Challenge

Prasad Purandare is building an AI startup for image identification in stealth. As a bootstrapped startup, controlling budgets and cost runoff is critical to ensure the success of their journey. 

As the startup started building and training ML models, they exhausted their AWS credits reasonably quickly. The startup was searching for a way to get a grip and cut its monthly AWS bills. 

They searched for an affordable, agile solution that required minimum effort from their side to reduce their costs. This is when they discovered Cloudshim and signed up as one of our early customers.

Solution

Cloudshim worked with Prasad and identified the main contributors of their cost were underutilized instances. The first step was to ensure EC2 instances were rightsized. They used the Cloudshim Cost Savings recommendations that evaluate CPU Utilization, Network & EBS activity to determine whether to upsize or downsize the instances. 

In the next step, they considered the activity in these rightsized instances by evaluating the heatmap of the CPU utilization over different periods. The startup has remote workers that log in at flexible hours and perform various tasks: labeling, training, testing, etc. With Cloudshim Resource scheduler, they could apply automatic Start/Stop schedules across all non-prod instances.

Results

The startup realized savings of 60% by Rightsizing and scheduling underutilized instances. With its budget back under control and a good understanding of its resource usage patterns, the startup is no more worried about its cloud bill.