Cloud may change the definition of “business critical” applications

Google outages are rapidly becoming as passé as earthquakes to native Californians; unless it’s a really big one, no one really pays much attention. So it shouldn’t be surprising that Google’s latest “crash” (caused by some interesting routing problems, apparently) evinced an attitude of nonchalance from Stanley.

Who is Stanley? I don’t know, except that he was quite vocal about the outage and his opinion that he was “not really bothered by it.”

Google Crashes Again on Friday

Stanley Was wrote: Wednesday May 27 from around 8pm till shortly after midnight, I was not able to check web statistics for wasseo.com through Google Analytics since this service didn't work from where I'm at (Grootebroek in the Netherlands). At first, I thought it might be due to the electrical discharges by the storm that was raging here. However, other sites were accessable [sic]. Thus, it must have been a problem with Google. This is the first time that this happened to me and I'm not really bothered by it. [emphasis added]

May. 28, 2009 05:31 AM EDT

I have to ask: is your employer bothered by it? Because there is certainly a financial cost associated with that outage. The question is not whether there was, but how much.


SOFT COSTS MAY BE HARD TO SWALLOW

Even though non-mission critical applications, like Google analytics, may not directly affect the bottom line in the same way as revenue-generating applications, they still cost the organization financially. The problem in determining just how great an impact the outage of a non-critical application may have on the bottom line is that the impact is generally a soft cost, meaning it is measured using factors that are not easily quantified. This includes productivity, under which Stanley’s scenario certainly falls. After all, if part of his responsibilities are to collect metrics via Google Analytics and he was unable to do so, what was he doing? We assume something else, but that’s not always the case.

Too, it’s often difficult to see how one small activity fits into a larger process, one that may in fact be delayed by such an outage. If a decision making process relies on the data Stanley was charged with collecting, it may be the case that this process was delayed, resulting in a delayed business decision that ultimate cost the company $X dollars.

And in the real-time reactive digital world, it may be the case that the data Stanley was trying to collect could have pointed to a problem either with the site – a large percentage of abandoned conversions, for example, could be due to poor application performance or an application error – or with the campaign itself. Both problems need to be addressed sooner rather than later but because of the outage may not have been addressed quickly enough, resulting in lost customers or revenue.


CLOUD CHANGES THE NATURE OF “BUSINESS CRITICAL” APPLICATIONS

The reliability of cloud-based services continues to be one of the reasons CIOs cite for not moving to “the cloud.” The inability to predict the availability of the underlying service coupled with many of the as-yet unknown reliability issues with cloud infrastructure, particularly those built on home-grown solutions rather than proven, established solutions, continue to be a challenge for cloud providers in convincing organizations that they are in fact stable enough for organizations to deploy critical business applications.

Given the potential impact on revenue due to downtime this should not be surprising. Consider this data from Assessing the Financial Impact of Downtime (white paper, PDF, a must read IMO):

cost-downtime

 

The financial impact of downtime on an organization is staggering, and there is very little accountability for such downtime in existing cloud provider agreements.

Stanley, as we know, wasn’t really bothered by the outages and choices made by Google – and yet the outage almost certainly resulted in costs that may not be accounted for. 

Reliability is a major IT concern, regardless of the data center model or location of that data center. The availability of applications – whether in a business critical path or not – is an imperative IT and cloud providers cannot ignore. In fact, it may be necessary as more and more “non” critical data stored “in the cloud” to redefine what is and is not a business critical application.

Consider the use of file and document storage as well as collaborative applications “in the cloud”. These are certainly not critical to the business in the sense that they directly generate revenue, but they can have an impact on the bottom line if they are unavailable deriving from lost productivity and wasted time. Certainly there is a cost associated with organizing and preparing for even “virtual” meetings and presentations that must be “eaten” by the organization when the virtual meeting place is unavailable.

The more we rely on cloud based services the more we need to focus on reliability and availability as a measure of effectiveness for those providers. It behooves cloud providers to ensure their environments are as reliable as possible -- which they certainly already do – but also that they recognize the growing dependence and importance of the services offered to the bottom line of their customers – organizations.

Stanley may not care about Google’s latest outage – or its next one – but organizations should and must care about the potential effects on its employees across a wide variety of functions and should begin considering what is/is not considered “proper business” use of cloud-hosted services; especially those that are free for use and come with absolutely no guarantees of availability.

Follow me on Twitter View Lori's profile on SlideShare AddThis Feed Button Bookmark and Share