Kasey Shah

People often ask me about the differences in architecture between Amazon Dynamo (the 2007 SOSP paper), DynamoDB (the AWS serverless NoSQL database), and Aurora DSQL (the AWS serverless SQL databases). I captured the key points on durability, high availability, programming model, and trade-offs in my blog: https://lnkd.in/gkMG_tVD TL;DR: Despite the shared name, Dynamo and DynamoDB have little in common architecturally, and make very different trade-off decisions.

632

8 Comments

Standing up an AI workload shouldn’t take months, but a steep learning curve and a rapidly evolving stack often stretch production LLM deployments far beyond plan. We recently launched OCI AI Accelerator packs to address the most common blockers we see in enterprise AI. These packs reduce the time required to deploy Generative AI applications, while helping enterprises reduce the learning curve and stack complexity that typically slows production and deployment. Learn more: https://lnkd.in/emiTNsRF

189

3 Comments

We recently announced the GA of Oracle Database@AWS. Customers can now run Oracle Exadata Database Service and Oracle Autonomous Database on dedicated infrastructure on OCI within the AWS U.S. East (N. Virginia) and U.S. West (Oregon) Regions. In addition, to meet growing customer demand, we plan to make the service available in an additional 20 AWS Regions around the world. InfoWorld has the details: https://lnkd.in/gswnvM9z

108

8 Comments

📣 kgateway now supports Kubernetes Gateway API Inference Extension! 📣 AI workloads backed by GPUs running on Kubernetes are NOT like typical webapps. You need to solve the GPU multiplexing problem, but you also need to solve the request load-balancing problem. AI inference requests have unique characteristics that make conventional load-balancing techniques less effective. kgateway helps with this problem. Make smarter decisions based on real-time metrics such as GPU cache utilization, LLM queue depths, model names/fine-tuned adapters loaded, etc, etc. 🔎 Read more in my latest blog: https://lnkd.in/gTsna8wj 🔎 Checkout the Inference Extension OSS community: https://lnkd.in/g6gqyBju 🔎 Get started with kgateway: https://lnkd.in/gqCVE29a #ai #inference #kubernetes #networking #gatewayapi #apigateway #kgateway #cncf #llm

43

3 Comments

I'm excited to share that our partnership with Microsoft will now bring you full support for Azure ML services on VMware Cloud Foundation. While our customers and partners continue to embrace #PrivateAI, they have also asked for broadening support for #HybridAI alternatives that give them choice of best-of-breed in the public cloud and on-premises or edge. We've been listening, and that's what makes today's news especially exciting. Check out the blog post below for more information and a link to our solution architecture authored by Fabiano Teixeira. https://lnkd.in/gEKzpiMA

614

8 Comments

The VMware Exodus: Why IT Leaders Are Taking a Fresh Look at Hyper-V Broadcom's acquisition of VMware has fundamentally reshaped the virtualization landscape. With 150-1000% price increases, 72-core minimums, and subscription-only licensing, IT organizations face an unprecedented decision point. But here's what most people missed: While the industry focused on VMware's chaos, **Windows Server 2025 quietly became enterprise-ready** with features that directly address long-standing Hyper-V limitations. **What Changed in Windows Server 2025:** ✅ 2,048 vCPUs per VM (enterprise database-ready) ✅ GPU partitioning with live migration (AI/ML workloads) ✅ Workgroup clusters without AD (edge deployments) ✅ Enhanced security enabled by default ✅ Hardware flexibility—no vendor certification lists required The Strategic Advantage: Perpetual licensing still exists. Your existing hardware investments remain valuable. You control when and how you adopt cloud services—hybrid on your terms, not vendor mandates. This isn't about abandoning VMware because it's expensive. It's about making an informed decision with complete information. I'm launching an 18-part series that covers everything from TCO analysis to production architecture. Part 1 sets the stage and explains why this moment matters. For infrastructure architects and IT leaders navigating renewal decisions: the virtualization landscape has fundamentally changed. Let's evaluate all options honestly. Read Part 1: https://lnkd.in/erkzD4Px #Virtualization #HyperV #WindowsServer2025 #VMware #AzureLocal #AzureArc #TierPoint

120

10 Comments

The AWS Asia Pacific (Taipei) Region is now open. It has 3 Availability Zones (AZs), and brings the #AWS global footprint to 117 AZs across 37 geographic Regions worldwide. Read Betty Zheng's new blog post at https://lnkd.in/gev94yyR to learn more.

514

5 Comments

I'm joined by Prashant Desai, VP of Product at NetApp, and Aung Oo, VP of Azure Storage to discuss the evolution of ANF (Azure NetApp Files), its role in hybrid cloud environments, and how it's enabling seamless data mobility and innovation in AI. From the origins of the NetApp-Microsoft collaboration 20 years ago to the cutting-edge features like SnapMirror, FlexCache, and the new codename “Atom platform”, this episode is packed with insights for developers, architects, and IT leaders looking to modernize their storage strategy for AI. #AzureNetAppFiles, #NetApp, #AzureAI, #HybridCloud, #GenerativeAI, #AIAgents, #DataMigration, #CloudStorage, #SnapMirror, #FlexCache, #EnterpriseAI, #SoftwareDevelopment, #AzureOpenAI, #itpro

61

2 Comments

We're excited to announce new capabilities in Google Cloud #VertexAI Training designed to simplify and accelerate large-scale model development. Key highlights from our latest update: 🔹 Flexible, Self-Healing Infrastructure: Leverage fully managed, resilient Slurm environments with automated failure detection and performance-optimized checkpointing. 🔹 Cost-Effective Scheduling: Utilize Dynamic Workload Scheduler (DWS) for fixed future reservations or flexible on-demand capacity. 🔹 Integrated Frameworks: Access optimized recipes for the full model lifecycle (including SFT and DPO) and seamless integration with NVIDIA NeMo. Whether you're fine-tuning standard models or training massive custom foundational models from scratch, these new features are built to get you to production faster. Read the full announcement to see how organizations like Salesforce and AI Singapore are already leveraging these tools to elevate their models. 👇 https://lnkd.in/gVVq3WCw

185

2 Comments

🚀 From Weeks to Days: AI-Powered Cloud Modernization Excited to share my latest blog on the AWS Migration & Modernization site about transforming cloud modernization using AI agents (Kiro/Cline) with AWS Model Context Protocol servers. The shift: Traditional modernization takes weeks of manual work. Agentic architecture automates assessment, planning, and implementation—delivering production-ready Infrastructure as Code in days, not weeks. Key benefits: ✅ Automated system analysis with visual diagrams ✅ Real-time cost modeling during design ✅ Auto-generated CloudFormation, CDK, or Terraform code ✅ Built-in AWS Well-Architected Framework compliance This approach fundamentally changes how we tackle cloud transformation by combining intelligent automation with AWS best practices. Read the full article: https://lnkd.in/ev89jAhk #AWS #CloudModernization #AI #DevOps #InfrastructureAsCode What modernization challenges is your organization facing?

27

2 Comments

Last week we announced the general availability of Nutanix Database Service (NDB) v2.8. NDB is the market leader in heterogenous database lifecycle management across hybrid clouds for Oracle, MS SQL Server, PostgreSQL, MongoDB and MySQL. It enables enterprises to build Database as a service (DBaaS) offerings for their organizations with ease. Leading organizations world wide including WellsFargo, Sogei, AAA, JetBlue Airlines, Rafael, Penn National Insurance, & Sub Zero use NDB for heterogenous database lifecycle management. With this release we are further expanding the platform to support more powerful database operations that include bulk patching, cluster capacity & recovery and enhanced support for database management on AWS with Nutanix Cloud Clusters (NC2). In addition this release delivers increased database compatibility and control for recent PostgreSQL, EDB and MongoDB releases. This release is packed with tons of new capabilities and features and you can learn more about them via the following blog: https://lnkd.in/guB9E4Cr You can also learn more about NDB and what's new in v2.8 by signing up for the upcoming live TechTalk on April 17th at 8am PST. https://lnkd.in/gYJcZDxh. You'll be able to engage in live Q&A with the creators of NDB.

45

2 Comments