Bastien Leblanc

When my colleagues Akshat Vig and Ashish Kumar joined MongoDB, they brought their extensive experience from Google and Amazon Web Services (AWS) to a rich tradition of MongoDB development. An aggregation, if you will... 🥁 In this new MongoDB Engineering Blog post, they share their vision of the key themes that shape how our Engineering team works to help developers get to production fast: Resilience - providing a system that anticipates failure and recovers automatically Intelligence - Vector search and AI capabilities built directly on your operational data Simplicity - Multi-cloud that works with a single API, turning operational headaches into simple deployment choices. The goal isn't just a better database but enabling true developer agility. Check out the full blog post: https://lnkd.in/gbA_kX-d

26

Amazon MSK now supports Apache Kafka version 3.9, bringing new flexibility for managing tiered storage and enabling consumer applications to continue reading historical data even after tiered storage is disabled at the topic level. This release also includes a range of bug fixes and improvements, making it easier than ever to run and scale real-time streaming workloads on a fully managed service. With compatibility across all AWS regions where MSK is available, teams can confidently migrate or upgrade their Kafka clusters and focus on building innovative streaming applications, while AWS handles the operational heavy lifting. #ApacheKafka #AmazonMSK https://lnkd.in/gSMJFsRh

21

.@Oracle Database@AWS Now Generally Available | https://lnkd.in/e7YaauZj >> Congrats. My take: "The general availability of Oracle Database@AWS completes the hyperscalers’ rollout of Oracle’s multicloud strategy. AWS customers can now access their enterprise data on their Oracle databases using their favorite Amazon analytics and AI services. With Exadata and RAC running on OCI in AWS data centers, Oracle databases can achieve levels of performance and availability not previous experienced by customers in AWS. All in all, the multicloud era proclaimed by Larry Ellison is well on the way." #NextGenApps #NextGenDB

24

2 Comments

Scaling with Confidence: Understanding HPA Tolerance Levels As enterprises modernize workloads on Kubernetes, autoscaling is no longer optional — it is a foundational design choice that impacts reliability, performance, and cost efficiency. The Horizontal Pod Autoscaler (HPA) is often the first scaling mechanism architects implement. Yet, the nuance many overlook is tolerance levels — the thresholds that define when HPA should react to metric changes. At first glance, tolerance might feel like a minor configuration detail. But from a solution architect’s perspective, it directly influences: User Experience → avoiding performance degradation during sudden traffic spikes. Operational Efficiency → preventing “flapping” of pods due to overly sensitive thresholds. Cost Management → balancing scale-up responsiveness with scale-down prudence. Why Tolerance Levels Matter HPA works by monitoring resource metrics (CPU, memory, or custom metrics) and adjusting pod counts. However, metrics naturally fluctuate. Without tolerance, HPA would continuously scale pods up and down, creating instability. Low tolerance (too aggressive): pods scale frequently, driving operational noise and higher costs. High tolerance (too relaxed): workloads risk performance drops before scaling reacts. The default tolerance (10%) is often a starting point, but real-world architectures demand tuning based on workload patterns. Architect’s Lens on HPA Tuning As a solution architect, I frame tolerance discussions around three dimensions: 1. Workload Characteristics Latency-sensitive applications (e.g., payments, trading) need faster scale-up → lower tolerance. Batch or background jobs can absorb fluctuation → higher tolerance for efficiency. 2. Business Priorities Enterprises prioritizing customer experience over cost may tune tolerance tighter. Those focusing on FinOps discipline may accept brief latency for smoother, cost-optimized scaling. 3. Operational Resilience Tolerance tuning must integrate with observability. Architects should pair HPA with dashboards, alerts, and predictive autoscaling to preemptively react to patterns. Leadership Insight Tolerance levels aren’t just technical knobs; they represent architectural trade-offs between agility, cost, and resilience. As solution architects, our role is to ensure scaling strategies align with the organization’s operating model — whether that means absorbing costs to protect customer experience or enforcing stricter efficiency controls. Done right, HPA tuning becomes a business enabler: smoothing seasonal demand, supporting product launches, and ensuring operational confidence. 🔹 Takeaway: HPA tolerance is not a “set and forget” parameter. It is an architectural decision point that shapes how your digital platforms handle unpredictability. Solution architects who master this balance demonstrate not only technical depth but also business leadership in scaling strategies.

