Both enterprises and SMBs report that one of the greatest challenges of launching a cloud project is overcoming misconceptions about the cloud. False expectations often lead management in the wrong direction, hamper early planning and distort the entire project, making cloud success unlikely. There are many myths about cloud computing, but these four are the most harmful to a successful
1. The cloud is always cheaper than internal IT.
Non-technical business managers and executives often hold tight to this disputed myth, and that has led to some truly disastrous cloud projects. The price of cloud services ranges from as little as a quarter that of internal alternatives to three or four times more expensive than maintaining a fully on-premises data center. Applications that require large stores of data, access and update database management systems (DBMS) in massive numbers and require tightly constrained response times likely will be much more expensive than uninformed adopters would hope.
For most cloud users, operations and support costs are more likely to make the case for private cloud than server or hardware costs.
And this ugly myth has spawned a “sub-myth” that server reductions of scale create the greatest argument for cloud. With blade servers costing around $1,500, it’s a hard myth to understand, particularly when multi-programming and virtualization allow businesses to reuse their servers with increasing efficiency. For most cloud users, operations and support costs are more likely to make the case for private cloud than server or hardware costs.
2. All clouds are built by connecting multiple virtualization-equipped data
This myth is likely perpetuated because virtualization-based cloud strategies like OpenStack receive major publicity and Infrastructure as a Service (IaaS), the most common cloud service, is essentially a virtual machine (VM).
In reality, if resource efficiency is the key to cloud success, then virtualization may be a bad place to start. Platforms that can properly secure VMs and application hosting at the application component level without the need for additional software will be more efficient. Microsoft Azure and a series of other platforms based on Sun’s Solaris operating system are good examples of this.
The notion that a cloud simply consists of multiple data centers is another sub myth. Cloud efficiencies are created by pooling resources, but there isn’t a direct correlation between an increase in size and efficiency. The relationship is actually opposite; efficiencies diminish as sizes increase. You may pay more by creating a multi-data center cloud than you would save.
Most successful cloud providers are already moving “up the stack” toward higher-level services like vertical or horizontal application Software as a Service (SaaS). The key attribute of any Platform as a Service (PaaS) or SaaS is multi-tenant support, or the ability to isolate users as though they were running on independent machines. While virtualization accomplishes this, there are other multi-tenant architectures, and the management and security of even virtualization-based clouds still require multi-tenant support. Look for that feature first.
3. Private clouds require special IT architectures to become efficient.
Many companies hope to see major gains in efficiency after deploying a private cloud. The economics of cloud architectures show that data centers that combine servers and storage into resource pools based on a common technology will have similar overall efficiency.
Virtualization alone will improve efficiency versus isolated servers, and multitasking or multi-threading OSes will efficiently use CPU and disk resources. The fact is most companies will not gain much in terms of efficiency even if they fully re-architect their data centers on a cloud model.
There are many cloud architectures and different business models. Before you look for a match, assess how much you should expect to gain.
A complicating factor here is that there is no single cloud model. While it’s noted in the second myth that most people think of private clouds as linked, virtualized data centers, this approach is only suitable for businesses that plan to consolidate IT operations based on many discrete servers. Companies with large, central computers will not find virtualization helpful at all. There are many cloud architectures and different business models; before you look for a match, assess how much you should expect to gain.
4. Cloud standards are not ready.
IT continues to advance at breakneck speeds; standards deliberations do not. That means standards in IT are never really ready.
Will the lack of fully mature cloud standards compromise its use? The truth is probably not. If you’re planning to build applications that are tightly coordinated between your data center and the cloud -- or between cloud providers -- cloud standards could become a roadblock that will impact your planning, but they likely won’t derail the entire project. You’ll probably have to tune your problem management and fault isolation practices for cloud applications, but that would be the case whether or not there were standards. Security standards for the cloud are clearly in their infancy, but how many data centers are secured in a standards compliant method?
Like all projects, cloud deployment can only succeed when expectations are realistic. If senior management’s views of cloud computing don’t match the experiences of IT admins, companies risk losing project momentum and support. IT organizations can’t focus on genuine problems when they’re swatting cloud-mythology flies along the way. Before you start a cloud project, plan an educational briefing of key IT teams and departmental or operational personnel to set the cloud record straight. That will align key people on accurate expectations and debunk common myths of the cloud.
ABOUT THE AUTHOR
Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982.
This was first published in January 2012