Contents |
Introduction
Although we seem to be moving somewhat beyond the hype cycle that defined “cloud” for the past couple of years, we have yet to move beyond significant levels of skepticism from CIOs about moving mission critical applications to the cloud. Many see challenges that have yet to be overcome and questions that have yet to be answered. By their nature, many CIOs are risk-adverse, charged with maintaining a well-oiled machine. While CIOs need to embrace some strategic risk in their efforts to move forward, four major issues keep many CIOs from embracing the risks – and opportunities – associated with cloud. Of course, other factors exist, but these are the four that I see relatively often and that need attention.
Outages
On February 22, 2013, Microsoft’s Azure cloud platform suffered what was termed as a “worldwide service outage” apparently due to an expired SSL certificate.
On December 24, 2012, during the height of Christmas season, Amazon Web Services (AWS) suffered an outage in its Elastic Load Balancing service due to developer human error. This outage affected NetFlix, one of the Web’s most bandwidth-intensive and popular properties. Other AWS outages affected many popular Web sites at various points throughout the year.
While these are just two outage stories, they are not the only ones out there. No system will ever be able to achieve 100% uptime, particularly a system that includes human beings at some point in the equation. People can and will make mistakes from time to time.
Unfortunately for cloud providers such as Amazon and Microsoft, their outages are both public and big news thanks to Twitter and news outlets looking for a scoop. And those very public outages can take a toll on CIO confidence in the entire market, leading CIOs to take what they believe is a more risk-friendly option of keeping critical services in-house. Enterprises themselves often have their own availability issues, but there is a sense of control inherent when one owns a problem as opposed to simply entrusting resolution to a distant provider.
There are ways to avoid the impact of service provider outages, but these methods are still not as well understood as the traditional arsenal of clustering and a standby generator.
Skills
One of the reasons that clouds are sometimes considered too risky for production is that IT professionals lack an understanding for how the services work. TrainSignal recently released a post outlining the top 10 jobs that will be in demand in the cloud era. In that article, TrainSignal states, “The cloud is drastically changing the way IT departments function, and there are a whole new set of skills IT pros need in order to stay competitive. After all, the cloud is projected to grow into a $240 billion industry by 2020.”
Cloud uptake – or lack thereof – is following a similar pattern for other upheavals that have occurred in earlier technology paradigm shifts. As these shifts happen, technology pros need to retool and change their thinking to accommodate the new technology and the way it works. We’re still very early in this migration, and a lack of skills is creating some trepidation about the opportunity.
This factor is very easy to fix. Get people trained, read and learn, and work with peers to gain a comfort level with the services that are available. Test services with free trials.
SLAs
In general, cloud provider service level agreements aren’t all that great and certainly don’t lend themselves to high levels of optimism. Most cloud providers will say that they offer a “financially backed SLA.” It’s only when you start to look at the details that you discover that the financial backing is often in the form of a service credit up to, but no greater than, the month’s charges. For organizations that lose millions of dollars per minute of an outage, saving $5,000 on that month’s cloud charges isn’t going to much consolation.
On the flip side, cloud providers can’t assume all of the financial risk, either. They have to make money to stay in business, and some offer geographically separated zones of service that are theoretically able to provide a high level of availability… except when they forget to renew their SSL certificate.
As standards between providers develop, organizations will be able to spread their risk not only between different zones on a single provider, but between providers as well. Of course, this kind of availability comes at a cost to customers.
Cost uncertainty
Ok, quick… how much data do you intend to transfer between your file servers and your storage array this month? How much of that will be incoming vs. outgoing traffic? As customers look at the various calculators that providers make available, it becomes apparent that, in order to even begin to understand how much a cloud service might cost, CIOs need to gain deep insight into the comings and goings of their data. Or, they need to do some trial runs with a subset of data and attempt to extrapolate how much that service may cost.
Even though cloud offers a lot of opportunity, CIOs cannot skip the ROI portion of their responsibilities. While in some situations the organization may choose to go with what might be a service that carries more direct cost, it may, for instance, be reducing capex or taking advantage of a benefit it otherwise wouldn’t have.
But, in some way, an ROI and cost analysis needs to be done and the costs for cloud can be complex and even intimidating, leading some CIOs to continue to go it alone.
Action Item: All of these factors have solutions, and those solutions boil down to further education about cloud provider offerings. Most providers offer small instances of their services for free ora free instance for a trial period. CIOs should charge someone with some deep testing to determine the potential fit for cloud solutions and begin to understand what it takes to keep cloud-based services available and how the services are charged.
Footnotes:
