How IronScales Reduced Kubernetes Costs by 21% With Automatic Pod Rightsizing
About The Customer
IronScales is trusted by over 10,000 global organizations to protect their inboxes from Phishing, BEC and QR code attacks. Using a combination of AI and human insights, the company’s platform empowers IT security leaders to spend less than 6 minutes each day on email security. But under pressure to increase economic efficiency, their own 3-person DevOps team was forced to spend dozens of hours each month manually right-size Kubernetes resources.
GlobalDots selected a solution that automatically adjusted CPU and Mem requests of Kubernetes pods during runtime – helping them reduce cloud costs and spend less time on manual adjustments each week.
The Problem: Manual Rightsizing to Manage Mini Peaks
The company’s kubernetes pods were routinely over-provisioned. They had strict SLA targets to meet, and without the ability to adapt resources in real-time, this meant they either risked breaching SLAs or used more resources than required.
Moreover, their cloud instances used a static configuration, which made managing sudden peaks in traffic difficult. When the company onboarded a large client or conducted a Proof of Concept (POC), the system would be overloaded, leading to errors and slowness. The DevOps team would have to manually update the resources every time this happened.
Under pressure to increase economic efficiency, the DevOps and Dev team began manually rightsizing instances and applied horizontal auto-scaling – which enabled them to achieve impressive compute optimization and cut compute expenses by 25%. But these efforts were ultimately a band-aid, not a cure, because they were not scalable or applicable for the long-term.
The DevOps team consisted of just 3 professionals, and the manual approach took up dozens of man-hours each month – and when they undertook an end-of-month analysis, it became clear there was still a lot of waste and Kubernetes inefficiencies.
The Solution: Dynamic and Automated Resources Optimization
Understanding that an automated solution was required, the team followed GlobalDots’ suggestion to solve the problem. Our experts saw that the company needed to simultaneously avoid over-provisioning and meet its SLA targets. The solution our experts selected ensured both could be achieved.
The platform simplifies Kubernetes resource management, dynamically scaling resources to match real-time demand. It continuously adjusts CPU and Mem requests of Kubernetes pods during runtime to not only improve efficiency, but also eliminate out of memory (OOM) and CPU errors. This enabled the DevOps team to optimize their resource usage while spending less time manually rightsizing.
Better still, the solution featured a transparent dashboard to easily analyze cost clusters and set alerts for budget deviation. This empowered the DevOps team to demonstrate the cost savings they were making and proactively manage their spending.
The Results
- IronScales cut kubernetes costs by 21% while reducing SLA breaches by 13%.
- DevOps and senior developers spent 18% less time manually adjusting the system.
- The application’s latency was reduced and performance improved.
Save Time and Cloud Costs with GlobalDots
IronScales’ push for economic efficiency led to a dramatic improvement in its overall cloud management. And many companies could see similar concrete gains simply by adopting the same solutions.
Want to learn how it could make your Kubernetes resources more efficient?
Related Solution
-
Cloud Cost Optimization Solution
Stay ahead of the evolving FinOps technology landscape, control rising cloud costs, and gain deep insights into your costs’ context.
This post first appeared on Web Performance, Security, CDN And Cloud Computing, please read the originial post: here
