Clouds Becoming Solid

As an IT manager, I always liked my servers nearby. Co-location made a lot of sense for some, but I wanted my servers (and phone systems) where I could see them. In one position, we had two computer rooms in a redundant configuration in different buildings with backup generators on the roof. But now I see things differently. IP, SIP, and other web protocols make the vast majority of services, systems, and applications distance insensitive. I find the vast majority of my desktop time spent in the browser accessing critical services that I can’t physically see and I don’t miss them. In fact, I enjoy the fact they are gone.

My point is broad, but let me illustrate it with a simple target; Microsoft Exchange. For most of my career, I could be described as an Exchange bigot. I use Exchange/Outlook at work and I use Gmail for personal and blog stuff. I use both fairly heavily every day. Both services have experienced outages, but only when Exchange is down is it really my problem. It normally goes down because it runs out of space, so we paid for recurring monitoring to help manage the server. But the solution really involves much more than managing the Sever. It also means managing Outlook on the desktops; and roaming profiles that seem to keep growing (largely due to email). I bought a Microsoft cell phone that connects to Exchange requiring Outlook Web Access on yet another server; we had to purchase public IP addresses and web "certs" with recurring charges. We also had to buy spam protection and virus protection for Exchange. We have an AC unit for the computer room; it failed and caused our last outage. The various add on tools, facilities (servers, a/c), applications (Exchange/Outlook), networking infrastructure (Firewalls, capacity, VPNs), and ongoing staffing issues makes running Exchange non trivial for a small business. Did I mention service packs and security patches?

Gmail on the other hand does not require a special client, nor special steps or software for remote access. The virus and spam protection are better than the products we purchased, and the phone integration via IMAP is fairly seamless. Additional applications within the Gmail family include Calendar and Contacts. Plus I can find things in my email client faster and easier than I can within Outlook. I can access my email from any desktop, anywhere, and I don’t have to deal with A/C and VPNs. The fee paid service level (Premium) includes technical support, APIs, and now Outlook support. In my situation, I have all the Microsoft infrastructure in place and operational, and I think it might be easier and cheaper to move to the cloud – presumably if it wasn’t in place the cloud proposition would be even stronger.

As a telecom type I’ve been drawing clouds on whiteboards since they were (black) chalkboards.  The cloud used to mean network, or specifically the network we didn’t care about… we could draw the in and the out and mentally outsource what happened in the cloud. Well that metaphor, mentally outsource, seems pretty accurate and attractive. Just as real clouds can take the shape of anything we can imagine – The cloud is now presenting virtually any combination of services we desire – a virtual line from complete hosting to simple "cool" with an infinite number of combinations and permutations in between.

Consider the components of a voice system – processor, directory, storage, gateways, call processing software, trunk hardware, and phones or other endpoints. Plus various applications such as messaging or speech recognition may duplicate most of these components. The cloud is evolving to a model where most of these components can be delivered as a service. The traditional “hosted” solution outsourced the whole thing to a per user charge per month – but now we can outsource the servers only (PaaS) and implement with purchased software. We can put some components onsite, some in "colo" facilities and rent the others as a service. .

Mitch Kapor, the founder of Lotus Software, recently commented that we have nearly arrived at a time where startup companies don’t need an IT department. His point resonated. Amazon’s EC2 service shows a published price of .10/GB/Mo for storage – That is $1/mo for 10 GBs of fault tolerant storage. Look ma, no IT! Obviously some expertise is needed to implement and support, but not anywhere near the effort required to maintain a datacenter. These costs and models are changing the dynamics of the cloud.

The fact is we are running out of objections to the cloud. We can’t argue network response times or bandwidth in our web based world. The argument over integration with other servers and applications are minimized with modern XML and simplified APIs. The rent vs. own arguments are becoming moot as the cloud adapts to support both. The best argument, cost, seems to be crumbling fast – forget parity, but increasingly cost arguments favor the cloud. Kapor was actually dead wrong; what I believe he meant was the IT skills required for an enterprise are changing from physical to architectural. That’s another post.

Another factor favoring the cloud is technology lifespan. Voice communications are experiencing a rapid amount of change. This isn’t like the normal tech always changes creed. In fact, voice technology has been relatively stagnant in terms of value. There are virtually no voice components right now that won’t be radically different in just five years. Products like Google Wave – still unreleased, attempts to redefine email and IM communications. Voice is just sorting out IM and presence. Don’t bet on PRI links continuing to be the default building block. The core call processing capabilities – not just the features, are changing the way we work. UC technologies are quickly impacting ROIs and internal operations. Telecom technology goes on the books typically with a five year depreciation schedule; services don’t.

While the recession has us looking short term, a perfect storm is forming for cloud computing long term. The perfect storm metaphor is perfect, but not because it’s another cloud pun. Look at the multiple fronts converging together: mobility, acceptance, economics, and uncertainty, IBM, and Google. For the first time, virtually all computing services can be delivered over the web, the world is flat.  The web is delivering quality and secure services to consumers, raising both awareness and acceptance to cloud services such as airline booking systems, banking, and personal email services. Economics for cloud computing are shifting to favor the cloud, intensified by a down economy. Uncertainty based on rapid technology improvements are keeping some from long term depreciation schedules.

In a recent NYT article, IBM is now legitimizing cloud computing. The article suggests that IBM legitimized Linux and the personal computers as a prerequisite to widespread adoption. But IBM is not alone. Google and Amazon are enabling self hosted clouds in fascinating new models. Google recently published a how to guide for creating “warehouse scale” machines. This is no longer bleeding edge technology seeking an application.

Personally, I tend to always almost get things right. My advice: buy stock in an umbrella company.