I’m sitting in my office on a very cloudy, indeed stormy day. Ironically I’m contemplating cloud services for one of our customers and wondering if the transition to cloud might be stormy for them. On one hand, they already subscribe to a few Software as a Service (SaaS) solutions to support specific operational functions and have had for years. Yet, in a meeting with some folks in IT, there were some strong opinions about “never” moving to the cloud. Despite the irony of these statements, having had several years of SaaS experience, I wanted to know a bit more.
After a little digging, I felt that what concerned them was really a move towards Platform as a Service (PaaS). This is where there seems to be potential for the cloud to turn stormy. Why might it turn stormy? Most (all?) software vendors that provide basic platforms for delivery: database, data integration, etc., are making intentional moves towards PaaS subscription based models. In an undetermined number of years, each vendor will prefer to move customers towards a PaaS model. Whether they will continue to offer on-premises software licenses and at what cost remains to be seen.
Why is there a visceral reaction to having cloud based PaaS versus on-prem software, when there is less resistance to SaaS? In part, I think it is about who is making the decision. For SaaS, the decision is made by, usually, well defined functional groups in the business looking for a discreet solution to an operational need. PaaS is an IT decision about what underpins, well, everything that glues the functional and information architecture together. There has typically been a large investment in platforms that have evolved over years to improve and optimize business function and integration. Whether an organization has single sourced it’s integration software or has multiple vendors, the messaging and batch data integration, web and data services took years to create and balance. Moving that, or even some of that, represents a huge change and risk.
But if change, as always, is inevitable, and the software vendors are moving towards PaaS, how can this transition be managed? If an organization is starting from scratch, deploying SaaS and PaaS, this would be easier. Even so, this still has challenges since different SaaS providers are just as difficult to integrate functionally and data as traditional on-prem enterprise software is. As organizations increasingly move discreet functions to the cloud via SaaS we already have an integration challenge. Throw in a transition to PaaS and IT has a bigger headache.
To make a transition, like most things this big, it is best to start small and prove out the concept. Don’t try to move messaging, batch, a service integration all at once. Even within these categories start small. For example, instead of pushing all your ETL into the cloud, start with a small project to move a few mappings. Test them out, find out what’s the same, what’s different between on-prem and cloud based data integration.
How does support of the operations change?
How does performance change?
How does development to deployment change?
How well do your ancillary tools that support testing, change management, etc. work with PaaS?
How well does your cloud based ETL work with SaaS and on-prem databases? PaaS databases?
This is just a start of the things to consider. Now you can see why IT feels a bit uneasy about this type of change.
In general I’m a supporter of PaaS since the long term cost benefits of subscription vs. on-prem are compelling for businesses. In particular, for new initiatives such as Big Data and Analytics projects where platforms to deliver these solutions may not yet exist. However as a replacement for existing platforms, the short or near term cost of the transition is often not a topic the software vendors openly discuss. But transition to the cloud for PaaS can potentially be substantial efforts given that is an overhaul of the base, fundamental architecture that provides the glue that holds an organization together.
So, organizationally as well as architecturally, the move to cloud could be stormy. Once we get over the rainbow to the other side it could open up flexibility and interchangeability, with less cost then we have had in the past.