Cloud cost optimization: 10 ways to save
Cloud technology allows scalable on-demand sharing of resources and costs among a large number of end-users. It enables end-users to process, manage, and store data efficiently at very high speeds.
Despite these numerous benefits, many enterprises are facing spiraling costs, and they’re looking for ways to reduce their cloud spend. Where can you find cost savings to make your cloud investments deliver on their potential without being cost hogs?
The two common pricing models for cloud computing are fixed and dynamic pricing. The more traditional model is fixed pricing, in which you buy a static amount of resources (compute, storage, networking, etc.) and project its resource needs for 3-5 years. This model has the potential to commit funds for capacity that isn’t used. (Server virtualization in the data center illustrates why by identifying underutilization of resources.) With fixed pricing, you pay for the resources whether you’re using them or not.
Dynamic pricing attempts to bring this to a more consistent and cost-effective approach—like a power company where you only pay for what you use. While pay-as-you-go can bring costs down, it does require accurate analysis of your usage of compute, storage, and network bandwidth. The pendulum can swing from cost-effective to costly if you’re not aware of peaks and idle times so that you can shut down applications and services when not in use.
In the fixed pricing model, you leave services running 24/7. In a dynamic cloud-pricing model, understanding when to turn off unused services is crucial to capturing the true cost savings.
Whether you’re in a fixed or dynamic pricing model, controlling costs is crucial and should be an ongoing effort. Below are 10 ways to find cloud cost savings now.
01. Conduct performance analysis
In many cases, switching to dynamic or pay-as-you-go pricing will save money. But to achieve those savings you need to utilize the cloud service provider’s (CSP's) reporting features to conduct an in-depth performance analysis. While the CSP data will tell you how well the application is performing in the cloud, it doesn’t measure the end user's experience. You will need to identify and configure your own reporting tools to focus on those performance metrics that mean the most to your users.
Your client service level agreements (SLAs) tell you what to measure. Everything in a data center or cloud deployment may be blue, but the end user’s perception of their experience is that performance is horrible. We’ve seen this perception cost companies significant dollars to address.
Once a proper analysis of application services has been done, the usage, resource peaks, etc. will show a clear picture of whether the application will be cost-effective in the cloud. For example, an application could could cost more in the cloud than the legacy-dedicated models based on the characteristics and makeup of the application's services.
02. Implement continuous monitoring
CSPs offer quality cost optimization tools, but it is the I & O team that should have the real insight into your environment. They should continually monitor what servers are being built and to what specs. The tools and reports you put in place for analysis form the backbone for continuous monitoring. Set up alerts and automated actions to ensure performance and control costs.
03. Invest in automation
With the wealth of data captured by your monitoring, take advantage of the auto-scaling and other automation services provided by your CSP. Make sure that the automation you configure scales down as well as scales up.
An example is development systems that are used only during the workday. You can set up automation to turn off these systems during nights and weekends so that you don’t pay for the 60 percent of the time when no one is using the systems. Automation can stand up the development environment for the start of the workday on the East Coast and power it down when the developers on the West Coast leave for the day.
04. Reduce excess storage
Storage utilization can be the biggest factor in driving up costs in the cloud. IT culture and business habits can push lots of duplicate data and create redundancies that will exponentially drive up costs. Simple adjustments in how data is copied and versioned will require reeducating staff on processes to make it more cost-effective when utilizing cloud storage.
Pricing models for cloud storage infrastructure include monthly fees for cost-per-gigabyte of stored capacity. Cloud storage must be optimized through use of the different storage tiers in the cloud as data ages in its life cycle. Make sure any provider has transparent pricing that clarifies each factor and rise in cost as usage grows.
05. Identify overprovisioning
The enhanced monitoring you’ve put in place to leverage dynamic pricing will identify where you’ve been overprovisioning resources. This is especially true when you’ve migrated applications from dedicated physical servers to virtual servers in the cloud. The average physical server supporting one application is underutilized in the data center. The widespread adoption of server virtualization demonstrates this. You may also see improvements in the CPU speed, the memory speed, and the data bus in your cloud environment.
06. Correct inefficient code
An example of inefficient code is having a web server run each line of a multipart query on the web server instead of taking the entire query and embedding it on the database server. The difference in performance is noticeable, and the reduced costs from limiting network traffic is substantial.
07. Assign an inventory owner
One of cloud computing’s essential characteristics is on-demand provisioning. CSPs are more than happy to provision as many servers and services as you request. Assigning someone to track these requests and the utilization of the servers or services is essential to controlling costs and managing shadow IT.
08. Manage shadow IT
Getting control of the IT spending outside IT is critical. As long as different parts of your company can individually spend money for cloud and other IT services without involving IT, you can’t control cloud computing costs.
Set up a service catalog that lists the types of cloud services available and make the process for ordering and approving the spend quick and transparent. Be sure to incorporate auditing so you can track who is spending money on the cloud.
09. Review support contracts
Once you’ve moved to the cloud, you will have hardware that was supporting the application on-premises. As you retire the hardware, don’t forget to retire the support/service contracts for that hardware. You should also look at staffing in the data centers because the skill sets required to manage cloud applications are vastly different.
10. Bring your own licenseWhile CSPs are happy to rent you licenses to run your software in the cloud, why pay twice? Software vendors have incentives to keep your business, and many have updated their licensing to enable you to use licenses you already own in the cloud. There are a couple of big companies that mandate using their cloud offerings for their software if you’re moving off physical systems. Companies we’ve worked with have been biting this bullet while they figure out how to replace the software with more supportive vendors.
RELATED CONTENT | White Paper
Nine Strategic Guidelines for Application Workload Placement /
Migrating to cloud requires a mindset and plan for continued investment and development. A phased approach can help your organization by migrating those application services found to be cloud-ready and building processes around them. Take those applications that are not cloud-ready and invest in a restructured development life cycle that brings all services into a cloud-ready state. Whether fitting the service for private or public cloud, the new model will keep services agile and more cost-effective for the business going forward.
The initial investments in operational processes, agile development, and architecture will pay financial dividends on the cost-center side and also make your organization more competitive and capable of responding to rapid changes in customer demands.
Key innovations such as serverless technologies like Amazon Lambda could spearhead some of the biggest operational cost savings, but only if your organization shifts their application designs to this architecture.
The path forward
Shifts in day-to-day IT operations could be a culture shock, and this can make or break cloud goals. Combining the technical methodologies, business processes, and development strategies into a unified understanding of the cloud and its benefits to all will be key to your success. Companies that embrace cloud—and the operational and culture changes that go with it—will position their offerings for growth in a competitive marketplace.