Matthieu SIMON

the need for faster data processing is only getting higher

46

The Architecture of Apache Druid #apachedruid #olap #history #architecture #storage #distributedsystems https://lnkd.in/egqai_ui

1

Microsoft's 𝐒𝐦𝐚𝐥𝐥 𝐋𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐌𝐨𝐝𝐞𝐥 (𝐒𝐋𝐌) 𝐏𝐡𝐢-3 𝐰𝐚𝐬 𝐣𝐮𝐬𝐭 𝐫𝐞𝐥𝐞𝐚𝐬𝐞𝐝! Interesting insights shared in this article by Product Managers at Microsoft. I'm incredibly excited about the potential benefits of Phi-3, particularly on edge devices such as mobile phones, tablets, or even devices used by farmers in the fields. Key Takeaways from reading the article: 𝐒𝐋𝐌𝐬 𝐚𝐫𝐞 𝐍𝐎𝐓 𝐦𝐞𝐚𝐧𝐭 𝐭𝐨 𝐫𝐞𝐩𝐥𝐚𝐜𝐞 𝐋𝐋𝐌𝐬. 𝐄𝐚𝐜𝐡 𝐡𝐚𝐯𝐞 𝐭𝐡𝐞𝐢𝐫 𝐨𝐰𝐧 𝐩𝐫𝐨𝐬 𝐚𝐧𝐝 𝐜𝐨𝐧𝐬. ✅ SLMs are designed to perform well for simpler tasks, are more accessible and easier to use for organizations with limited resources and they can be more easily fine-tuned to meet specific needs.  ✅LLMs are more suited for applications that need orchestration of complex tasks involving advanced reasoning, data analysis and understanding of context.  ✅𝐏𝐨𝐫𝐭𝐟𝐨𝐥𝐢𝐨 𝐨𝐟 𝐌𝐨𝐝𝐞𝐥𝐬: Some customers may only need small models, some will need big models, and many are going to want to combine both in a variety of ways, to optimize costs. ✅𝐈𝐧𝐝𝐮𝐬𝐭𝐫𝐢𝐞𝐬: SLMs also offer potential solutions for regulated industries and sectors that encounter situations where they need high quality results but want to keep data on their own premises. ✅𝐄𝐝𝐠𝐞 𝐂𝐨𝐦𝐩𝐮𝐭𝐢𝐧𝐠: SLMs will be suitable for edge computing. Consider applications like car computers, PCs without Wi-Fi, traffic systems, smart sensors on factory floors, remote cameras, or devices monitoring environmental compliance. ✅ 𝐐𝐮𝐚𝐥𝐢𝐭𝐲 𝐯𝐬 𝐋𝐚𝐭𝐞𝐧𝐜𝐲: Latency is the delay experienced when LLMs access the cloud to generate responses to user prompts. While high-quality answers can justify waiting, in many cases, speed is crucial for user satisfaction. SLMs will shine here in terms of speed. #slms #llms #generativeai #cloudcomputing

6

1 commentaire

Performance-wise, the new protocol has reduced dataplane latency by 11% on average, and by 17.4% at the P90 mark. In addition to making SQS itself more performant, this change benefits services that build on SQS as well. For example, messages sent through Amazon Simple Notification Service (Amazon SNS) now spend 10% less time “inside” before being delivered. Finally, due to the protocol change, the existing fleet of SQS hosts (a mix of X86 and Graviton-powered instances) can now handle 17.8% more requests than before. #AWS #AmazonWebServices #AWSBlogs #Cloud #CloudComputing #Serverless #SQS

2

Master Service Discovery in Microservices | Eureka and Java Spring Boot #systemdesign #microservices #servicediscovery #eureka #springboot #java https://lnkd.in/eUnAzXf3

