quo in software and service provision.

The evening kicked off with a set of speedy presentations: 5 minutes per presenter, brutally speedy. Their velocity and compactness made them hard to summarize, but let me try:

Simon Wardley, now with Canonical, was first on stage. Always an interesting presenter (I might have been biased by his love of ducks and the fact I that I was using his power supply, stealthily borrowed). Actually, the fact that I could use his power supply on my laptop is a testament to the importance of portability and the power of re-use, which was the theme of Simon’s presentation. It continued his XaaS theme from the last CloudCamp, making the case for the importance for standards. Simon also argued for the importance of transparency in cloud services (cf. recent happenings in the financial world as the result of opacity). Who owns and operates the equipment isn’t always obvious in cloud services. An immaculately timed 4 minute 59 second presentation.

Joe Bagley, CTO Europe of Quest Software, was next up, asking if the cloud is really green. Here we are, putting 20kw of power demand into a rack, then trying to keep it at  22′c. Less than 0.3% of the electricity used makes it through to the processor. Rather than looking at how much processing each watt gives, Joe argued that the new question people will ask is: for each service in use, how much energy is consumed? Server power consumption doesn’t scale linearly with use. An idle server still consumes huge amounts of power. Even so, virtualization doesn’t give the power savings many expect. Even worse, while VDI (desktop virtualization) is trendy, it moves even more processing into the datacentre and uses lots of RAM, and therefore even more energy. Then you throw away the old desktops… definitely not very green. Joe’s advice? Go green: take a hybrid approach, re-use and rethink. Oh, and make sure you get the (power) bills. (4 minutes 50 seconds).

Paul Watson, of Newcastle University, has been doing some work on behalf of Arjuna.  Throw away the cloud infrastructure, or get cloud from existing infrastructure? Paul suggested that you should create a private cloud, and spread the load. Put service agreements on sharing resource in place, and then share computing resources between different departments to make efficient use of (unused) computing power. Then link that to public cloud services to cope with peak demand. Paul’s model suggests the idea of many federated clouds (both public and private) – a kind of cloud of clouds. Interesting.

Duncan Johnston-Watt, Enigmatec CTO, performed a canned demo (using Elastic fox), showing how the cloud might be used for disaster recovery. Apart from his very correct use of the work “momentary”, the demo was a little pedestrian. Thankfully Cisco’s Phil Dean hopped on stage to tell us what CIOs want from the cloud. A reminder from Phil that CIOs are seeking to be business leaders. Accord to him CIOs like Cloud’s offer of consistency for all users, service simplicity (cost and 24×7 operation) and service orientation (for business agility). They don’t like the loss of control, risk management issues (business continuity and security) or migration and hybrid operation. I found myself thinking that perhaps CIOs don’t understand Cloud Computing yet, and are still wrapped around the axle trying to understand Web 2.0. Never the less, cloud needs a business focus says Phil.

Zimory’s Philipp Huber used his commanding stage presence to talk about what pervasive cloud computing might look like. He wheeled out comparisons with the energy market. Why does the energy market run so smoothly? Multi-tier supply – energy products, distributors, resellers – and 100+ years to get where they are. Both private and commercial demand, with well educated customers, and well established infrastructure standards (fuses, plugs… ). SLAs and quality standards, operating in an open market. How does cloud computing compare? New cloud providers emerging daily. Increasing demand, but driven by early adopters. The early majority still has major concerns about security and stability. Standards are emerging, but still not providing a seamless experience. Full interoperability is still “in the clouds” – fundamentally it works against the cloud producers, since it reduces stickiness. Billing models need to be transparent between clouds too, and easy to understand – again, something that might not be immediately attractive to providers.

Rhys Jones (from RBS) laid out an IT department perspective in “Clouds are cool (so why aren’t we using them yet)”. Power is nothing without control was his cautious note, before laying out some key benefits from his perspective: Switching from capex to opex is attractive. Cloud can do things that couldn’t be done before (due to resource constraints). Cloud gives ability to scale down (relevant currently). Someone else can do the optimisation, leaving you free for higher value activities. Rhys doesn’t see the cloud as just another form of box rental. Because of the cost structure, it enables rapid, low-risk trials – that opens up new possibilities for innovation. He noted that moving to the cloud is a cultural shift, and changing culture is hard. In corporate IT, demand outstrips supply, and the demand has always been upfront. That steers against Cloud technology.

Wayne Horkan, Sun Microsystems CTO (UK and Ireland), talked about the global cloud infrastructure build out. A little unsurprisingly, he sees cloud computing as becoming dominant. He cited bandwidth figures from Amazon showing S3 using more bandwidth than Amazon web sales – I’m not sure that’s a proof point, but interesting none-the-less. Wayne said, “This is the Klondike gold rush…” – I agree that the building wave is definitely starting to feel that way. Wayne drew out a nice layered model, from network up to operations…. “What’s the next stack?” open source he says. Wayne ran out of time and was gracefully ushered from the stage by a gracious host.

