October 3, 2013 | Written by: David Cox
Share this post:
Typically, the Thoughts on Cloud blog discusses the benefits of moving to the cloud, but when I ran across the following post “Why Some Startups Say the Cloud Is a Waste of Money,” I thought it would make for an interesting blog post topic. In this post, I would like to explore some of the reasons why people say you shouldn’t be using the cloud and then use these concerns to show how to leverage the capabilities of the cloud. Three of these concerns are cost, availability and performance.
Cost: There is a reason that car dealers are pushing leasing—they can take more of your money. This is the same in the cloud service environment. If you are only looking to host a computer on the cloud, in the long term, buying will be cheaper. Cloud makes sense if you are constantly changing workloads, upgrading systems or if you need the latest system. If you just have a static system that doesn’t change for several years, then buying and hosting your own server is going to work out in your favor.
Availability: If you are just hosting the same system on the cloud that you previously hosted locally, the cloud will be less reliable. You need to take advantage of the capabilities of the cloud to increase reliability. The abilities to spread a system across multiple geographies, monitor and provision systems based on workload, and have multiple systems clustered together can all increase reliability. If you aren’t doing this in the cloud, then you’re just adding another layer, a virtual machine (VM), which increases your chances of system failure.
Here is an example of how this would be calculated. Let’s assume that all components (server, VM and application) have a 99 percent availability. If these are all running on one system, the overall availability would be 97 percent (0.99 x 0.99 x 0.99). If you take out the VM component, you can achieve a higher availability of 98 percent (0.99 x 0.99). Cloud does have the advantage of easily spinning up new systems that can be clustered together. If you have two parallel systems, the availability would go up to 99.9 percent (1-(1-0.97)2 ).
Performance: Quality of service can be a critical requirement for many applications. For applications that are very calculation-intensive, hypervisors may impair the overall performance. Additionally, other systems running on the same hardware can influence the performance of other applications. Running these types of applications on standard VMs may not be the optimum architecture choice.
These types of applications should be run in a hybrid environment, with standard VMs running non-computational-intensive components like web servers and bare metal servers running computational-intensive components like database servers. SoftLayer is one of the few cloud offerings with this type of flexibility. SoftLayer provides the capability to manage bare metal builds and traditional VMs through the same cloud portal.
Hopefully, this blog post has been successful in prompting you into action to implement some of the more advanced features in the cloud. Is your company implementing cloud and reaping the benefits, or are you just following the crowd into the cloud? I look forward to hearing your comments.