9

One of my beefs with terraform is it’s trying so hard to be flat. Some of us want very dynamic scripts. But when you create a k8s cluster, form the kubexonfig file, then need to use the kubexonfig file to load in the rest of the parts like ingress, having a depends_on would be nice for providers. I did cheat a little and abuse the variable system to force the provider to wait, but then you run into what about a workspace that will have kubernetes set to false and get a zero count for your k8s module? Then it gets angry because the module output isn’t there. I still like Chef. Chef let me do what I want when I want. Just was bad on creating the underlying infrastructure but great at configuring it once it was up. *shrug* #terraform #kubernetes #k8s #eks #automation #burgerkingmodel

3

1 commentaire

The flexibility Kubernetes provides for the RPCs Exec, Attach and PortForward is truly outstanding compared to other methods. Nevertheless, container runtimes have to keep up with the latest and greatest implementations to support those features in a meaningful way. The general effort to support WebSockets is not only a plain Kubernetes thing, it also has to be supported by container runtimes as well as clients like crictl. ✍️✍️✍️ Container Runtime Interface streaming explained https://lnkd.in/gfgnqDmd

3

Great podcast about Kubernetes Policy as code ! Who knows more about Kyverno than Jim Bugwadia 😜? “Jim Bugwadia, CEO of Nimrata and a committer to the kyverno projects, joins host Robert Blumen for a discussion of policy-as-code and the open source kyverno project. The discussion covers the nature of policies; policies and security; policies and compliance to standards; security scans”

8

Fantastic data points by Ajit Inamdar on Amazon ECS !! Using Amazon ECS from tiny workloads to large production grade deployments is a no-brainer, and you let AWS handle all the heavy lifting of container orchestration and yes, you can also leverage your existing AWS skills to get started. #aws #containers #amazonecs

17

2 commentaires

🔒 How do you manage secrets in your Kubernetes environment? 🤔 Kubernetes is challenging when it comes to secret management, that’s a fact. How do you ensure your secrets are secure? Do you use a centralized or decentralized approach? Where do you keep your secrets? One effective tool is Sealed Secrets by Bitnami. This tool offers a secure way to encrypt secrets before they reach your cluster, ensuring that sensitive data is protected at all times and can be safely stored in a Git repository. #devops #kubernetes

6

🚨 I just published an in-depth article on Medium covering - 𝐑𝐞𝐚𝐥-𝐓𝐢𝐦𝐞 𝐔𝐬𝐞-𝐜𝐚𝐬𝐞 𝐨𝐟 𝐅𝐫𝐚𝐮𝐝 𝐃𝐞𝐭𝐞𝐜𝐭𝐢𝐨𝐧 𝐢𝐧 𝐅𝐢𝐧𝐚𝐧𝐜𝐢𝐚𝐥 𝐓𝐫𝐚𝐧𝐬𝐚𝐜𝐭𝐢𝐨𝐧𝐬 𝐰𝐢𝐭𝐡 𝐊𝐚𝐟𝐤𝐚 𝐚𝐧𝐝 𝐒𝐩𝐚𝐫𝐤 𝐒𝐭𝐫𝐞𝐚𝐦𝐢𝐧𝐠 🚨 Fraud is a big challenge for financial institutions, with countless transactions happening every second. In this article, I show how to set up a real-time fraud--detection pipeline using 𝐀𝐩𝐚𝐜𝐡𝐞 𝐊𝐚𝐟𝐤𝐚 for streaming data and 𝐒𝐩𝐚𝐫𝐤 𝐒𝐭𝐫𝐞𝐚𝐦𝐢𝐧𝐠 to analyze and flag suspicious transactions in real time. Here’s what I cover: ✅ 𝐒𝐞𝐭𝐭𝐢𝐧𝐠 𝐮𝐩 𝐊𝐚𝐟𝐤𝐚 𝐭𝐨 𝐬𝐭𝐫𝐞𝐚𝐦 𝐭𝐫𝐚𝐧𝐬𝐚𝐜𝐭𝐢𝐨𝐧 𝐝𝐚𝐭𝐚 ✅ 𝐁𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐚 𝐒𝐩𝐚𝐫𝐤 𝐒𝐭𝐫𝐞𝐚𝐦𝐢𝐧𝐠 𝐣𝐨𝐛 𝐭𝐨 𝐝𝐞𝐭𝐞𝐜𝐭 𝐟𝐫𝐚𝐮𝐝 𝐩𝐚𝐭𝐭𝐞𝐫𝐧𝐬 ✅ 𝐑𝐞𝐚𝐥-𝐰𝐨𝐫𝐥𝐝 𝐟𝐫𝐚𝐮𝐝 𝐝𝐞𝐭𝐞𝐜𝐭𝐢𝐨𝐧 𝐫𝐮𝐥𝐞𝐬 - 𝐥𝐚𝐫𝐠𝐞 𝐰𝐢𝐭𝐡𝐝𝐫𝐚𝐰𝐚𝐥𝐬, 𝐫𝐚𝐩𝐢𝐝 𝐭𝐫𝐚𝐧𝐬𝐚𝐜𝐭𝐢𝐨𝐧𝐬, 𝐡𝐢𝐠𝐡-𝐫𝐢𝐬𝐤 𝐥𝐨𝐜𝐚𝐭𝐢𝐨𝐧𝐬 ✅ 𝐎𝐮𝐭𝐩𝐮𝐭 𝐚𝐧𝐝 𝐚𝐥𝐞𝐫𝐭𝐢𝐧𝐠 𝐬𝐞𝐭𝐮𝐩 ✅ 𝐒𝐜𝐚𝐥𝐢𝐧𝐠 𝐮𝐩 𝐟𝐨𝐫 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧 This guide is ideal for those looking to build scalable, low-latency systems that can handle millions of transactions with ease... 🚀 🔗 Read the full article Below Whether you're just getting started or looking to take your real--time analytics skills further, there’s something in it for you! Happy coding!! If you found this valuable, don’t forget to like and subscribe for more insightss! #Kafka #SparkStreaming #FraudDetection #RealTimeAnalytics #BigData #FinancialTech

