How Purplle Saved More Than $200,000 on GCP In Just One Month

Purplle is India’s largest online beauty store that caters to both the beauty and wellness needs of people. It has a user base of more than 4 million users and is continuously growing and acquiring customers at a rapid pace. In addition to metros and tier one cities, Purplle taps into tier two and three markets that are major factors of their success.

Industry

Retail and Consumer Goods

Type

For Profit

Location

India

Challenge

With Covid-19 and the lockdown that followed, Purplle faced a huge influx of orders. They had to ensure a seamless shopping experience for their customers and this required a robust infrastructure which they are managing on Google Cloud Platform. In the midst of increased sales, their GCP bills were soaring high and the GCP console was really becoming a bottleneck for them when they carried out the activity cost optimization.

We have an year long engagement with OpsLyft. The results in the first two months itself are really amazing. I don’t know if I have come across any startup team that offers the customer support as they do.

Nilesh Garg

Engineering Manager

Challenge

Suyash and his team analyzed their GCP bills but couldn’t figure out where a major portion of their cloud bills was going as the GCP console was not showing them the full breakdown of their cloud costs. When Purplle reached out to OpsLyft, they found out the cause of this cost surge in an instant.

A significant portion of Purplle’s cloud bills was idle and unused resources. The lack of dashboards and alerts for idle and unused resources in the native consoles be it AWS Cost Explorer or the GCP console, has made it a pretty common problem faced by many organizations and with Purplle it was no different.

Solution

Suyash and his team tried Cost360, a cloud spend management platform by OpsLyft that analyzes the cloud resources in real-time to give software developers insights and suggestive actions for making cost-friendly infrastructure changes. The real-time dashboards in Cost360 showed Purplle the idle and unused resources they had in their account and the cost associated with them within a few minutes of them first using it.

With Cost360, they were able to identify what all resources were idle and unused such as compute engines which have less than the threshold of CPU utilization (you can set your own customer threshold using Cost360), VM disks that are not attached to any instance, etc. Not only this, Cost360 showed them since when were the resources idle, who provisioned the resources, the utilization pattern of the resources, when was the APIs last called on the resources and many other minute details.

In addition to getting what all idle and unused resources are there, Suyash and his team were able to get the cost of these resources and how much they can potentially save by optimizing these resources through Cost360. To make their job easier, Cost360 provided them with steps to take care of these resources so they don’t have to figure out how to troubleshoot them.

However, this is just the beginning of what the OpsLyft is going to help Purplle achieve. In the longer run, Purplle team needs a single system on the product that can do the following:

• Provide full-stack observability on GCP resources for cost, performance, and uptimes.
• Create and destroy sandbox environments on the fly based on specified user configuration of VM and storage capacities along with software modules that should provision the machines with.
• Create preemptible VM clusters in production environments and ensure a zero downtime for the applications by developing an intelligent cluster auto-switch mechanism to standard VMs in the events of unavailability of preemptible VMs

OpsLyft aims to deploy these solutions in production for Purplle by the end of Q4 this year.

Results So Far

The folks engineering team at Purplle with Cost360 have seen the following changes in their operations:

• DevOps team is spending zero bandwidth in identifying unused and idle resources and are able to take quick actions as and when they receive alerts based on the same.
• Engineering managers have access to state of art visibility over GCP infrastructure and are able to stay within their allocated cloud budgets on a per team basis.