10

2 Comments

It all starts with data, clean data and in real time. Sharing Confluent’s latest ebook on how real-time data is driving real AI results. It covers why legacy systems are holding AI back and how a data streaming platform can unlock speed, scale, and smarter decisions. Download it here: https://lnkd.in/gxGxu3Ff

23

Wouldn’t it be nice to code and manage resources all in one place in VS Code? Microsoft's Catherine Wang explains how to set up VS Code to connect with Redis MCP Server and Azure Managed Redis. #developer #AMR #MSPartner #Redis https://lnkd.in/gM4ajQme

11

Most Oracle → PostgreSQL programs start with a license conversation. Few end with an architectural transformation. Migration reduces cost. Modernization increases leverage. I’ve seen enterprises move off Oracle successfully — and still preserve the same structural fragility, just on a different platform. That’s not modernization. That’s relocation. In my latest article, I introduce a dual-layer framework: 🔹 MIGRATE+ — engineering rigor, governance, validation, operational discipline 🔹 MODERNIZE — modular architecture, AI-readiness, autonomy, and platform leverage Because PostgreSQL is not just a database replacement. It is an opportunity to redesign your digital core. The real question is not: “Can we migrate?” It is: “Are we willing to modernize?” If you’re leading a migration initiative at scale — especially in regulated environments — this perspective may help you think beyond the cutover window. #PostgreSQL #Modernization #DatabaseMigration #EnterpriseArchitecture #CIO #CTO #DigitalTransformation #OpenSource #DataStrategy

16

2 Comments

Great news for Spanner and BigQuery users. Datastream now provides a seamless, serverless bridge from Spanner’s transactional power to the high-performance analytical capabilities of BigQuery and BigLake. Get the most scalable and available database that runs Google with BigQuery and BigLake’s serverless AI-Native Lakehouse!

514

3 Comments

AWS launched Amazon EBS Volume Clones, a new capability that allows users to create instant point-in-time copies of EBS volumes within the same Availability Zone with a single API call, eliminating the previous multi-step process of taking snapshots and creating volumes from them. https://lnkd.in/gWt5r3GR

14

1 Comment

I've been very impressed by the automation and AI integration that nodemation (n8n, or "Nathan" as some like to call it) has developed. For software that can be self-hosted as open source, it is incredibly powerful. I wrote up a post going into great detail about how to use infrastructure as code, Kubernetes, and automated builds to deploy n8n into your own AWS EKS environment. The post includes a full script to automate this process, including using a load balancer with SSL and a custom domain. Enjoy!

36

2 Comments

A #database architected and built for #Kubernetes #CockroachDB: A #DistributedSQL database built for Kubernetes CockroachDB is the only database architected and built from the ground up to deliver on the core distributed principles of #atomicity, #scale and #survival so you can manage your database IN Kubernetes, not along the side of it. https://lnkd.in/gE7Hp4KW

9

Your hosted/managed Kafka is not as hosted/managed as you thought and a lot of costs add up quickly as well as frustration? :-) Read At first glance, cloud-hosted Kafka looks easy. But if you’re still rebalancing partitions, scaling infrastructure, patching versions, and troubleshooting failures, you’re not getting a fully managed experience. This blog breaks down the 3 big challenges of hosted Kafka and how Confluent Cloud does it differently.

12

🚀 Three new Kafka topic APIs now available on Amazon MSK 🚀 Customers can now use the ListTopic, DescribeTopic, and DescribeTopicPartition to get details of their Kafka topics. You can also now view your topics through the AWS Console. Available for standard and Express clusters using Kafka version 3.6 and above! https://lnkd.in/g692an2b #awscloud #amazonmsk #apachekafka

105

2 Comments

💥 My first official #MicrosoftFabric blog on #AIFunctions You may be aware of .ai.extract() which is great for extracting independent values. But what if you wanted to extract complex nested objects for unstructured data ? Use ExtractLabel. Extremely powerful, super handy, verifiable, modular and ease to use. In the blog, I show a detailed example on how to use it. You can also try it in a sample notebook. Join me at FabCon next week to learn how to work with unstructured and multimodal data. 🔗https://lnkd.in/gHtzDDNt 📝https://lnkd.in/gPrbtjCr

121

3 Comments

Dear #AWS #EBS friends, this is a game-changer announcement for storage! EBS gp3 volumes can now scale up to 64 TiB in size (4X the previous 16 TiB limit), up to 80,000 IOPS (5X the previous 16,000 IOPS limit), and up to 2,000 MiB/s throughput (2X the previous 1,000 MiB/s limit). https://lnkd.in/epgshUw2

14