By Daniel Galecki
The cloud computing market reached $121 billion in revenue last year (2015) and is growing at an annual rate of 26%. Public cloud offerings provide an attractive value proposition. They allow IT to get out of the business of running data centers, where for years most of the budgets have been spent on hardware and software maintenance and upgrades. Instead, IT can focus on innovation, looking for better ways to support the business.
Cloud-based Windows or a Linux servers can cost less than $1.00 USD per hour – even after a year, the cost of that server is less than $8800 USD (£6,152). Considering that cost in contrast to the hardware, space, power and cooling and other costs that the cloud service is replacing, a dollar per hour is quite inexpensive. The same comparison applies to SaaS applications, like CRM applications, that cost as much as $64 (£45) per user per month (or about $2.13 per day). You can have your CRM application in the cloud with no hardware and software maintenance or upgrades to worry about.
There is no question that Cloud services provide real benefits.
In Infrastructure as a Services (IaaS), you have the ability to only incur costs when you need to, which is great for test workloads – you don’t need to buy a permanent piece of hardware and have it sit idle most of the time. You simply turn on the cloud instances (typically virtual machines) for the amount of time you need it. When you turn the instance off, you stop paying for it. However, we are not used to turning virtual machines off – which means oftentimes companies pay for more cloud service than they actually need. The cost of the wasted usage appears small – pennies per hour. But, because a large organisation may have hundreds or thousands of these…the pennies add up quickly.
For many production loads, the machines always need to be on – so, is there a way to save money here? Yes. Depending on the Cloud provider, you may have access to subscription pricing (e.g. Amazon EC2), that gives you lower prices for “reserved” instances. The savings compared to pure on demand pricing can be significant – 40-50% over time, so if you are running continuous production loads, consider looking for these pricing options.
Another way to save money is by ensuring that you are using the right type of virtual machine. It is tempting to get the virtual machines with lots of RAM and CPU, because “we may need that power.” But, do you always need all that CPU and memory? If not, consider using smaller machines. If your CPU or RAM requirements suggest short time spikes in demand, look for instances that allow for scaling up. In the long term it will save you money. And, if your company has reserved instances, make sure they are all in use before you subscribe to additional “on demand” instances – utilize the capacity you already have.
But, of course, just getting an IaaS instance doesn’t mean there are no other costs. With IaaS, you get a virtual machine with an operating system. You are still responsible for installing your software. Here, you have to make sure that you are allowed to install your software in the Cloud environment. For example, if you use IBM SoftLayer, you are not allowed to run Oracle DB in that environment, as it is only allowed on Oracle, Azure and Amazon cloud environments. And the licensing terms change, so keep checking as conditions may change over time.
You should also investigate whether you should use your own license for the software, or perhaps, it is more cost efficient to use a cloud instance that includes that software license in its subscription. In some cases, you can "bring your own software and license" (BYOSL) under programs provided by some software vendors. This allows your organization to leverage its existing investment in on premises software when moving to the cloud. In any case, if you are providing your own licenses for software that you are running in the cloud via IaaS, then you are still responsible for maintaining license compliance with those license agreements. Its no different, in that respect, from running the software in your own datacenter.
The benefits of Software as a Service (SaaS) do seem significant. Everything is contained in the subscription price; there are no software upgrades or infrastructure concerns. You subscribe to a capacity (per user, etc.) and simply use the software.
However, with SaaS, the challenges come when you closely examine the vendors’ rules regarding how the software is used. Some SaaS vendors require that you buy the same subscription level for all users, regardless of how much of the application they use. Since you need to provide power users the ability to do their jobs, this results in subscriptions being under-utilized by other, casual, users. Even if you have the ability to have different levels of access, how do you make sure the right subscription level is associated with the right user? After all, users will complain if they don’t have access, but won’t let you know when they have too much access.
Cloud services can provide real benefits, but, regardless of the type of Cloud services used, organisations need to consider employing software license and subscription optimization processes and technologies to optimize their investment in cloud services and software. These tools will help identify idle and under-utilized instances and subscriptions, and identify license compliance issues. This will help you maximize your return on Cloud services investment, whether in IaaS, SaaS or other types of cloud services.