We are all familiar with the triangle of price, quality and schedule. The conventional wisdom is you can have any two out of three -- but rarely can you achieve all three. If you want it fast and cheap, it isn't going to be very good. Add a strong desire for quality into the equation and keep the time frame short, and costs are likely to soar.
Enter the "Cloud." This is a latter day timesharing, service bureau or remote computing model (for those of us who have been around a while) with a bit of a twist. The platforms, tools and software systems available today truly enable computing power to be delivered as a utility to anyone, anywhere and at any time.The cloud presents an opportunity to push routine or commodity services out the door, reducing the burden on internal resources, increasing agility, flexibility and predictability of costs. The cloud makes it easy to scale up and scale down as needed, and can free limited internal resources to focus on higher value projects.
But with these benefits come new concerns. The top of the list is always security. How can we be assured our data will be safe from theft or misuse when we put it in the hands of some outside data center operators? Clearing that hurdle brings us to other make or break decisions. Does the provider offer the platforms, API or software tools that are needed to run our applications and support the business? Can I meet all of my regulatory obligations?
Assuming we can identify a player or players that meet these and other necessary criteria we eventually turn our attention to cost.
Here is where it becomes really interesting because costs are not always that predictable. For Software as a Service (SaaS) we can apply the rates per user per month and have a reasonable estimate of the cost. However, this is not as simple in other types of cloud services such as Infrastructure as a Service (IaaS) where costs will depend heavily on the actual resources consumed by our application.
We can estimate resource consumption and, using published rate charts, try to predict the cost. But it turns out this may not be a fair basis for comparison. Research shows not all clouds are created equal. A series of benchmarks run by Krystallize Technologies demonstrated the same workload run on identical machines provisioned at different cloud service providers will yield different performance levels. This would suggest we will can expect different levels of performance for our application depending on its characteristics and the provider we choose.That was not too surprising given different providers will have different equipment, architectures and design. What was very surprising was the performance varied within provider. The representative work load executed on several identically provisioned machines at the same provider also yielded different performance levels. Moreover, these performance levels varied significantly over time. Keep in mind these cloud service providers operate data centers that are in a constant state of change.
While there are plenty of tools to simulate workloads, monitor the performance of an application or the network, and monitor costs, Krystallize CloudQoS™ provides the kind of visibility into the cloud that no other monitor delivers. You will be able to detect when the "sand" under the platform has shifted.
Whether you are first choosing a cloud service provider, managing an existing provider or just trying to maintain a quality of service, having the ability to measure the true performance of the platform supporting your application will be essential.
With the proper visibility, you may be able to rest easy knowing your cloud service will remain as fast and cost effective, and will continue to provide the same high quality service throughout the life of your application.
Captain Joe
Follow me on Twitter @JPuglisiLLC
