Recently, I read an interview with fellow IBMer, Bruce Otte, in which he laid out IBM’s Roadmap to ‘Smart Clouds’ by highlighting five key areas. I’m not going to enumerate those here, if you are interested you can check out the interview, but there is one thing that especially caught my eye: the notion of cloud appliances.
Even before my day job revolved around one of our cloud appliance offerings, the idea of cloud appliances, and actually just appliances in general, intrigued me. Coming from a background that was decidedly hardware agnostic, which is not very uncommon when you are developing application infrastructure software, the thought of combining software with fit-for-purpose, enhanced hardware was a bit foreign. Further, since I spent my days (and nights) developing application middleware that was targeted to run anywhere and support untold variations of usage scenarios, the idea of a device built for a very narrow set of usage scenarios was completely foreign. Quite honestly, I did not understand the appeal of appliances at first.
Fast-forwarding to now, after a year and a half of chatting with users about an appliance, my views are quite different. To be clear, my views are not different because they have to be. My views are different because of the direct feedback I get from users that leverage these kinds of appliance in their enterprise. By distilling what I hear from users, a majority of the value of an appliance seems to owe to its built-for-purpose nature (important to note that this feedback is usually in the context of our various WebSphere appliances).
Appliances intentionally support a constrained set of usage scenarios. On the surface, this may seem bad, especially to those who immerse themselves in the ‘One tool to do it all’ mindset, but that is clearly not the case. First, because the appliance tackles specific usage scenarios, optimization at the software/hardware integration layer is possible. Implemented correctly, the result is performance hard to match by traditionally installed software.
Additionally, because providers build appliances for purpose, they can load the devices with out-of-the-box goodness. Consider appliances that provide data transformation capabilities. Many times, these types of appliances come with assets that handle transforming data from one format to another without the need for any kind of user coding or scripting. Any built-in expertise is an advantage and a bump in time to value, because it means users do not invest initial development resource, or ongoing maintenance resource, to provide these capabilities.
This is not all that I hear from users regarding value derived from appliances’ purpose-built nature. Users like the hardened approach to security resulting from being able to do things such as dedicating hardware within the appliance for security processing or decreasing attack surfaces by only initiating critical processes. They enjoy the user consumability afforded by appliances. They do not have month’s worth of features to learn, rather a narrow set of functionality to learn and exploit. Quite simply, in a world full of born skeptics, there seems to be a real soft spot for appliances.
All of this brings me back to Bruce’s interview and his mention of cloud/cloud-enabling appliances. I cannot help but wonder if appliances will become a central, de-facto part of cloud architectures (or maybe they already are)? Certainly, the affinity of typical issues faced in the cloud (connectivity, resource management, data integration and management), towards purpose-built appliances seems strong. In addition, the purpose-built nature of appliances really looks like a microcosm of the *aas trend that we see in cloud. Essentially, these appliances provide some type of functionality as a service, with little effort on the user’s part. So, what do you think? Will appliances be an integral part of cloud computing going forward?