Chris Jones

🔥🔥🔥 #Grafana - Monitor #GenAI App/ #LLM with AI-Observability in Grafana Cloud 🔥🔥🔥 Topics: - Performance monitoring:  - Cost optimization (LLM) applicatons - End-to-end tracing: - Prompt and response tracking

10

1 Comment

Learn about the evolution of the Elasticsearch .NET client and how it will gradually phase out the old NEST library: https://gag.gl/tCCrQX #ElasticSearchLabs

2

Earlier this week, I shared a LINBIT blog post on the subject of Kubernetes. On the same theme, I noticed a YouTube video by The Linux Foundation: ‘Kustomize Your Operator’ by LINBIT’s Moritz Wanzenböck. The presentation is from the recent Open Source Summit. As the video description says, ‘More and more Operators automate deployment of cloud native applications in Kubernetes,’ and if you’re deploying in Kubernetes, watching this video will definitely aid your understanding! https://hubs.ly/Q02WLYXY0

4

Incredibly proud of the latest Kinesis Client Library version (3.0) which, when compared to previous KCL versions, reduces compute costs by up to 33%. KCL 3.0 introduces an enhanced load balancing algorithm that continuously monitors resource utilization of the stream processing workers and automatically redistributes the load from over-utilized workers to other underutilized workers. This ensures even CPU utilization across workers and removes the need to over-provision the stream processing compute workers which reduces cost. Additionally, KCL 3.0 is built with the AWS SDK for Java 2.x for improved performance and security features, fully removing the dependency on the AWS SDK for Java 1.x. #Kinesis #DataStreaming #KCL

94

Each phase of an #AI pipeline has different storage requirements. Learn about the data size, access requirements & recommended type of storage for each AI phase in part 1 of our guide to storage for AI: https://eqix.it/45Erly5  

1

In this Datadog video we’ll continue looking at how Kubernetes handles authentication with a look at bootstrap and static token authentication. https://lnkd.in/e4Gz_VKV

4

"Turning off observability" to reduce metrics (and ultimately, costs) is a problem that's more common than you'd realize, and one that Grafana Labs is focused on solving with Grafana Cloud. This case study is a good example. SailPoint reduced their metrics volume by 33% with the Adaptive Metrics feature which measures your metrics' actual usage and recommends which ones to aggregate. They now have 50 million total active series managed within Grafana Cloud. And the best part is that it's been quite "low-risk" for them. "We haven’t had any complaints about dashboards not working, or a query. Everything runs like it did before. Adaptive Metrics [in Grafana Cloud] has been a great way to show engineering that we can make changes to metrics, and they don’t have to worry about things breaking. And I don’t have to worry about causing outages or downtime. It is a really low-risk solution.” Read the full case study here and reach out if you'd like to learn more about Grafana Cloud's Cost Management capabilities like Adaptive Metrics: https://lnkd.in/gFc4bj5Z

71

2 Comments

Container orchestration platforms, like Kubernetes, were, from the beginning, an ideal solution for microservice architectures running a lot of stateless services. This was also the case for Datadog, which is run on dozens of self-managed Kubernetes clusters in a multi-cloud environment, adding up to hundreds of thousands of pods. But what about stateful applications? What are the best practices to run and scale those without losing data? The team at Datadog owning our Kafka clusters has been running big business critical storage workloads on our Kubernetes clusters for a long time. Over the years, they have gained experience on how to run this type of workload at scale, and have created tooling around it. The team at Datadog owning our Postgres databases, on the other hand, is currently working on the transition to move their workloads from managed cloud instances to Kubernetes. In this live session from the Datadog London Summit, Ara Pulido, Staff Developer Advocate, chatted with Martin Dickson, Senior Software Engineer in the Datadog Kafka team, and Edward Dale, Engineering Manager in the Postgres team at Datadog, about their experience, their tooling, and their stories (good and bad) on running stateful workloads on Kubernetes. https://lnkd.in/eFQyKhKM

6

🔥 🔥 🔥 Faster #RCA and #anomaly detection in #Grafana application observability based on #OpenTelemetry 🔥 🔥 🔥 With the result of: - Faster #RCA (Root cause analysis) - Reduce #MTTD / #MTTR - Preventing negative effects on the business and more #apm #opentelemetry

4

Market Context: Kubernetes, Hyperscalers, and Red Hat OpenShift My take on the news announcing the integration of Red Hat OpenShift on Oracle Cloud Infrastructure (OCI), merging a leading Kubernetes platform with a top distributed cloud. This collaboration supports configurations of Red Hat OpenShift on OCI, providing customers with flexible, cost-effective, and secure deployment options across multiple cloud environments. The partnership enhances enterprise capabilities for managing scalable, distributed applications, and supports AI and ML workloads, signifying a substantial advancement in cloud computing and container orchestration technologies. Read the full article here 👇 https://lnkd.in/eheeyHmG

8

3 Comments

The Kubernetes ecosystem may get too complicated for its own good. Picking stack components has become a top Kubernetes pain point! You can learn more from this new survey by Spectro Cloud. #openshift #aro #rosa #osd #rhoic https://lnkd.in/g8T88f4a

18

Please check out this talk at KubeCon with Mitch Becker and Tom McDonald. When deploying Container based applications, storage is often thought of last, yet can have a big impact on the performance, scale, and recovery of your stateful applications. This talk helps align applications to storage classes to shorten your deployment times. Check it out!!

15

#AzureAI + Weights & Biases = 🤝 Read the full guide: https://lnkd.in/gRYxXQ-3 Thanks for the close collaboration, Felipe Benavides, Takuto Higuchi

8

This was great InfoQ reporting on Karpenter 1.0 from Matt Saunders: "Karpenter automatically provisions new nodes as needed, responding to unschedulable pods in real-time. By dynamically adjusting cluster compute capacity, it optimizes both performance and cost-effectiveness." https://lnkd.in/ezJYJUJN

3

II'd like to share a blog post which is written by one of the best Solutions Architects at #LINBIT, Matthew Kereczman. Matt explores a stretched cluster site architecture for high availability and disaster recovery (DR) in Proxmox. Offsite replication for DR isn't something that Proxmox can do on its own, so this design could be welcome news for #Proxmox users.The blog looks into the necessary steps and settings for creating a stretched #LINSTOR cluster to provide disaster recovery (#DR) capabilities for Proxmox VE #virtual machines (VMs).

6

2 Comments

In this lightning talk at AWS re:Invent 2024, explore how to enhance Apache Spark using Amazon EC2 G6 GPU instances, demonstrating speedups of up to five times and cost reductions of 80% on Spark operations. bit.ly/41dv3yH

5