Probably my last post on PAAS, as I think it’s time to move on talking about PAAS – there are more interesting cloud happenings that are exciting. But, to bring closure to a twitter discussion, I wanted to get this post done.
I have written previously about Platform as a Service. Had hoped that things have changed, but there are no signs that they have. Also have done a quick and dirty survey and discovered PAAS is not exactly thriving, but is not dead either. PAAS vendors for most part have not done a good job of articulating the value prop – ignoring that many of us don’t even understand what value these vendors bring to the table, I will outline few of the points that make me uncomfortable about PAAS.
1. Lack of industry accepted definition of what PAAS is.
Each vendor defines what PAAS is differently. If you peel back the original messaging, the root messaging seems to be “Trust us. We know what we are doing. Lock yourselves to our platform and you will do well”. This is what I call a Papal message. In fact, some of the PAAS evangelists are somewhat similar to religious evangelists in their passion and in asking followers to suspend their disbelief. I see tweets like “The blind man in Chicago was cured of blindness by Brother X” – ok, may be not that, but something like my prospect was so impressed he jumped out the window thinking PAAS lets him fly.
If we look past the antics of vendor posturing, it does reveal a structural weakness in the market. There is no interest or motivation to sit at a table and discuss what a PAAS is and if it would make sense to have a reference implementation. As much as we all hate J2EE, we have to admit that they guys at least tried to do a reference implementation and several had published J2EE patterns that served as guidance for developers. This does not exist for PAAS.
Some would argue that .NET did not follow a reference implementation – well that strategy works if you have a segment of the market that believes in your ecosystem – this would indeed work well for Microsoft and Salesforce and not for others.
You could argue that a foundation or a task force was formed. I think we as an industry have seen enough attempts at a committee or a foundation or a task force. These are where vendors get together, drink expensive wines and sign on a dotted line – nothing good ever comes out of these. If this was driven by an end customer, I would have some hope. When we were young, we used to call it “smoke and mirrors” show. Boy, for all failings of Sun Microsystems, I wish we had someone like that now.
2. Vendor Lock-in
Firstly, here is a fact you need to memorize.
Open Source <> Blank check
Just because your project is open source does not impress me. You have to prove value. There are more open source projects than developers out there and one could argue that not all open source projects are same.
I know this has nothing to do with my main argument, but I wanted to get it off my chest :).
Now, coming to vendor lock-in – a good PAAS is the one that does not dictate your choice of IAAS. It should work well on AWS, Azure, GCE and other IAAS options out there. Most of the PAAS vendors do have some level of support for other IAAS options, but there is an uncomfortable realization that their mother ship prefers using a particular infrastructure.
One thing most PAAS vendors are similar is in their universal hatred of AWS. I am not holding my breath that they will treat AWS as a first class infrastructure option. Time will tell, but just a vanilla support is not enough of an assurance that they will continue to put resources into AWS interop.
Also, early indications are that to get full value of PAAS, you should probably consume all the services that PAAS provides and this may mean ripping out what you have and replacing them with new services provided by the PAAS vendor.
One thing I would say is that smaller PAAS vendors actually have an advantage here as they are not tied to a larger mother ship with different agendas. They may actually build a platform that helps their customers.
3. Maturity of PAAS
It is my opinion that PAAS is for bleeding edge adopters and not for mainstream yet. I keep hearing that PAAS demos well, but may not deploy too well in production. There are definitely early adopters who have through sweat and tears managed to make it work, but for a mainstream app development shop that is racing to meet business needs, wise thing to do may be to hold off committing yourself to a particular platform. (Exceptions: If you are already a MS shop, you would be foolish not to use Azure – similarly, if you already use SFDC to build your apps, you should fully leverage SalesForce1).
These are some things that make me nervous about state of PAAS currently, but I do believe we could be at the beginning of a massive opportunity or it could just fizzle out. I am no longer in the camp of believing that PAAS will be subsumed by IAAS or PAAS does not have any value. PAAS has value, but we are probably in top of the first inning in a nine innings game. It’s time to go grab some peanuts and beer and enjoy conversation with friends. Think the SF Giants will get it done this year?