Optimising Your Cloud Spending: Five Tips

August 29, 2019 in Cloud Computing

Optimising Your Cloud Spending: Five Tips

Not too long ago, organisations relied on on-premise infrastructure for their storage, databases, networking, software, and analytics. Costs were high, and maintenance was a prime factor of concern. However, organisations spanning various industries and measuring across all sizes have shifted their focus towards Cloud Computing since then. Firms are procuring cloud infrastructure and increasingly shifting their workloads to public clouds, while optimising their existing on-premise virtualised environments as well. Moving the workloads to the cloud helps in reducing costs only if proper optimisation of cloud resources takes place.

Cloud optimisation isn’t a simple case of reducing costs by simply picking the least expensive options. It depends on several factors, and the smart decision-making ability of how, when and where to host your workloads.  

Cloud computing enables us to minimise capital expenditure on setting up data centres, electricity supply for power and cooling, and IT experts for managing this infrastructure. The cloud offers advantages of economies of scale. The infrastructure being flexible and scalable, there’s a good match between needs and resources . A report from Flexera states that the top two priorities remain – moving workloads to public clouds, and optimising cloud costs. More than 50% of the enterprises are spending in excess of $1.2 million per year on public cloud in 2019. Much of the cost incurred is on the heavier side, largely due to resource wastage. Let’s take a look at the steps that can help optimise costs on the cloud:

Identifying and monitoring expenses

The most dominant reason that prevents the organisation from identifying and controlling wastage is the complexity involved, along with its pricing and billing. There are hundreds of price points to ponder upon and a staggering amount of line items that are beyond comprehension. Besides, there may be unfavorable price changes, and the introduction of new features and updates can make it nearly impossible to keep track. Therefore the primary need that arises is the identification of the problem areas and their controlling:

  • Identify unused instances and other on-demand IT services that are being paid regularly but seldom used. All those instances that are unhealthy or inactive should be removed manually or automatically. 
  • Identify excess capacity of on-premise virtual machines, so that correct instance sizing can be undertaken. Right-sizing may involve larger size, smaller size or a different set of instances to substitute the previous ones. 
  • Unattached persistent volumes of storage disks like Amazon EBS volumes can be deleted. Attached volumes do not tend to get deleted automatically and may take up unnecessary storage space and increasing costs.
  • Explore inexpensive alternatives – like going serverless instead of actual VMs, or reserved instances..

Implementing automating policies to minimise waste  

Cloud inefficiencies and excess capacities often lie unattended or idle, and costs pile up. Manual identification and checks are not feasible for large scale computing operations in organisations, but automation features can be leveraged to optimise costs:

  • Users can set instant alerting policies to avoid any compromise on data arising from misconfigurations or weak configuration of settings such as asset control list, etc . 
  • Automated checks may be scheduled on a daily, weekly or monthly basis 
  • Automation can be used to deal with more complex scenarios like Reserved Instance planning.  You can automatically match requirements with resources to reduce wastage over long periods.
  • Automation includes shutdown workloads after hours, rightsizing of instances, specification of expiry dates, among others.

Determining where to run workloads

Organisations looking to relocate applications from data centres to the cloud need to adopt a methodical approach, involving rational decision-making with respect to the portfolio of applications. The key criteria to be taken care of is whether the application can run on a virtual machine stack.  Sometimes migration or movement of workload involves a change in infrastructure while the business logic remains unchanged. While for others, changes may be made to the application to let the workload function in a virtualised cloud environment. The next critical area of decision making is to decide on the deployment model- public, private or Hybrid. The factors involved in such deployment include:

  • Privacy and security regulation
  • Unique technology needs
  • Agility and elasticity required of the workload.

Optimising for public, private and hybrid clouds

Under a public cloud, the vendor takes responsibility for developing, managing and maintaining the computing resources pool. It is the most popular choice for cloud computing since it offers vast solutions, and computing needs to suit all organisations because of its flexibility and scalability, reduced complexity and flexible pricing options. Private cloud is for the exclusive use of a single organisation via a secure private network. It falls relatively on the expensive side. Hybrid cloud, as the name suggests, is a mix of both public and private cloud solutions. You can distribute the workload between public and private clouds. Organisations may try to leverage the benefits of all three to suit their complex procedures or workflows.

Optimising Cloud operator offerings

To identify the right mix of resources and cost, organisations may need to run some load tests. Once it is determined what kind of tailored mix is required – whether public, private or Hybrid or a combination of these, users can pick from a selection of vendors that provide these services at the possible prices along with discounts and regular updates. Different applications and cloud workloads require varying levels of data storage. For this, firms can take the benefit of tiered storage classes. Most public cloud vendors offer many types of storage classes. Depending upon a particular cloud type the specs may differ. Costs may also be reduced by moving data (that is likely to remain unused) into cold storage. 

Organisations that are seeking the highest levels of optimisation may likely opt for multi-cloud architecture. It may require more effort in setup and maintenance; however, it boasts the potential to lower the cloud computing costs to a significant extent. An ideal state can be achieved by maintaining a balance between performance and cost by going ahead with multiple cloud or single cloud, single vendor or multiple vendors, tiered or untiered structures in clouds to name some. Orchestration (i.e. codifying the steps and processes involved in management and deployment of virtual workloads) accompanied by automation, ensures zero human intervention. This reduces scope for errors, wastage and inefficiency. The result? Predictable and reliable workflows in an organisation helping it attain higher levels of IT and corporate governance.