Simplifying cloud native for all | for sponsorship queries contact kubesimplify@gmail.com
India
Based in India
- Tweets 5,156
- Following 3
- Followers 12,720
- Likes 9,008
The checkout failed. Inventory was fine. Where did it break?
In this simulated OpenObserve demo, the trace leads to payment.charge and its injected timeout.
Follow the request, then inspect the failing span.
#OpenTelemetry
3
1
3
415
kubesimplify retweeted
So after I wrote the blog on open observe and I read the benchmark post by OpenObserve benchmark blog got me thinking into actually give it a spin and seeing the compaction plus cost savings myself.
I did a detailed walkthrough of OpenObserve 1.0 explaining the features and some numbers too. I think we all look out for complete package and tbh I do think OpenObserve is kind of that complete package in the observability ecosystem.
My recommendation is to read their benchmark blog with elasticsearch and then see the video, try out(you can try in < 60 seconds on your mac) and then try on your systems.
Also if you like the video, you know the drill? Like share comment :)
I was happy to explore the observability space again after sometime, its not over, I have more tools to explore which I will be sharing soon only on Kubesimplify.
Link to blog and video is in the next post(again don't forget to share for good karma :D )
#openobserve #kubernetes #observability
4
4
17
1,924
Blog link openobserve.ai/blog/elastics…
1
2
97
A storage benchmark can be accurate and still be easy to misread.
OpenObserve’s published run reports 118 GB of compressed data, or 211 GB of actual disk use including indexing. Elasticsearch used 375 GB.
But the systems accepted different amounts of data: 100% of documents for OpenObserve and 38% for Elasticsearch in that mapping configuration.
Before comparing the headline numbers, check what each includes and how much data actually made it in.
Which benchmark detail do you check first?
#Observability #OpenObserve #CloudNative
3
2
4
501
Tokyo, here we come 🇯🇵
AGNTCon + MCPCon Japan 2026
📅 Sept 10–11
Building AI agents, MCP servers/clients or agentic AI infrastructure? This is one to join.
🎟 ¥44,000 JPY
Code: KUBESIMPLIFY
#AGNTCon #MCPCon #AgenticAI
1
2
4
607
The first v1.37 patch release is due Sep 15, alongside September patches for v1.34, v1.35, and v1.36.
Patch releases are the cheapest insurance in Kubernetes: mostly CVE and regression fixes, tiny blast radius.
If you upgraded to v1.37.0 on release day, v1.37.1 is your friend.
2
456
Kubesimplify is teaming up with ContainerDays & AI Context Singapore as a media partner this year.
October 27–28, Suntec Singapore.
Two days that are actually about the work , production Kubernetes, platform engineering, AI infrastructure, and where agentic tooling and MCP are heading next.
60+ sessions from people who are building this stuff, not just talking about it.
If you're in cloud native or AI infra, this is the one to be at in Singapore this year.
Check the Link below👇
1
2
8
432
etcd runs your entire cluster and most people treat it as a black box.
7 things worth knowing, one per tweet:
1
3
8
432
etcd quiz: what actually creates a new revision in etcd?
A) every read
B) every write
C) every watch event
D) compaction
Your cluster's entire state history hangs on this one.
Answer tomorrow at 1:30 PM IST.
2
373
Are you ready for part 2 ?
Session 1 of the KubeSimplify Book Club is done. We're reading Inference Engineering by Philip Kiely and this week covered the Preface, Chapter 0, and Chapter 1.
Chapter 0:
- open models (Hugging Face, Nvidia) are closing the gap with closed frontier models, a big reason inference engineering is hot in 2026
- local inference means running models on your own hardware
- every decision comes down to latency, availability, and cost
- the pipeline: training, learning, inferencing, serving
on the ground: teams renting GPU racks from providers like NeevCloud, enthusiasts buying DGX Spark and Nvidia GPUs
Chapter 1:
- model weights are layer 1, inference runtime is layer 2
- speculative decoding to speed up generation
- the sumo wrestler vs swimmer analogy for throughput vs latency tradeoff
- metrics to know: TTFT, TPS, P90
- tensor parallelism to scale across GPUs
Recording is up, link below. Next session Friday 8:30 PM IST with Saiyam.
