Cost Optimization Assessment
Whether you’re migrating to the cloud or are already there with your enterprise-level applications, understanding application demand and cloud capability is key. CloudQoS enables you to identify the ‘Best Fit, Best Cost’ baseline of cloud capability tuned to your services. Understanding what Cloud and what specific cloud configuration to use can have massive impacts to your cloud and associated software costs.
For example, a leading financial service company recently used CloudQoS® to achieve an overall Cloud cost-savings of 43%. The client used CloudQoS® to baseline a number of their Oracle based applications and then synthetically test these workloads in multiple cloud environments and configurations. After testing the workloads, it was determined that pruning low-performing virtual machines from their existing cloud service provider and using only the highest-performing virtual instances would lead to a 43% cost savings. The client ultimately leveraged workloads across a smaller set of machines, leading to less costly virtual cores and associated Oracle licenses.
Cloud instances are regularly over provisioned with very low utilization, and sizing decisions being made through relatively arbitrary decisions based on ‘previous numbers of CPU’s’ or software vendor recommendations (which in turn are often overstated and/ or out of date). The best way to size is to test your application performance for fit on a number target state cloud(s) and configuration(s). Historically this was complex and would take significant cycle time and in some instances, negating the savings. However, now with Krystallize Cost Optimization Assessment, that leverages CloudQoS, we can rapidly profile your servers or applications to create a synthetic replica that can be tested to find the most optimal location to run that service. Typically, the Krystallize team will advise on a base set of configurations to test (based upon the observed performance) and then examine fit based on performance and cost. Our approach creates a uniform measure that can be applied across all clouds (even bare metal).
Working with the Krystallize team allows for better decisions based upon future growth – the world of multi-cloud is complex with many millions of configuration parameters. Choosing the right SKU can save significant costs especially when growth is taken into account. The figure below illustrates the overall impact that choosing the right SKU has on overall cost. When selecting an optimal size for individual working nodes, the AWS r3-large, though individually more expensive than the alternatives, illustrates a much more cost-effective means to address operations per hour. By requiring fewer nodes, the overall cost can be significantly reduced when compared with the smaller and cheaper alternatives.
Taking advantage of Cloud fluctuation:
Whilst the proven variability in the cloud can be a problem and have negative cost implications, the inverse is also true. By understanding that there are high performing cloud instances out there, it is possible to reduce your instance sizing and gain more performance. Quite simply a fast-performing small instance may be far more performant than a poor performing large instance. An engagement with a Fortune 500 Financial Services Transaction Processing Company demonstrated that more than 50% of their estate failed to meet an adequate performance threshold that was derived by application requirement. By using CloudQoS we were able to ‘prune out’ these poor performing cloud instances and only choose high performance instances. This allowed this client to save 45% savings from their IaaS costs and improve the consistency of their clearing and settlement processes.