New post on why modern observability is so expensive.

mattklein123.dev/2024/04/03/obs…

Nothing spells out how broken the U.S. immigration system is than learning about how exceptionally qualified, highly skilled and highly paid people weren’t selected in the H-1B lottery.

I can’t think of any country having a more self-destructive immigration policy. It’s tragic.

Every time I open this app, I feel like I’m drowning in ads and bots.

Not only does it totally kill the vibe, it doesn’t even want to make me post anything.

Tech twitter seemed like a magical place until 2022 (at least for me), and I don’t think we’ll ever get that back.

“Automatic Root Cause Analysis via Large Language Models for Cloud Incidents”

It was only a matter of time before we started seeing papers using LLMs for infrastructure and operations tasks, and here’s the first (to my knowledge) from eurosys

arxiv.org/abs/2305.15778

“AIOps”

Even being able to answer questions based on a stream of unstructured logs will be useful.

But if a monitoring tool can understand your codebase, and for every error or crash reported, if it can extract additional context, that’d be a game changer. Anyone building this?

I feel the so called “death of Google” has been greatly exaggerated by people with their own ideological axe to grind.

It’s wild that Gemini’s groundbreaking achievements have been soundly ignored, whereas everyone and their dogs seem obsessed about the image generation gaffes.

Maybe it’s just me, but these obviously look like AI generated images?

I’m not scared, tbh.

More like, sad that we’ll be drowning in a sea of this stuff for eternity.

And as I’ve said before, ironically, this will make the genuine article all the more coveted and high-value.

One of the best investments to make when it comes to ensuring reliability is either buying or building a reasonably sophisticated deployment tool, especially one that can handle atomic rollbacks.

It’s the best way to ensure you’re building fast without breaking things (too much)

End of ZIRP era had also meant many tech companies whose engineering blogs I used to follow now publish considerably less.

If one were to look back at what some of these companies built circa 2015-2022, I fully expect to see a ton of NIH projects that’d not be approved in 2024.

I’d seriously consider paying for Twitter Blue if it allows me to get rid of the “For You” tab.

The content there is days old posts from people whom I follow that I’ve missed (that somehow never showed up on my “Following” timeline ever), or random junk of the lowest quality.

Given the discourse on this hellsite about DEI, you’d think companies are jam packed with unqualified DEI candidates who’re all getting hired and promoted ahead of more qualified people. 🙄

Is that the case at a *single* corporate organization? Can someone name one? Just one?

Less discussed are the second and third order impacts of this era. It impacted everything from:

- kind of tech that was built
- the kind of engineering leaders and senior engineers who emerged
- career ladders
- pricing of dev tools SaaS
- workforce composition
- quality of life

Mission/Mexican hot chocolate in oat milk from Dandelion Chocolate on a cold, drizzly January afternoon is one of the small delights of winter in San Francisco.

Speaking of the tech job market: people with strong networks, and specifically, strong ties to executives are not having much trouble finding new jobs, from anecdotal evidence. This was always the case, but becomes more brutally apparent during a downturn.

Being an immigrant tech worker means having decades old paystubs from previous employers, scans of every passport you've ever had (including all blank pages), I-94 records (before it went digital), and more, all tucked away in Google Drive in pdf format for repeated use.

When I think of “hall of fame” blog posts (had an enduring influence on the industry, made a real difference in the way people think about a problem), I think of:

- use boring technology by Dan McKinley
- being glue by Tanya Reilly (now at @[email protected])
- the engineer manager pendulum by Charity Majors

Every year I post a “best tech/papers of the year” list.

I try to write a detailed blog post, with a few lines on why I liked or what I learned from each of the items on the list.

Here’s my list for 2023, if anyone is interested. I’ll have the blog post ready by early Jan

I distinctly remember 2020 as the year of covid (and the year my dear grandma passed away), but 2021 - 2023 feels like a big indistinguishable blur to me.

It's kind of shocking to realize this blur encompasses 3 freaking entire years!

Really hoping for a more vibrant 2024.

New blog post, on cache eviction, the SIEVE algorithm, and some variant with interesting properties: brooker.co.za/blog/2023/12/1…

Amazon microservices architecure - 2008 vs 2023

If you thought old school microservices “death star” architecture was bad, lo, behold, we have now have a “galaxy brain” architecture. 😢