Enterprises said they believed that three years from now, most of their cloud dollars would go to players other than those they’re spending on today.
Everyone knows who the "top" cloud providers are today--Amazon and Rackspace. Many, even most, believe that other players like the telcos, cable companies, or even IT or software giants like Oracle, have a formidable challenge catching up with these guys. If the 122 enterprise users of cloud computing I've surveyed are right, the "everyone" and the "many" are wrong. The reason is that the shape of the cloud market, the real one, will be very different from the market we see today. In that real cloud world, the big network operators can--and likely will--be players.
Enterprises worldwide spend about a trillion dollars a year on IT equipment and software. The total cloud computing revenue stream this year is hard to define because providers don’t break it out, but the number I've seen batted around in the financial industry is just a tad over $500 million. So what we're all getting so excited about is spending equal to one-two-thousandth of global IT spending. Even my estimates of how much cloud computing could earn, which purists consider highly pessimistic, say that a full 24% of that trillion bucks is on the table. If that's a realistic projection, we've so far realized about one five-hundredth of the market potential. Aren't we getting a little ahead of ourselves calling winners in the cloud space?
Enterprises say the great majority of that potential market will come from their use of the public cloud to offload work from their own data centers during peak periods and failures, and the rest will come from support of highly bursty and unpredictable applications. To get these applications onto the public cloud, we need three basic things:
* The cloud has to be at least nearly as reliable as internal IT.
* The cloud has to be cheaper than providing the overflow resources by buying more servers.
* The cloud and internal applications have to be able to share work dynamically.
In my survey, I asked about these three areas, and enterprises reported some pretty radical conclusions based on their own cloud experience.
First, 86% of enterprises said that their public cloud services were less reliable than their own data centers were, enough less reliable to make them unsuitable for data center offload in any form. The problem most often reported was unacceptable variability of performance levels, but over a third of enterprises said they’d had availability problems with public cloud services as well. This doesn't mean that you can't create or offer a reliable cloud service, but it does mean that the current offerings aren't meeting stringent standards for mission-critical applications, and these are the applications most businesses spend their IT dollars on.
Second, enterprises said that on the average for offload/backup applications, Infrastructure as a Service (IaaS) offerings from the current market leaders were 75% more expensive than an internal solution would be. Basic cloud services don't provide the tools needed, advanced cloud features are more expensive, and nothing addresses data security problems. There's also a much larger perceived risk to the applications from the reliability difference I already mentioned. This all adds up to a demand for a savings premium from the cloud that's just not being delivered today.
The final issue in cloud offloading of applications is support for a dynamic reassignment of resources when offloading is needed. Think about this one a minute. Your primary transaction processing system goes down, so you need to switch to the cloud. In the traditional IaaS model, you'd spin up a set of VMs and somehow redirect your users to those systems (forget how that happens for now). That's something that doesn't happen immediately, so you really don't have dynamic offloading. OK, then you keep your virtual machines hot 100% of the time so there’s no delay. That pushes your costs way up.
You might wonder at this point how I ever thought that even 24% of IT dollars could go to cloud services. The interesting thing is that enterprises say that one cloud provider already offers a lot, if not all, of what they need. It's Microsoft Azure. Azure is a platform-as-a-service cloud that's designed to help enterprises build either cloud-resident or cloud-cooperative (hybrid) applications, and that can be made to share data center files with cloud applications fairly seamlessly. For SMBs, this is considered a really good combination, and Azure pricing isn't bad when compared to SMB internal IT costs, but for enterprises the pricing is still too high.
That's where enterprises think the telcos will come in. Some are already trying services from carrier cloud providers (they're not talking about the specifics), and they believe that for their applications the telco clouds will be reliable, perform as needed, be priced effectively, and be easily integrated with their IT operation to support dynamic offloading. Those 122 enterprises said they believed that three years from now, most of their cloud dollars would go to players other than those they’re spending on today. While they're not saying telcos will be those players, they think telcos have a shot.
Don't count Ma Bell out yet; she may morph into Ma Cloud.
