CEO Perspectives 2: Changing Misconceptions

Expanding into the cloud is like pioneering into new terrain.  Some have gone there and come back with amazing stories.  Others have moved out, lock, stock, and barrel, and keep writing letters, telling us how wonderful it is.  We may experience a longing for adventure and profit, but there are doubts and fears to overcome.  How much will it cost?  What will I have to give up?  What if something goes wrong?  How do I know I’ll be safe?

New terrain.Since cloud computing is new and unknown, it is natural that these questions arise.  For answers, we turn to seasoned travellers who have explored the new terrain and who also understand our concerns.  Last week we mentioned one such specialist, Andrew McAfee, a principal research scientist at MIT who studies how technology is changing the business world.  In his article  What Every CEO Needs to Know About the Cloud, McAfee addresses the concerns of the business community: cost, reliability, and security, in a somewhat unexpected way.  His insights suggest that we may be moving into a space where the old rules don’t always apply.

Take cost, for example.  It’s hard to predict.  There are some studies that show costs double when you move to the cloud, and yet other studies indicate that it is cheaper in the long run.  Which to believe?  McAfee says that it really doesn’t matter.  IT is such a small part of a company’s budget anyway–about 3.2% on average, according to Gartner–that cost is not a big issue.  Furthermore, with cloud providers taking advantage of the economies of scale, and with hardware prices going down all the time, the costs of cloud computing will be continually decreasing for the expected future.  So cost is not as big an obstacle as we might expect.

Reliability is another example.  People look at the Amazon outage of last year and ask, “How can we trust the cloud?”  McAfee points out that at least one company glided right through the crisis: Netflix.  They had anticipated such a scenario, and built redundancy into their system to withstand a service interruption.  So the show went on for Netflix, without missing a beat or dropping a customer.  What’s more, argues McAfee, many cloud services have a higher reliability record than on-premise implementations.  According to the Radicati Group, Gmail’s available up-time is over 99.9%, making it more than 30 times more reliable than most corporate email systems.

Whoa, hold on a minute there, partner.  We’re talking real-time here, not email.  Let’s bring the discussion up to speed and consider how all this applies to a mission-critical real-time system.  OK, the lesson from Netflix is useful.  A real-time cloud implementation should be able to provide some kind of redundancy, preferably through multiple vendors, or if not, at least through functionally isolated, physically separated systems.  At the same time, doing better than a corporate email provider sets a pretty low bar.   It begs the question:

Is it possible for a cloud system to be reliable enough for real-time data?

Which brings us to a final misconception, not specifically mentioned by McAfee, but important to us.  We need to clear up the either/or mindset.  We have to stop asking: cloud or no cloud?  Should I drop everything and move on to the bold, new frontier, or stay here with the wimps and Caspar Milquetoasts?  This is a false dichotomy.  Nowadays the cloud is more than an open public space “out there” that somehow receives and delivers data in vague and mysterious ways.  We have options, such as private clouds for tight control, and hybrid clouds to isolate the machinery, PLCs, and SCADA controls of an industrial plant system from managers and analysts who are authorized to access the plant data via the cloud.  In either scenario, the vital heart of the control system is not exposed to the cloud.

The important thing to remember is that the new frontier out there beckoning us onward is not so scary or inaccessible as it may seem.  It is possible to build an outpost, and still hold the main fort on your real-time data.  However, there is at least one more area of misconceptions to look at: security.  We’ll talk about that next week.