Thanos Baskous

I spoke with Tom Wilkie, CTO of Grafana Labs about what it’s like scaling open source projects while building a commercial company around them. Discussed the challenges, lessons, and what goes on behind the scenes of a tool so many of us use: https://lnkd.in/gD92CXuD

587

9 Comments

Terraform plans are noisy. We made them visual. When a terraform plan has dozens (or hundreds) of changes, it's hard to quickly understand what's going on. We fixed that in the Terrateam UI. Inspired by feedback from Derek Morgan, Terrateam now breaks down plan outputs by resource, showing what will be created, updated, or destroyed, with visual diffs for each field. #terraform

51

2 Comments

Accidentally discovered a new way to work with GitHub Copilot today: write new tests based off test coverage reports. Copilot (or I) run this command to generate a report of all the uncovered lines in the current diff: pytest --cov --cov-report=xml && \ diff-cover coverage.xml --html-report coverage_report.html && \ open coverage_report.html Then Copilot reviews the report, summarizes the uncovered situations, and proposes new test cases to add, or if the code is truly unreachable, suggests deleting it. Just another reason why you should add diff-cover (or similar tool) to your projects!

115

4 Comments

The migration that nearly locked out 3,000 developers. At Palantir, we were moving Elasticsearch health data for thousands of microservices. Tests looked clean. Auto-promotion to prod kicked off at 4am. Then — chaos. Migrations looped endlessly. Permissions only applied after “completion.” Developers suddenly couldn’t see their installs. 300 environments × 500 microservices each = ~3,000 devs blocked. The culprit? One extra parameter in the "finish" call. In test, global permissions masked the issue. In prod, it broke everything. We rolled back fast. Workflows recovered. But I walked away with two lessons: - Test is not prod. Your fixtures will never match real-world messiness. - Scale amplifies small mistakes. One parameter can ripple into thousands of outages. What you need in production isn’t paranoia. It’s humility.

41

1 Comment

Observability has always been about understanding complex systems and turning signals into action and resolution. Today’s systems are more complex (and the signals more abundant) than ever, making AI-powered observability a force multiplier. Today, we’re excited to launch Grafana Assistant in public preview. Built into Grafana Cloud, it’s a context-aware, AI partner that helps you build dashboards, investigate issues, explore metrics, logs, and traces, and accelerate root cause analysis faster than ever. When it comes to AI in observability, we believe in designing for outcomes, not job descriptions, keeping humans in the loop, and using the power of open source to accelerate meaningful innovation. We’ve built Grafana Assistant with all this in mind. Mat Ryer and Maurice Rochau’s blog post highlights how we at Grafana Labs think about AI-powered observability. You can read it here: https://lnkd.in/erBBTQFW

235

7 Comments

OLTP databases and analytical systems both try to minimize I/O, but they do it through entirely different abstractions. B-trees and secondary indexes serve transactional workloads where selectivity and random access dominate. Open table formats like Iceberg and Delta rely on partitioning, sorting, and metadata-driven pruning to achieve the same goal at analytical scale. In this post I explore what “index” really means in this world, and why we can’t just bolt secondary indexes onto analytical systems. https://lnkd.in/d6wrgD9t

209

18 Comments

I've seen a simple infra change take weeks before. Not because it was hard, but because the ticket bounced between Dev, SRE, and Sec multiple times. That’s orchestration debt: the hidden cost of work getting stuck in silos, with no shared flow to move things forward. - Flow without orchestration: Every change bounces Dev → SRE → Sec → Ops through tickets and handoffs. - What it creates: Delays pile up, scripts get duplicated, and nobody’s sure who owns what. - Flow with orchestration: One intent declared upfront, carried through automatically so all teams stay aligned without the back-and-forth. At one enterprise we worked with, eliminating orchestration debt cut DevOps tickets by 95% and sped up developer productivity by 20%. We've seen how this debt compounds: slowing releases, frustrating engineers, and inflating costs quietly in the background. If you want to see how this works in practice, read our blog post Unifying Your Toolchain: Introducing the Facets Orchestration Platform (link in comments). It walks through how Facets sits on top of your existing tools yet brings orchestration across the entire delivery lifecycle.

43

1 Comment

The myth of the “AI SRE” is everywhere. Most of it is noise. Engineers are under pressure to move fast and prevent downtime. But when it comes to “AI SREs,” not all approaches are equal. Language models can help, making it easier to summarize observational data and suggest possible explanations based on history, but they will only get you so far without the structured causal knowledge Causely provides. In my latest piece for CIO Dive, I break down how to separate hype from reality in #AI for reliability #engineering: https://lnkd.in/euPpscGp For a deeper, research-backed perspective, our Senior Applied AI Scientist explains in InfoQ why causal reasoning is the missing piece for those chasing the promise of an “AI SRE”: https://lnkd.in/eHZECD8a

64

9 Comments

In 2023, the DORA State of DevOps report found that using LLMs to generate code exacerbates the real bottlenecks in the development process, with quality problems canceling out the gains for most teams. "They'll get more reliable at 10x the scale. Just you wait and see!" In 2024, the DORA State of DevOps report found that using LLMs to generate code exacerbates the real bottlenecks in the development process, with quality problems canceling out the gains for most teams. "They'll get more reliable at 100x the scale. Just you wait and see!" In 2025, the DORA State of DevOps report found that using LLMs to generate code exacerbates the real bottlenecks in the development process, with quality problems canceling out the gains for most teams. "Ah, hmmm. Yeah, about those quality problems we said scaling would fix...?" This is it, folks. The tools are about as reliable as they're going to get in the foreseeable future. Perhaps it's time to start addressing the /real/ bottlenecks in your software development process? (Hint: it's got nothing to do with "A.I.")

416

40 Comments

Every computing paradigm has an open source reference application. There was the JPetStore for Java EE. There was the Sock Shop for cloud native computing. And now we have coffeeAGNTCY for multi-agent computing. coffeeAGNTCY is the fastest way to experiment with the agentic open source ecosystem. It's a complete reference app that shows how to build multi-agent software using #AGNTCY, Agent2Agent, MCP components. Think of it as the "Hello World" for the Internet of Agents. Most agent demos are smoke and mirrors. This one is a reference application - not a demo! You get to see the guts of the code. You get to extend it in interesting ways. You get to attach any software component to it to see how it would behave, benchmark it. It actually handles the unglamorous stuff - discovery, cross-framework communication, scaling past two agents, scaling across organizational and vendor boundaries. Start with Corto to see A2A and SLIM in action, then graduate to Lungo for the full multi-agent multi-org orchestration, including MCP endpoints. Read the blog by Luke Tucker and try it yourself: https://lnkd.in/grcVA96j

51

4 Comments