April 2022 Summaries

AWS Glue is a fully serverless, cloud-based data integration service that automates the ETL process for businesses and enterprises. It offers faster data integration, a serverless environment, and data automation for big data. The pros of using AWS Glue include automatic code generation, serverless design, debuggable logs, and suggested data schemas. However, potential downsides include limited script compatibility, restricted control over resources, compatibility issues with non-AWS systems, and a steep learning curve for new users. Overall, AWS Glue is an efficient solution for managing big data in most enterprises, especially those already using AWS services.

Cloud cost reporting is a must-have to track expenses and prevent unexpected bills like the $80,000 per day incident at Adobe's development team. However, relying solely on cost reports is insufficient for effective cost optimization. Automated cost optimization techniques are essential to make a real difference in reducing cloud spend. By generating detailed reports and sending alerts when certain parts of the infrastructure exceed set thresholds, teams can identify areas for improvement. Additionally, automated cost optimization tools can analyze cloud resources and provide actionable recommendations to reduce costs. A case study demonstrates how automated cost optimization on Amazon EKS reduced costs by 66% in just 15 minutes. By running automated cost optimization, teams can significantly reduce their cloud bills and improve their overall cloud spend management.

Amazon Web Services (AWS) experienced a second outage in two months on January 4th, affecting businesses such as Slack, Zoom, and Notion. On November 25th, AWS had another major outage impacting various services like Amazon API Gateway, Lambda, DynamoDB, CloudWatch, CloudTrail, CloudFormation, Autoscaling, Athena, among others. The repeated outages highlight the need for companies to adopt a multi-cloud disaster recovery and failover strategy to prevent downtime when cloud providers experience issues. CAST AI offers solutions that automatically shift resources between Azure or Google Cloud when AWS faces troubles, ensuring infrastructure is never down.

** The world is facing a growing need for cybersecurity professionals to tackle the rising threat of cyberattacks as the adoption of cloud computing continues to accelerate. The shortage of 2.7 million cybersecurity experts globally makes automation crucial in tackling this challenge. Nation-state actors and critical infrastructure are identified as three major threat vectors with potential for large-scale and global impact in 2022, including tensions between Russia and Ukraine and China's growing cyber capabilities. Security risks associated with moving government workloads to the cloud, containers, and multi-cloud environments also pose significant challenges. The "curse of abundance" caused by numerous cloud security services can lead to complexity and misconfiguration issues, while organizations must navigate multiple cloud vendors and implement CASB solutions to detect multi-cloud configuration drift. Governments are also adopting multiple clouds, making nation-state actors a potential threat in this new attack vector landscape.

The global chip shortage has been affecting major tech companies, including Samsung and Apple, which may be impacted by supply chain disruptions. On the other hand, Apple's new M2 chipset entered mass production in April and will start being used in MacBooks scheduled for release later this year. TSMC, the world's largest chip producer, announced a $100 billion investment program to stay ahead in the industry, which is becoming increasingly expensive with each new generation of chips. Other companies, such as Amazon and Nvidia, are also investing heavily in chip production and research, while IBM acquired AI-powered application resource management solution Turbonomic and Dell sold its cloud platform Boomi for $4 billion. Microsoft's cloud sales increased by 33% in Q3, surpassing 20% market share and closing in on Amazon's leadership in cloud computing.

The cloud space has seen several significant developments in June 2021, including a major outage at AWS, which highlighted the need for companies to invest in a multi-cloud strategy. Additionally, Deloitte acquired CloudQuest, Microsoft secured an agreement with AT&T to host its 5G network on Azure, and Broadcom forecasted impressive quarterly sales driven by growing demand for chips used in data centers. In developer news, GKE Analyzer is now available publicly, allowing users to optimize their cloud costs, and CAST AI has introduced new features such as support for organizations, paused and resumed clusters, and accessible node lists.

CAST AI is a platform that helps developers deploy, manage, and cost-optimize their Kubernetes clusters across multiple cloud platforms. It aims to end cloud provider lock-in and enable multi-cloud deployment. CAST AI provides managed Kubernetes services with a vanilla Kubernetes experience, ensuring maximum observability while maintaining excellent network performance. The platform comes with predetermined configuration choices, including auto-healing, logging, metrics, and health checks, making it production-ready. CAST AI also offers built-in observability features such as Prometheus, ElasticSearch, Filebeat, Kibana, Grafana, Kubernetes Dashboard, and Hubble UI for network and service connectivity visibility. The platform is free to try with up to 20 CPU's, and developers can join the community on Slack to share thoughts and ideas.

The incident involved the Google User ID Service, which detected outdated data and triggered an automated quota decrease due to its migration to a new quota system. This resulted in various Google services, including YouTube and Google Workspace, becoming inaccessible globally. The outage highlighted the risks of over-reliance on a single cloud provider and the importance of having a multi-cloud disaster recovery strategy to mitigate such failures. CAST AI's approach allows companies to run applications on multiple clouds, shifting resources to alternative providers when needed, thereby minimizing the impact of outages caused by a single cloud company.

Kubernetes has announced that it will deprecate Docker as a container runtime after version 1.20, but this change does not affect users who manage their own clusters or use managed services like CAST AI, which can continue to run Docker images without issue. The reason for this change is that Docker's API is incompatible with Kubernetes' Container Runtime Interface (CRI), and Kubernetes wants a unified API for different container runtimes. This means that users will need to switch to other compliant container runtimes like containerd or CRI-O, but their existing development flows that include building Docker images for running on Kubernetes do not need to change. However, relying on the underlying docker socket may break certain workflows, and there are alternative options available for this specific use case.