Optimizing public cloud costs- dos and don’ts

by | May 28, 2020 | Blog | 0 comments

Realization of cloud economies lies in the way it is implemented and managed. With Cloud deployment models i.e. Private Public or Hybrid, most of the companies do not realize hidden cost when going cloud. These costs include the manpower to migrate to the cloud, size of instance, data center selection, OS and bandwidth costs etc. When comparing the cost of a cloud deployment, one must take into account the costs of power, cooling, administration, staffing, and data center real estate for deploying the same application in-house or in traditional hosting. But incurring cost in cloud pays back remarkable intangible benefits, increased agility, responsiveness and several other business benefits against which investment costs can be overlooked.

Now purely from public cloud perspective below mentioned steps can help reduce the recurring costs or bill. Selection of instances must happen purely based on the utilization requirements and this becomes critical to select the right instance type.

Various instance types based on the various cost models can be classified as:

  • On demand instances: High cost, for short term usage
  • Reserved instances: Low costs, for long term commitment
  • Spot instances: Lowest cost: for non-critical and short term usage

Note: Free Usage: Cloud provider like AWS also provides free instance hours as well.

Now, depending on the needs of your application, one can optimize your costs for instances by purchasing reserved Instances or Spot Instances. On-Demand Instances are a good option if you run Instances in less than 20% utilization; however, if you plan to run your Instances more than that, Reserved Instances can save you money. To obtain Reserved Instances, you make a low, one-time payment for each instance you want to reserve, and in turn you receive a significant discount on the hourly usage charge for that instance.

Besides selecting the right instance size and type, there are several toolsets that continually collects utilization, consumption, availability, capacity, and cost metrics for your monitored cloud resources: Compute (e.g. AWS EC2), Database (e.g. AWS RDS) and more. These toolsets continually seeks the optimum balance of available pricing options and potential cloud configurations in consideration of your historical usage pattern.

These toolsets must be deployed in the public cloud environment to create the balance between cost and performance. These tools provide recommendations f to determine the most suitable server or database instance size and compares actual cloud resource performance to sizing benchmarks that are synthesized from all the resources we observe. They also provide detailed downscaling recommendations and reports suspected under-utilized resources and unused instances. Hence, using the tool recommendations and right selection of instances, one can reduce the public cloud bills drastically.

By applying the public cloud provisioning best practices, methods and right toolsets for cost and performance optimization, public cloud bills cans and resource selection, it is sure that bill of public cloud services can go down.

Most likely, if the best practices are followed as mentioned below, public cloud environment turns out be business friendly, optimized and cost effective.

General

  • Design for optimized architecture that includes DR, Load balancing, NW transfer, Storage and backup etc. to support dynamic business requirements ensuring optimized performance and service levels.
  • Right instance size and type selection
  • Reclamation processes/de-provisioning processes for ensuring underutilized resources are not charged needlessly
  • Usage trending to keep track of historical usage
  • Forecasting tools to predict the future requirements and associated costs

Monitoring

  • Usage: Who is using what, where and when?
  • Performance: What is the utilization rate?
  • Cost: How much does it cost us?

Environment Optimization

  • Usage: Can we retire or reuse existing resources?
  • Performance: Can we size resources better (up or down)?
  • Cost: Can we pay less for each compute unit we use?

As mentioned above on cost and performance optimization concerns, organizations must remember that achieving the balance between these two is the key to create the cloud success story. Cloud economies can be best realized through getting the following aspects rightly:

  • Resources rightsizing
  • Price optimization
  • Deployment reallocation
  • Simulation & planning

Enterprises need a constantly updated and complete breakdown of their costs. They want this information so that they can be alerted to cost anomalies in real-time. So by setting up billing alerts, enterprises can help in setting up cost thresholds to avoid last minute surprises. Also they need to understand, allocate, and track the costs associated with their various tags. The information is essential for them to properly measure their business functions, measure their efficiency, and make informed business decisions

Lastly but most importantly, they need to ensure that they are not wasting money through over provisioning or poor purchasing habits. Every business faces a competitive landscape and inefficient or overspending puts companies at a competitive disadvantage.

Along with this, education regarding cloud spending is a must to endure cloud economies. Ease of provisioning is leading to issues like sprawl management. This brings down the cloud benefits in long run. Workforce must be educated around these aspects and there must be controls set up around provisioning process. These controls can be implement thought advance cloud orchestrators that have features like approval workflows, automated resource reclamation procedures, threshold management, self-learning for advance analytics etc.

Besides this, workforce must be educated on:

  • Clearly see and understand the way the company uses its cloud resources over time
  • Dynamically quantify the optimal provisioning safety-margin, and right size accordingly
  • Stay up-to-date with changing pricing plans and rates
  • Continuously choose the relevant price plan to match application needs
  • Include cloud economics metrics when engaging in capacity planning for new cloud applications
  • Use the most intelligent and actionable tools available to aid your cloud resource management process

Concluding

Every IT driven business is excited about the utility based consumption of IT services, thanks to Cloud computing. Word “pay-for-what-you-use” is making every CIO think about adopting cloud computing solution for obvious reasons. But real Cloud economics are only achieved by dynamically optimizing the balance between Operational and Financial objectives. Subscribing to cloud service is just the initial step in this journey. Simply over-provisioning until an intuitive feeling of comfort and safety is achieved is no more than fantasy cloud economics, and to achieve “perfect provisioning” companies need highly intelligent tools which can not only iteratively learn resource usage patterns, but can provide actionable recommendations for how to optimize cloud deployment by rightsizing and choosing the most economical pricing plans.

Hence, real cloud economics are only achieved by dynamically optimizing the balance between operational and financial objectives of using the cloud. This can be achieved through using the following techniques:

  • Real-time Analytics for view and control usage
  • Right-sizing for downscaling, resource optimization and avoiding over provisioning
  • Price Optimization for selecting right pricing plans
  • Planning & Simulation for having realistic view of future cloud service consumption

When above mentioned points are used together, these can deliver greater cost savings. Cloud service providers offer tools and support service to review single or multiple optimization scenarios based on predictive analysis of future spending. Hence, by taking the advantage of automated optimization services, Enterprises can maximize cloud investment and reduce the cloud spending at the same time.