Approximately two years ago, I sat in a stylish, yet reasonably priced restaurant in Ashburn, VA, taking a relaxed post-technical-chinwag luncheon with some of the jolly decent folks from AWS. Before I go on, I will say that I am not going to name names of the attendees from AWS, but suffice to say that we had, for some time, enjoyed a great working relationship with the folks there (despite our account manager being in Seattle and us being on the East Coast), so the conversation, as you could imagine, was fluid, open and frank. I will endeavor to replay some of it to you here, by means of a scene setter:
Me : “EC2 is fantastic, but what I really really want, more than anything, is the ability to have the AWS orchestration platform, API’s and all, running in my own data centers, so that I can use a standard tool-set to provide an easier mechanism for our administrators to manage deployment of workloads on both the internal and public cloud” (Ed: nobody called them private clouds back then)
AWS : “Ha ha. No chance. Are you mad ?”
Me : “Erm. No. But, what you’re telling me is that either a) we can’t have it for free, b) you don’t have a model for selling / leasing / renting it, c) it’s incredibly complex, held together with string, gum and is basically not-repeatable outside of your environment or d) all of the above ?”
AWS : “It is fair to say it’s complex, yes. And no, we don’t have any way, nor intention, of making it publicly available. It’s a nice idea, but not going to happen”
Me : “You do realize that this makes it a burden on the data center ops staff ? If we want to put workloads in EC2 / S3 now, then we have to use two or more management frameworks to accomplish that” (Ed: EBS wasn’t invented yet)
AWS : “You can use the Eclipse SDK tool kit to build your own interface into AWS services”
Me : “That’s nice, but I think you’re missing the point. I have nothing in my data center to manage orchestration end-to-end. I have Virtual Center and XenCenter, but I want whatever it is that stitches the resource pools together with the requests. You must have something that does that, or, do you have a bunch of people running around behind the scenes manually spinning up servers and storage every time a new request comes in ?”
AWS : “Actually, there is some of that…..(jokingly)”
OK, so let me cut to the chase. It’s two years later and I am still searching for Utopia. I have kept my ear to the ground and my eyes on the prize, but I still can not come up with a workable solution that fits my absolute goal. I know why. It’s because it doesn’t exist, yet. I say “yet” because I really hope that someday it will become a reality.
Yes, we have tested a lot of what’s out there and I know that there are good solutions out there that do “on ramping” for public cloud – not quite what I want. I know there are good solutions that allow you to build a private cloud and then broker workloads out to the public cloud – close, but no cigar. What I want is a TRUE standards-based platform (not just an API) that is 100% deployable for a private cloud and 100% compatible with ALL major cloud providers, so that I can reduce the complexity overall across my estate (wherever that may reside).
I audibly groaned in dismay recently when I read that OpenStack is dropping the work for the Amazon API because of fears around this, that or the other. I am equally concerned that Amazon’s stance seems to be that their platform is “good enough not to be shared” and from what I can gather (disclaimer : I could be wrong) it looks like they will not participate in any of the various open initiatives. Is that what happens when you have the best and most successful “tool of the time” ? If so, then I hope that AWS doesn’t catch an debilitating case of the paralyzing innovator’s dilemma.
My dream is simple in my mind and on my whiteboard, yet seems far away from being readily attainable. It is nothing more than a wish for the “global DNA” of cloud services to become as well standardized, accepted and as widely available as the Universal Serial Bus. I would urge the vendors (you know who you are) not to view it as a lost opportunity (by staying proprietary) but look at it as an opportunity gained – after all, how much revenue has been made over the years on the back of the burgeoning USB device market ? The world has changed and the days of enterprises employing the acquisition methods of single vendor, proprietary solutions are gone, and they are definitely not coming back – not in any company that wants and values agility as a core business differentiator.
But hey, why would any vendor listen to me ? Well, for one reason and one reason alone. I am the customer. That may not mean much any more in the world of fast-paced, ultra hi-tech and “who can out-do who with new product names and glossy advertising”, but I think there is a clear and present danger that those (customers) who must try to implement the “next big thing” can only do so if it is simple, economical and above all, portable.
Changing times are upon us, faster product cycles, unbelievable rates of innovation and a choice like never before, but it really doesn’t mean much if the customers a) can’t keep up and b) can’t or won’t implement what is available because it doesn’t do what I need – it does what you think I need – and at the end of the day, it’s just too damn hard