I responded recently to an e-mail from a good friend who works for one of the big database applications vendors. He wrote me expressing discomfort and dismay over hype that swirls around cloud computing. He pointed out that no one vendor had yet put forth a full-scale vision for a shared-responsibility, open-access "cloud." He went on further to say that vendors wouldn't know how they would make money if a true open-access cloud were to come along.
My response was that the appeal of the "plug in, turn on, get IT service" mantra is compelling in an economy that forces companies to look away from capital acquisition in favor of off-balance sheet methods like leasing and the cloud "pay-as-you-go" model. Users will be drawn to both compute and storage clouds and vendors will want to sell them clouds and/or cloud infrastructure stuff. Concern over whether they are "open" has yet to put the brakes on this accelerating force. But there is growing concern among users about cloud vendor lock-in.
Amazon S3's application programming interface is the defacto standard for storage clouds--a fact that neither Amazon nor storage cloud user communities seem to mind much at the moment. Nonetheless, there is now a collaborative effort dedicated to furthering the adoption of cloud standards, storage and otherwise.
Openness aside, there are a number of assumptions floating around about storage clouds that have yet to be tested. These are assumptions like infinite scalability, unlimited availability, and increased responsiveness. Nothing in computing is infinitely scalable. There have been and will continue to be storage cloud outages. Response times are not deterministic. And nobody seems to be really talking about security all that much, let alone legal issues surrounding personal data in the cloud, SEC and other compliance issues, and potential chain of custody issues.
The biggest concern I have right now is for independent cloud storage vendors, particularly some of the small but rapidly growing start-ups. At CloudCamp Boston, someone provocatively mentioned that compute clouds commoditize IT operations. Interesting observation. One could argue that IT operational commoditization at least started when the help desk went overseas. But let's push that thought forward a bit more with regard to the storage cloud.
Amazon's S3 service was carved out of an existing IT operation that supports a different core business--retail. In essence what Amazon did originally was to monetize both spare capacity and the operational support staff which presumably was already being paid for by the retail side of the house. Is Amazon the sole owner of a secret formula that turns internal storage into cloud storage gold? No. Any reasonably large IT shop with an entrepreneurial bent could do the same.
Enron (remember Enron, the energy company?) was once a storage service provider (SSP), but not like their many SSP rivals at the time. Enron wanted to create an arbitrage opportunity around the excess capacity held by their SSP contemporaries. They proposed to buy and sell storage capacity in the same way they were buying and selling energy at the time--as a fungible commodity. Yes, Enron believed that storage sitting on a raised floor inside a data center or a co-lo was fungible.
Too bad for the smartest guys in the room. They were about eight years too early. It took an Amazon to prove that storage is indeed fungible. Could other big retailers do an S3 equivalent? Why not? There's nothing really stopping them. And why not financial services firms? Manufacturers? This list goes on. Nothing stopping them either. And I believe we will soon see brokers of cloud services in the same way we saw brokers of time-sharing services back in the day of the mainframe.
Hybrid clouds--ones that are both internal and external to the enterprise--will eventually dominate. There is no reason why a large internal IT services provider couldn't do what Amazon is doing, namely partition-off some idle capacity and offer it up to a cloud-based services consortium (the "open" cloud?) for a fee. Huge investments in computing and communications infrastructure have already been made. We will see an open cloud for example when a group of entrepreneurial IT executives form a consortium that offers utility cloud-computing services. These will not only be open, but their services will be priced to market. I hesitate to say "commodity" but that's where I think utility cloud services are going.
I believe it will happen this way because a historical precedent has already been set. Remember the old mainframe time share days? When IT represents a significant investment to the enterprise, IT executives look to maximizing its efficiency. One of the ways they did this back in the day was to offer up idle mainframe compute cycles for sale. They sold cycles decades ago. Today they can sell compute cycles, storage capacity, and the operations staff to manage it, all delivered over the wire. Amazon isn't the first, nor will they be the one and only.
This economy will force big enterprise IT to look at replicating the Amazon EC2/S3 model for two reasons:
Amazon can deliver IT services more cheaply than most large enterprise IT departments so the IT executives managing large IT shops will want some of what Amazon is smoking. They'll imitate the infrastructure. They'll buy cloud services. They'll sell cloud services.
Amazon has proven it can make money off its idle compute capacity. Every IT shop runs through computing peaks and valleys. Why not sell the idle capacity during the off-peak times just like Amazon. An "open" cloud would allow them to do this.
In the end, computing vendors make money from clouds by selling the underlying infrastructure. Enterprises make money from clouds by selling access to idle infrastructure offered up to the "open" cloud. Brokers of cloud-computing services make money by inserting themselves in between the buyers and sellers of cloud-computing resources. And, all of the foregoing means that the independent cloud services providers will live in an increasingly commoditized world where the price of the commodity is set not by those selling services as a core business, but by those selling cloud services as an adjunct to their core businesses.
That in my mind is the threat that the smaller independents may or may not see coming. As an independent, having an application wrapped around a cloud infrastructure is one way to forge a more sustainable cloud business model.