20

2 commentaires

Really interesting read on the AI assited coding platforms. GitClear research talks about impact of code quality. Here is my summary on the research #copilot #codequality #softwareengineering https://lnkd.in/e23uUTiR

11

1 commentaire

New! In this video, we will show you 2 methods how to migrate from Google Cloud VPC Firewall Rules to Firewall Network Policy using the VPC FW Rule Migration Tool. https://lnkd.in/eYBVtkRt Googler specialist Nicolas #vpcfirewallrules #googlecloud #cloudfirewall #migration #terraform #cloud #google

1

Performance-wise, the new protocol has reduced dataplane latency by 11% on average, and by 17.4% at the P90 mark. In addition to making SQS itself more performant, this change benefits services that build on SQS as well. For example, messages sent through Amazon Simple Notification Service (Amazon SNS) now spend 10% less time “inside” before being delivered. Finally, due to the protocol change, the existing fleet of SQS hosts (a mix of X86 and Graviton-powered instances) can now handle 17.8% more requests than before. #AWS #AmazonWebServices #AWSBlogs #Cloud #CloudComputing #Serverless #SQS

3

Exhaustive whitepaper on the value of a decoupled CI/CD architecture. Worth the read. A decoupled CI/CD architecture is easy to scale and makes the CI/CD pipelines more resilient to failures, which is critical as the end-to-end software production and delivery pipelines grow more and more complex, not least in a microservices architecture with thousands of independent pipelines. https://lnkd.in/e594C7iX

5

Toying around with KrakenD CE API Gateway + JHipster + Entra ID. I had built an end-to-end POC with Kong + JHipster + Azure AD (when Entra ID was still "AAD") and I want to look at this again. KrakenD is licensed Apache 2.0 and they are a member of the Linux Foundation. Add Microsoft's MSAL libraries have really matured in the two years since that POC (that relied Spring Security's OIDC/OAuth2 support and Microsoft's default support of the same). I may have a repository pushed soon enough for you to look at. But first I'm going to look at that GitLab CI/CD blueprint in JHipster. Maybe rewrite it for GitLab's CI/CD catalog support of To Be Continuous (TBC). If there's one thing I love more than GitLab, it is TBC 🫶. #JHipster #KrakenD #OSS #OIDC #tobecontinuous

14

1 commentaire