In our last blog I said to watch for our guest blogger Randall Becker. As much as I would like to take credit for his writing, this is Randall’s entry. I hope you enjoy it as much as I did.
What is “The Cloud?”
Put simply, a computing cloud is a set of resources that run one or more applications without concern as to where the CPU cycles or data actually come from. Using a cloud rather than individual computers and disks enables “elasticity” of your computing assets.
Rather than allocating the maximum amount of CPU and storage resources that a business process might need at a peak,
cloud infrastructure allows you to minimize resource costs by only buying the resources you need, when you need to use them, by sharing your computing assets (figure 1).
In most computing environments, not all applications need the maximum amount of resources at the same time. For example, company email on a major holiday will be at a minimum, while the ATM and credit card applications might be at maximum. CPU cycles can be “stolen” from the email application and given to the ATM and credit card applications. Similarly, end-of-month processing might need a lot of disk space and CPU cycles only while it is running, and otherwise is dormant.
There are private clouds and public clouds. A private cloud is a set of storage and computing resources that you own and manage, while the resources in a public cloud are owned and managed by someone else. Typically, you pay only for the resources that are actually in use over a specified period of time.
This blog is about public clouds and how your data can become lost in the fog—in a large variety of ways.
Service Levels
If you want to use a cloud service, you make a list and go shopping. Maybe you’re looking for computing capacity, a software application, or just storage. In any event, you’ll end up signing an agreement that looks a lot like a software license agreement, and unfortunately is about as useful. That is, it is full of caveats and disclaimers, and us often subject to change with little notice or negotiation.
“Reliability” service levels are rarely specified, and even if they are, “five nines” is still over five minutes of downtime a year and usually doesn’t include scheduled downtime for maintenance (figure 2).
What Are Some Service Level Risks?
Threats that can affect access to public cloud infrastructure include war, asteroids, earthquakes, tsunami, hurricanes, and children. Let’s take a closer look at each of these risks.
Unlike business, where war has economic impact only, war risk to cloud infrastructure is high. We’ll examine this in detail in the next post and you might be very surprised at what the risks are.
Asteroids are very much out of our control. Their risks can be assessed in terms of probabilities and severity, but not location. Fortunately for us, the big ones are not common and as far as we know, the dinosaurs were not using The Cloud when they were wiped out. While earthquakes, tsunami, and (to a lesser degree) hurricanes cannot be predicted with any accuracy, we are learning rather rapidly about their effects, the risk zones, and how to mitigate risk through the use of geographically dispersed data centers.
As for children, this risk is part of the overall category of “human error.” Somebody changes a setting on a computer and suddenly nobody in North America can access his or her data. If you have access to manage your infrastructure from home, do you lock your PC before leaving it even for a minute? How quickly could a family member take down your data center by accident? What about a cat walking across your keyboard – Do you know where that unexplained online salmon order came from? The Uptime Institute estimates that human error causes roughly 70 percent of the problems that plague data centers today.
Hacking can be included in the human error category as well if the intent was just to steal data and not take down the infrastructure.
Other risks might include:
- Internet backbone outages
- Denial of service attacks
- Cloud company goes out of business
- Service incompatibilities
- Economic conflict between countries
In our next blog, we’ll introduce the imaginary countries of Clientada and Servia and explore how access to your public cloud can be lost, through no fault of your own or from a failure of your cloud provider. This is a study of international intrigue at its finest, so stay tuned.
CEO and founder of Nexbridge Inc., Randall Becker has been providing software solutions since 1983 across industries for organizations needing the next level of reliability where downtime is unacceptable. He has focused on the HP NonStop™ Server and UNIX platforms since that time and his involvement has included major infrastructure projects for financial, retail, and law enforcement sectors. An expert in Indestructible Computing and Software Development Lifecycle (SDLC) methodologies, Randall has been involved in all aspects of NonStop™ solutions development. More information is at http://www.nexbridge.com/