1
6
501
First v1.37 patch release lands next week: v1.37.1 targets Sep 15, 2026 (cherry pick deadline Sep 11), alongside the September patches for 1.35/1.36.
A 1.34.12 may follow only if a CVE or critical fix warrants it, 1.34 entered maintenance mode on Aug 27 and goes EOL Oct 27.
Patch releases are the cheapest insurance in Kubernetes: mostly CVE and regression fixes, tiny blast radius.
If you upgraded to v1.37.0 on release day, .1 is your friend.
3
406
🚨 Big one for the community!
PyTorch Conference North America 2026
📅 Oct 20–21 | 📍 San Jose
✅ 7 technical tracks
✅ 3 co-located events
✅ Demos, posters & networking
Learn from Meta, UC Berkeley, Pinterest & more on Cudagraphs, TorchDynamo & multi-node training.
🎟️ 40% OFF with code: KUBESIMPLIFY
🔗 Registration link below 👇
#PyTorch #ML #AI
1
1
2
441
KubeCon + CloudNativeCon North America 2026 is coming.
November 9 to 12, Salt Lake City.
CNCF's flagship conference brings together adopters and technologists from across the open source and cloud native world, all working to advance cloud native computing.
What's inside:
- Expert-led talks
- Latest innovations from leading vendors
- 11 CNCF-hosted co-located events
- Parties and unmatched networking with TeamCloudNative
- Swags
Community discount: 20% off current pricing.
Code: KUBESIMPLIFY20
Registration Link pinned below👇
1
1
3
421
EKS access entries, in 20 seconds:
API-driven replacement for the aws-auth ConfigMap.
Grants live in the EKS API: auditable in CloudTrail, managed by IAM, free of hand-edited YAML in kube-system.
Migration is incremental. Both can coexist while you move.
1
2
425
The New Stack's writeup: thenewstack.io/kubernetes-fl…
2
120
81% of EKS clusters still route access through the deprecated aws-auth ConfigMap (2025 Kubernetes Security Report).
A hand-edited map from IAM identities to cluster permissions that anyone with edit rights can quietly change.
That is not an auth system. That is a shared text file.
2
10
890
Session 1 of the KubeSimplify Book Club is done. We're reading Inference Engineering by Philip Kiely and this week covered the Preface, Chapter 0, and Chapter 1.
Chapter 0:
- open models (Hugging Face, Nvidia) are closing the gap with closed frontier models, a big reason inference engineering is hot in 2026
- local inference means running models on your own hardware
- every decision comes down to latency, availability, and cost
- the pipeline: training, learning, inferencing, serving
on the ground: teams renting GPU racks from providers like NeevCloud, enthusiasts buying DGX Spark and Nvidia GPUs
Chapter 1:
- model weights are layer 1, inference runtime is layer 2
- speculative decoding to speed up generation
- the sumo wrestler vs swimmer analogy for throughput vs latency tradeoff
- metrics to know: TTFT, TPS, P90
- tensor parallelism to scale across GPUs
Recording is up, link below. Next session Friday 8:30 PM IST with Saiyam.
2
2
10
1,032
🦚 Happy Janmashtami from KubeSimplify!
Today, we celebrate Lord Krishna and the values he represents -- wisdom, compassion, courage and doing what is right. 🙏
May this Janmashtami bring peace, happiness, prosperity and positivity to you and your loved ones.💙
Enjoy the festivities, spend time with family & friends, and make the most of the day! ✨
And if you have some time tonight, join Saiyam for the first session of our AI Inference Engineering Book Club. 📖
🕣 8:30 PM IST
🔗 Register now, link below👇
Once again, Happy Janmashtami! 🦚💙
1
1
5
344