Pinterest encountered a significant performance issue during the migration of its search infrastructure, Manas, to Kubernetes, where one in a million search requests experienced latency spikes. The investigation revealed that cAdvisor’s memory monitoring processes were causing excessive contention, leading to these delays. The team resolved the issue by disabling a specific metric in cAdvisor, allowing them to continue their migration efforts without compromising performance.

The article discusses the importance of tuning Linux swap settings for optimizing Kubernetes performance, particularly in environments with limited memory resources. It provides detailed insights into how swap can affect application performance and offers practical recommendations for configuring swap to enhance Kubernetes workloads.

The blog discusses the introduction of configurable tolerance settings in the Horizontal Pod Autoscaler (HPA) for Kubernetes version 1.33, allowing developers to define how aggressively the HPA should respond to changes in resource demand. This enhancement aims to improve application stability and performance by allowing more fine-tuned control over scaling behaviors.

GKE Data Cache is now generally available, enhancing Google Kubernetes Engine's performance for stateful and stateless applications by utilizing high-speed local SSDs as a caching layer for persistent disks. This solution provides significant improvements in read latency and throughput, making it easier to manage data access while potentially lowering costs. Users can configure caching for their workloads with straightforward setup instructions and options for data consistency.

By implementing a php-fpm-exporter in a Kubernetes environment, the author identified severe underutilization of PHP-FPM processes due to a misconfigured shared configuration file. After analyzing the traffic patterns and adjusting the PHP-FPM settings accordingly, memory utilization was reduced by over 80% without sacrificing performance. The article emphasizes the importance of customizing configurations based on specific application needs rather than relying on default settings.

The webinar focuses on how to effectively deploy scalable SQL databases on Kubernetes, discussing key strategies and tools that enhance database performance and management in cloud-native environments. Attendees will gain insights into best practices and real-world use cases for leveraging Kubernetes for database scalability.

The blog post announces the release of etcd version 3.6, highlighting its new features and improvements aimed at enhancing performance and reliability for distributed systems. It emphasizes the importance of etcd in Kubernetes and other cloud-native applications.