The TCO of Cloud

This is an article I had published in October on The Rust Report

ImageWhile it is perfectly understandable that Larry Ellison missed his own Keynote at Openworld so he could watch his yacht do battle in the America’s Cup; what is harder to understand is that the entirety of the content – ultimately delivered by his chief engineer, Thomas Kurian – was about Cloud.

Cloud, after all, only represents between 3 and 10 per cent of Oracle’s overall revenue (depending on how you crunch the numbers).

It seems only five minutes ago that – quite rightly I think – Larry Ellison derided the IT industry for being “worse than the fashion industry” in its obsession with the Cloud.  But more telling perhaps was his comment that, “I’m not going to fight this thing”.

He certainly hasn’t and of late he has become more the “Sugar Daddy” of Cloud.  The original angel investor in and Netsuite a few months ago announced Cloud-based deals with many of his competitors including Salesforce and Microsoft.  Larry is having a bet both ways.

But is Cloud all it has cracked up to be?  I want to talk in particular about applications delivered as-a-service, rather than merely “hosted” computing resources.  SaaS brought much needed nimbleness to the industry.  Deploying it is cheap, there’s no hardware to buy and the subscription model famously makes it easy to buy on a corporate credit card, circumventing roadblock IT departments.

However, as many companies are beginning to discover, Cloud isn’t the panacea it was once thought to be.  Did anyone do proper Total Cost of Ownership (TCO) studies?  What were once quite manageable self-contained deployments of CRM-on-demand, for instance, are now having to be deployed company-wide with a huge associated integration costs; not to mention the pain of harmonizing data models and user environments.  (Let’s not even get into the security and sovereignty issues!)

What does this cost in terms of consultancy hours, lost opportunity and the agony of internal political division?  How does this compare with the visible and predictable – albeit expensive – costs of existing legacy on-premise deployments?

Also, it isn’t just CRM now, with vendors like Workday or Marketo becoming so mainstream, and then established players like Adobe and SAP moving into this space; identity and access is a nightmare for most IT departments.  At least with on-premise the same IT people had a strategy.  Now there is a plethora of applications breaking out that those required to manage them didn’t even chose.

And SaaS vendors are now just as guilty of three-year license deals and a “good luck with that” attitude to roll-out success.

So perhaps it is too soon to write off the on-premise model – secure, predictable and reliable as it is.  Lets consider hybrid models more carefully and lets treat Cloud computing as an option, and not the promised land the industry would have us believe.  There is of course an important role for cloud and particularly SaaS, but the business case should be thought through more soberly and the parameters of that role should be more closely defined.

Because caught in a private moment, many Oracle executives will confess: “on-premise still keeps the lights on”.