How to optimize thousands of resources
with one platform
Unused & Orphaned Resources
The eToro trading and investing platform boasts an impressive user base, with over 33 million registered users from more than 100 countries. Being a B2C company, their usage patterns exhibit significant hourly fluctuations.
Their technology stack encompasses a variety of components, including K8S, legacy systems, databricks, data lake, .net, and more. Additionally, they employ hundreds of distinct services that run on the Azure cloud, extensively using both PaaS (Platform as a Service) and IaaS (Infrastructure as a Service) solutions.
Approximately half of eToro's usage relies on virtual machines, while the remaining portion heavily utilizes managed PaaS services such as databases, service buses, event hubs, application gateways, firewalls, and others.
eToro's applications are deployed across over 30 clusters, equipped with thousands of cores, and are spread across various regions and environments. The utilization of eToro's virtual machines is exceptionally dynamic, making it challenging to manage and optimize efficiently. The usage can swell or shrink within minutes, contingent on stock market volumes.
Recognizing these complexities, eToro sought centralized solutions for cost optimization, reserved instances management, and waste cleanup. These challenges encompass multiple applications and departments, all laser-focused on cost reduction while ensuring zero downtime and minimal operational changes.
eToro identified CloudHiro as a viable solution due to its comprehensive cost optimization features tailored specifically for Azure.
Incorporating K8S, CloudHiro's cost visualization tool, eToro could easily pinpoint the most expensive and inefficient deployments. This allowed their R&D team to concentrate on improving memory and CPU utilization, resulting in more efficient scaling.
To tackle the issue of unused resources, eזoro leveraged CloudHiro's WasteMaster. With this tool, they were able to identify over a hundred distinct types of resources that were not being utilized, along with the associated periods of inactivity. By implementing CloudHiro's predefined rules and automated policies, they established an efficient mechanism for resource cleanup.
While reducing their reliance on Virtual Machines, partly achieved through K8S optimizations and other efficiency measures, CloudHiro's RI Wizard enabled eזoro to maintain a high VM coverage. Remarkably, this was achieved while increasing their Effective Savings Rate (ESR) to nearly 50%, even as their VM usage experienced significant reductions.
*ESR = Effective Savings Rate - savings gained by using financial instruments such as reserved instances and savings plans
In a very short period, eToro experienced a substantial cost reduction and achieved greater efficiency in managing their workloads. This reduction in costs occurred without any downtime, enabling them to expand more rapidly than ever before.