Neil Hutson, senior director at Microsoft, outlined their vision. Drawing a linear platform evolution from mobile, to client, to server, to cloud, Neil outlined the changing economics of software towards a consumption based model. He pointed out that Microsoft already run their own data centres, and have done for a long while, supporting their Live offerings. He outlined the Microsoft Azure announcement, and the themes contained in it: standards and simplicity, and citied support for HTTP, REST, SOAP…. The Azure announcement has been covered just about everywhere, so nothing new to add. Only time will tell where it goes…

[Open Spaces Round up next...]

Related Posts

• Britannic Technologies – Convergence in Communications
• Identity Management
• London CloudCamp Update

• Acquired on-line storage company Jungle Disk.
• Acquired specialist hoster Slicehost. (see Rich Miller’s post)
• Struck a CDN Deal with Limelight Networks (detail at CloudPundit)
• And launched a cloud strategy with Cloud Sites, Cloud Files and Cloud Servers.

I think they’ve rather swallowed this cloud computing thing, and swept in talent from the space. This week’s ecommerce expo in London gave me the chance to catch up with Fabio Torlini, Marketing Director of Rackspace.

We had a very feet-on-the-ground discussion around cloud computing. Fabio agreed that cloud wasn’t the future for everyone, but that it made sense for some. Rackspace sits very much at the premium end of the market, catering to mission critical needs. This has kept them clear of the commoditization of hosting, letting them invest in quality support and adopt a packaged approach to meeting customers’ needs.

I know from my time in the Valley that many a start up owes its success to Rackspace offerings, keeping servers going in all Internet weather. Cloud services haven’t had a great reputation, with some big players suffering a number of big outages. Fabio sees cloud services as a lower-SLA form of service. I’m not sure that I totally agree – cloud brings with it a very different philosophical approach than the one used to build web-applications on dedicated services. But there is a good argument for using cloud-based services for applications with elastic capacity demands and less stringent availability requirements.

Hosted and cloud services alike are a good way of moving spending from CAPEX to OPEX, smoothing big capital expenditures into recurring operational costs. That makes even more sense in times like today, when budgets are tight. It is also a more success-based model, since you buy more services as and when you have the business growth to justify them. No point being stuck with a huge, out of date building and servers for a business that didn’t hit its growth targets.

Another advantage of moving applications into a hosted environment, or for that matter a cloud, is gaining the economies of scale that a company like Rackspace have. Very few businesses would be able to afford to build their own state of the art data centre to match the recently opened facility in Slough. Rackspace understand dealing with customers directly, and also working with partners. According to Fabio, they have a strong channel model, with about half of their business being indirect.

Competition in the space is intense. When the likes of Amazon (with their S3 offering now out of beta) and Google appear in the competitive landscape, it is time for some strategic thinking and action. Even more so when Microsoft decide that they want to play in the cloud too. While many providers at ecommerce expo were holding their heads in their hands, Rackspace seem to be keeping their head above the clouds (don’t groan!). Their very high service levels, and the addition of some leading-edge cloud-based offerings, keep them at the top of the game.

Related Posts

• NetBenefit – UK Hosting
• Start ups – Get out of my Cloud
• Why the Cloud Need Standards

If you are in the UK and interested in finding out more about Cloud Computing, this is probably the best place to come and listen. The previous event was excellent (and packed). As ever, I’ll be there with my camera, and blogging what I learn…

Related Posts

• A Cloud Computing Tour – London CloudCamp
• Start ups – Get out of my Cloud
• Why the Cloud Need Standards

The post pivots around an interview with Harry Debes of Lawson Software. He predicts that SaaS will go nowhere, putting it in a bucket with ‘service bureaux’ and ‘application service providers’: 

“SaaS is not God’s gift to the software industry or customer community. The hype is based on one company in the software industry having modest success. Salesforce.com just has average to below-average profitability… …One day Salesforce.com will not deliver its growth projections, and its stock price will tumble in a big hurry. Then, the rest of the [SaaS] industry will collapse.

Salesforce.com’s Q2 Fiscal 2009 Financial Results (PDF) put them at $263 Million in revenues and GAAP EPS of $0.08. They were guiding at >$1Billion for FY09 revenues. Lawson’s Q4 2008 revenues were $233 million, so this is big boys talking big talk. Debes again:

“People are stupid. History has shown it repeats itself, and people make the same mistakes.”

Selling services and selling products are fundamentally different. Yes, the big software houses sell services to back up their product offerings, but that is very different than being a services-lead business. Customers expect different things, which makes it very hard to transition from one to the other. Few businesses have done it successfully.

If you are looking to SaaS, you want to choose a supplier that has a good attitude towards data portability and security and an exemplary ability to manage and run data centers to strict SLAs. Either that or back-off to someone who has, and then have robust processes between the customer and that third party. That’s a very different set of attributes than you’d be looking for in an on-premise software vendor.

