May 2025 Summaries

AWS EDP is a savings program for enterprise cloud users with significant AWS cloud consumption, typically over $1 million per year. It provides discounts on total AWS billing that increase with total spend and commitment period, typically 1 to 5 years. The discount is flexible and based on specific agreements. To qualify, companies must meet certain eligibility requirements, including a minimum annual commitment of $1 million or more. However, negotiating AWS EDP can be challenging due to forecasting growth, compute, and use demands, understanding workloads and AWS costs, taking into account AWS Marketplace applications, optimizing EDP targets, and considering alternatives such as Reserved Instances and Savings Plans. Committed spend in the cloud is risky due to forecasting challenges, high cost for changing requirements, vendor lock-in, and the need for flexibility. An AI-powered cloud cost optimization solution can help automate VM and workload rightsizing, detect unused components of infrastructure, and optimize configuration for best cost and performance.

Cast AI, the leading platform for Application Performance Automation, has partnered with Amazon Web Services (AWS) to help customers optimize their cloud environments more efficiently. As part of this collaboration, Cast AI has earned the "Deployed on AWS" badge, indicating that its platform is hosted on AWS and aligned with best practices. This designation brings added value to customers by making them eligible for the AWS Private Pricing Agreement (PPA) policy, which unlocks greater flexibility, procurement efficiency, and accelerated commitment consumption in AWS Marketplace. By deploying its platform on AWS, Cast AI empowers customers to automate cost, performance, and security optimizations-seamlessly, at scale. Organizations purchasing Cast AI in AWS Marketplace now benefit from PPA/Commitment Drawdown, Consolidated billing and simplified procurement workflows, Accelerated adoption of Kubernetes cost, performance, and security automation, and Continuous innovation aligned with AWS infrastructure and best practices. Cast AI is built to optimize AWS Kubernetes Workloads by continuously analyzing resource utilization, cost drivers, and security risks, then automatically taking action to ensure clusters are always right-sized, secure, and efficient. The platform offers key capabilities such as autoscaling, intelligent spot instance automation, real-time rightsizing of nodes and pods, built-in Kubernetes security, continuous performance tuning using advanced ML models, and more. By automating critical levers of cost, performance, and security, Cast AI enables cross-functional teams to work together seamlessly, aligning business and technical priorities while maximizing the value of every workload within the AWS ecosystem.

We've recently announced Series C funding to fuel Cast AI's innovation and product expansion, including the launch of Database Optimizer (DBO), an automated caching solution designed to streamline database performance and simplify operations. DBO aims to address the challenges of managing database performance and scalability by providing a seamless, no-code caching layer that automatically optimizes cache management and invalidation. With DBO, customers can experience reduced database load, enhanced application performance, and responsiveness, as well as improved query observability and scalability. Early adopters have highlighted the ease of setup and immediate performance improvements as standout benefits, with some achieving cache hit rates of 80-90%. As we evolve DBO, our roadmap includes enhanced support for complex queries and stored procedures, identifying inefficient queries and automatically improving performance, better query indexing, and expansion to additional database types.

The guide provides actionable enterprise Kubernetes best practices for resilient, secure, and cost-optimized deployments. It covers key areas such as resilience engineering, security hardening, resource optimization, GitOps-based disaster recovery, and implementation roadmaps. The guide emphasizes the importance of a phased approach, with each phase focusing on critical aspects of Kubernetes deployment. By implementing these best practices, organizations can achieve significant improvements in availability, cost efficiency, and security, including up to 99.99% availability for critical services, 40-60% reduction in cloud infrastructure costs, and up to 50% fewer security incidents.

GPT-4o-mini is a powerful generative AI application that offers fast and high-quality outputs across various workloads, but its frequent inference cost can add up quickly. In contrast, Alibaba's Qwen2.5-14B, an open-source alternative, provides comparable results at a significantly lower cost when hosted in-house. A switch to Qwen2.5-14B enables teams to support flexible LLM choices and take advantage of automated solutions like AI Enabler for deploying and testing models, as well as dynamically routing requests for cost and performance optimization. Benchmark tests revealed that Qwen2.5-14B is 2.3 times less expensive than GPT-4o-mini at full capacity, while its cost-effectiveness varies depending on utilization levels. By using Cast AI's platform and following a few simple steps, teams can test and deploy the most optimal LLM model for performance, cost, and security, making it an attractive alternative to running proprietary models.

In-place pod resizing in Kubernetes now allows users to update container resource requests and limits without restarting the pod, making it possible to adapt to real-time workload demands with minimal disruption. This feature is a game-changer for platform engineering, DevOps, and SRE teams, especially when paired with automation platforms like Cast AI that can fully leverage this capability to adjust resource allocations on the fly. With in-place resizing, teams don't need to resort to overprovisioning to avoid risk, leading to wasted resources and inflated cloud bills. Instead, they can now optimize their workloads continuously, even as they evolve, without manual intervention or complex restart automation logic. Cast AI takes full advantage of this feature by automatically detecting resource inefficiencies and applying the right adjustments at the right time, eliminating the need for manual monitoring and decision-making.

Containers and Kubernetes have transformed the way teams build and deploy applications, but this agility comes with increased security risks. Container image scanning is a critical security measure that ensures container images are clean and safe before deployment, minimizing the risk of security breaches. However, it has its limitations, including false positives, reactive approach, and complexity. Cast AI stands out with its advanced approach to Kubernetes and container security, transforming container image scanning from a reactive task to a proactive, automated security workflow. It proactively suggests secure, updated base images, automates remediation, and provides attack path visualization to prioritize and remediate vulnerabilities efficiently. Organizations can achieve robust, automated security measures with Cast AI, ensuring vulnerabilities are identified, prioritized, and remediated effectively.

This article explores the concept of Kubernetes ExternalName services, which expose a DNS name outside the cluster to consumers inside the cluster. This allows for abstracting the details of external services from applications, making it easier to manage and maintain complex deployments. The article highlights various use cases for ExternalName services, such as hybrid deployments, external dependencies, migration or restructuring, configuration centralization, and multi-cloud setups. It also provides a practical example of using an ExternalName service in a disaster recovery scenario, demonstrating how the service can be used to propagate changes quickly and transparently to clients. The article notes that ExternalName services have some limitations, such as not providing load balancing out of the box and relying on DNS resolution, which must be carefully managed to ensure performance and security. It also provides best practices for using ExternalName services, including understanding traffic patterns, making DNS resolution dependable and quick, making adjustments in advance, putting security first, managing traffic and load balancing efficiently, testing external services, monitoring and troubleshooting, and wrapping up with a summary of key takeaways.