Anyone who followed Cloud Computing last year watched several changes occur throughout the space. These changes were happening in the view of what Public, Private, and Hybrid Clouds were defined as, the interest in IaaS solutions in the Enterprise and within Service Providers, and finally a renewed interest in the potential of PaaS.
Currently Public PaaS solutions are focused on core compute functionality and persistent services all of which are being backed by some mix of IaaS (with or without Virtualization, more with than without though). Why are so many Public PaaS focused on their IaaS underpinnings? Two reasons come to mind, IaaS is easier for customers and IT to understand and yet generic enough to have the broadest appeal. This is important in the continuing evolution of the Cloud market as a whole, we need the widest adoption possible early on for any Cloud Platforms to take off.
Now, focusing in on the set of events around PaaS in the past year brings us to the rumors that surrounded EngineYard’s acquisition by VMware (which never happened and now we see that Amazon is working with them on a Ruby on Rails Elastic Beanstalk), Heroku was acquired by Salesforce.com, VMware announced the Open PaaS solution, and Microsoft will fight with anyone who calls Azure anything but a PaaS. All of this Public facing PaaS news sounds great, but I’m sure you are thinking how does this translate to Enterprise?
I wondered this same thing and began to try to figure out just where it all leads.
What is the Future of Enterprise PaaS?
Public Cloud Services are always different from Private Cloud Services, many people think that Hybrid is something easy but miss the nuances involved with implementing Hybrid Clouds. I’m not saying that achieving a Hybrid model is impossible, just that there are many hurdles and few have working solutions that seem to address most or all of them.
PaaS in an Enterprise will operate as a Service Oriented Platform. That sounds like a silky smooth sentence doesn’t it? But there is substance behind the statement. Today the most advanced Public PaaS platforms are focusing on generic, on-demand, multi-tenant, infrastructure components, and other core capabilities. What Enterprises will need for PaaS will be this plus many other services including specialized components such as vertical specific ERP and CRM solutions exposed as services, legacy solutions with service bridges, Big Data service connectors, and many more. As these capabilities get built and each former “Silo” gets exposed as a service a Platform will emerge.
About the Service Oriented Platform:
Something to understand is that each business is different, it was formed differently and while it operates most likely in a similar fashion to other businesses in that market, it is still different. This difference becomes even more pronounced when the needs are pushed to IT (Operations and Development). This would be the equivalent to the Butterfly Effect in IT.
Why does the Butterfly Effect matter for an Enterprise Service Oriented Platform?
Because it means that each Enterprise Service Oriented Platform will be DIFFERENT than all others (maybe only marginally, but still different). Each service exposed to the ESOP will be a new capability that Devs can leverage as they put together Apps on top of the Enterprise PaaS.
Something that may be important to realize is that IT Operations may view the solution as an ESOP while Devs may view it as an Enterprise PaaS. I think both of these are actually correct and will talk about the same thing, it is just how it is being viewed and by who, but it is still the same thing.
Hybrid Clouds will be exposed as another service capability:
Hybrid Clouds will be Public Cloud services that are exposed either directly or indirectly to the ESOP. The ESOP will have the control, authentication, metering, and likely the QoS. The EPaaS will have the allowed/exposed services and resources from the Public Cloud. This will be at a different level than simply pushing and pulling VMs, Virtual Hardisks, or Creating VPNs.
I’m interested in people’s thoughts and comments:
Comment on this Post or Follow Me on Twitter – @mccrory
I completely agree with SOA and PaaS merging. Nice!
I agree with your assessment. Many of my customers want their PaaS to “inherit” capabilities from their IaaS. Features like auto-monitor, recover, scale, etc. are all core to IaaS but serve as the foundation for a well designed PaaS.
Even Amazon focused on “Simple Platform Services”, that is, services which could scale horizontally. They intentionally removed requirements which would hinder linear scale: cross table joins in SimpleDB, once-and-only-once, in-order reliable message delivery, etc. These “Complex Platform Services” can be built on top of the Simple Services which leverage the horizontally scaling services.
Only recently have we started to see the higher class of cloud infrastructure services become available: Fibre, SSD, SAN, very-large memory, etc. This could spark the next generation of more sophisticated data services.
Great stuff. It’s going to be an exciting year. BTW – I’ll be out at Dell next week.