With the SaaS market growing so rapidly, that’s going to result in green eyes over in the on-premise software camp. Vendors can’t run build out SaaS on a product business model. The economics are different, and are currently changing rapidly. Back to Debes again:

“all your costs are up front and your revenue is over a five-year period, the more you sell, the more you lose… …You don’t break even till the four-and-a-half-year mark, but here’s a bigger problem: there’s no guarantee that that customer is still going to be yours in four years’ time”

The last part is exactly why businesses should be looking at SaaS – more leverage over the vendors. Yes, that’s bad news if you are a vendor, but only if you are afraid to play in a more competitive field. As to the economics, the SaaS vendors now have the opportunity to back the up front costs off to a Cloud service. These days, if a service is loosing more the more it sells, it’s time to take the product manager outside for a quiet word. There really is no excuse.

What we have is a win-win for the user and the application provider. The service bureaux was a rather wonderful thing, I’m just about old enough to remember them. They had a long and profitable life, and let businesses get on with their business. New applications could be turned up rapidly, and you had the benefit of their scale and buying power.

The ASPs got killed in the aftermath of the dot com bust, essentially collateral damage. So I’d say Debes is wrong, dead wrong. Far from off-premise software being a blip, I’d argue the exact opposite – our love affair with the PC and on-premise software is the blip. SaaS and Cloud Computing are simply the industry reverting to sanity.

Related Posts

• Start ups – Get out of my Cloud
• CloudForce – SalesForce.com Jumps about in the Clouds
• Why the Cloud Need Standards

Clouds in the Setting Sun

Cloud computing blogs are popping up all over (James Urquhart’s “The Wisdom of Clouds” being a personal favorite - check out his updated “Cloud Computing Bill of Rights“). But what exactly is cloud computing?

Cue “Cloud Computing Defined” on Mashable. The post went up before the weekend, its the conversation in the comments I’ve been watching. Cloud Computing is somewhat blackbox-like for those trapped outside. All the more reason for a good definition, and a good understanding of you might be dragging your business into (or at least the IT infrastructure).

Starting for what it isn’t, Mark Hopkin, on the Mashable blog, quotes Mike Elgan (the names are dropping left, right and center(re) ):

“Cloud computing” has been used to mean grid computing, utility computing, software as a service, Internet-based applications, autonomic computing, peer-to-peer computing and remote processing. When most people use the term, they may have one of these ideas in mind, but the listener might be thinking about something else.

Or to put it another way, it is the marketing umbrella of the day (Kool-Aid if you must). OK, so what should we be looking out for in cloud then? In the 90′s, in my consulting engineering days, we would draw grandiose architecture diagrams. In the middle would be clouds.

The clouds represented the ‘something magic happens in here… will figure out how later’. And for me, that is part of the essence of cloud computing. It takes a boat load of complexity that deals with storage, process distribution and many other things too hard and dull to think about and puts them in a neat box. As a user (in the sense of programmer, as well as end-user), you simply don’t have to worry about what happens in the middle. It just happens, and it is 24×7. I like Doug Klein’s comment:

In the networking cloud bad things happen all the time. Routers die, packets get dropped, loops occur. What has evolved over the decades, however, is a sophisticated and mature set of recovery tools; applications, protocols, processes all designed to detect, work around, patch and recover from failures. For cloud computing to realize its full potential we have to go through the same learning curve. It’s not impossible but it is certainly an order of magnitude more complicated give the much more varied nature of the situation. 

Mark harks back to those early networking days (with a quick name check for MCI nee Worldcom). The fact is, from a technology standpoint it might be tricky to put a finger on exactly what cloud computing is, other than a moving feast, but from a business one it is this: The next generation of IT outsourcing, staring us right in the face.

A little addendum… Even the register is on the case: “Cloud computing: A catchphrase in puberty“.

Related Posts

• Start ups – Get out of my Cloud
• Why the Cloud Need Standards
• A Cloud Computing Tour – London CloudCamp

Before I ask the question, let me answer it: ‘yes, but it depends’. There you go, one sentence and you are totally clued in – rush out and share your new found expertise with whomever you meet. Ok, may be I should wind back a bit.

In the 90′s, IT departments installed LAN servers by the crate-load, giving a central place to share and store data. However, it also meant that you needed network access to get to it. When the network went down, access to your data went with it. Hence Cisco’s slogan in the 90′s: The network works, no excuses.

These days, the defacto network platform is the Internet. We have more and more ubiquitous connectivity, enabling a new generation of apps that have escaped from behind the firewall-locked server farm, into the free-world of the Internet.

It is now easier than ever to process and backup data, share it between companies and individuals, and access it across multiple devices. These applications are still vulnerable to network outages, but the more distributed nature of cloud-based services should, in theory, make them more robust than the previous generation of applications that lived in the server-farm.

Many of the initial applications have been consumer orientated, but that is changing. On one side, IT itself is increasingly consumerised, and on the other, applications themselves are becoming more business centric (SalesForce.com ecosystem, backpack and highrise HQ, etc…).

Businesses now have low-cost access to what were once very expensive applications. We are now in the perfect storm, where IT departments can leverage Internet technology (standards, bandwidth and computing power) to deliver new applications to users.

More on these new services and platforms in the coming weeks…

Related Posts

• Start ups – Get out of my Cloud
• Why the Cloud Need Standards
• A Cloud Computing Tour – London CloudCamp