Monday, December 16, 2013

Clouded judgment, slip in the rain

He was one proud adopter of cloud technologies and had moved almost all applications to the public cloud. It was a case study written about and discussed in forums and publications. He was hosted, flown across cities by the provider to talk about his experience and give advice on why everyone should consider the cloud. His stature had grown and he advocated the use of clouds for scalability, lower TCO, variability of cost, almost like the poster boy that the industry was looking for and had found one in him.

I met the CEO of that company last week and engaged him in some small talk on the business and how my cloudy CIO friend was doing ? The CIO had reverted all applications back to the on-premise data centre in a hurry. One fine day the cloud provider declared an outage and it just happened to be the day when the load was high; the revenue impact was substantial. The CEO had then warned the CIO to take care of such eventualities. Then of all things the hosted gate pass application stopped working.

Last week was one with unseasonal rain which surprised everyone with every falling drop. Across multiple discussions with different groups in formal and informal settings, adoption of cloud was omnipresent. How many of you have adopted the cloud for your workloads ? Which apps would you move to the cloud in the future ? Why is manufacturing so averse to adopting the cloud ? Even ERP is now available on the cloud ! Vendors in some discussions offered to conduct a “Cloud readiness assessment”.

On the table were many views on all kinds of cloud: highly virtualized data centres masquerading as private clouds, to Infrastructure, Platform and Applications as a service. All flavours of solutions and orchestration engines including a couple which promised to manage hybrid clouds. Voices raised incidents of hybrids not working effectively or seamlessly, leading to clouds not working with random and frequent outages, leaving the CIO smarting with no real recovery option except to wait it out.

In a discussion, taking examples from Sales Force Automation and some ecommerce applications doing well, CIOs from manufacturing were challenged to find business cases within their industries and organizations. None forthcoming, the vendors provoked the group to stop hugging servers and let go. Chastened the group retaliated and one CIO raised a question to a hardware vendor: “Have you move your ERP that runs your factory and supply chain to your Cloud ?” Sheepishly the answer came “no”.

Many vendors will sell the cloud with financial metrics that belie any rational thought; cost of variable fractional CPU saved and hundreds of MB of storage across 17 applications adds up to some cost, or the additional hour spent by the engineer over a weekend to ensure that the full backup was successful. Do we really pay that way such that we can apply microeconomics to calculate real savings ? I have yet to come across anyone who did and saved big bucks. Decision between capital investment and operating expense is a CFO call.

Coming back to the gate pass application, why is it such a critical application ? Entry and exit of every person, vehicle and goods from the factory premises depends on it. It was the most innocuous and the most critical application for continued operation of the factory. Amidst the lot of hue and cry, the CIO had no option but to come down to terra firma. He admitted that he should have been pragmatic in his approach and not moved everything to the cloud. DR to the cloud was not envisaged and there too laid his folly.

Everyone is experimenting and exploring the cloud in some shape, form or avatar. Many have moved non-critical workloads or external apps; not many references of other kinds of legacy or ERP type apps for largely centralized enterprises. Hybrids remain experimental for now while start-ups are enjoying the benefit of no upfront investments. Pure play technology companies find clouds viable, the rest of the old and large businesses continue to tread cautiously and take a step-by-step approach to the cloud.

You don’t slip in the rain that way !


  1. I am sure the executives do the Application Rationalization before moving any workload to the Cloud and also have DR plan. Despite of such detailed planning, it is possible to a hit bend and thats why while hiring his commanders Hitler's final question used to be: "You are very talented and I like you but are you lucky?" :-)

  2. cloud or on premises: there is absolutely no difference in the backup strategy. This is off topic.

    Infrastructure in the cloud offers different benefits that on premises can't offer: extensive scalability
    However nothing can proove that on premises is more reliable than cloud. Or vice versa.
    Last, you can definitely run on multiple cloud, or combo of on premise and cloud.
    Sounds like this CIO didn't have the right team for the job. Maybe his team thought Cloud would mean they don't need to worry about redundancy, backups which does not make any sense to me.
    We need more info to understand what happened