iiinova — where your digital self keeps evolving

Almog @almog.xyz · 23/04/2026

A deep dive to how metrics are stored and queried. Inverted indexes, roaring bitmaps, and ... gorillas? www.bitsxpages.com/p/how-metric...

bitsxpages.com

how metrics are stored and queried

the data structures powering your metrics dashboard

070

Almog @almog.xyz · 16/04/2026

I'd argue that's exactly what's in the price tag ;) that's because the price tag of managed services include that, which I'd argue aren't necessary for opendata timeseries

000

Almog @almog.xyz · 16/04/2026

The team worked insane hours to get this release ready and we've finally shipped it! We built a first of its kind, prometheus-compatible database that is object native. ~22x cheaper than AWS AMP and stupid easy to operate. www.opendata.dev/blog/introdu...

opendata.dev

OpenData Timeseries: Prometheus-compatible metrics on object storage | OpenData

An MIT-licensed, Prometheus-compatible timeseries database built on SlateDB, bringing the operating model and cost structure of object-store-native systems to the Prometheus and Grafana ecosystem.

2195

Almog @almog.xyz · 27/03/2026

I know "measure twice, cut once" applies to performance optimizations, but where's the fun in that? I am quite proud of my lopsided planter boxes and custom memory allocators.

020

Almog @almog.xyz · 24/03/2026

Thanks! I haven't listened to that talk, adding it to the backlog

010

Almog @almog.xyz · 23/03/2026

Databases have strange economic incentives that end up making them "worse" over time. The problem is an asymmetry between who benefits from innovations at first (database companies) and who benefits in the long run (hyperscalers that sell hardware). Blogged: www.bitsxpages.com/p/the-broken...

bitsxpages.com

the broken economics of databases

Why database companies charge so much, earn so little, and keep making things complicated

181

Almog @almog.xyz · 17/03/2026

I regularly get asked "how do you create your diagrams for your blogs"? Recently I've been using my own tool, and I'm open sourcing it! Yuzudraw is a visual editor for ASCII that has a token-efficient DSL that lets agents generate/modify your diagrams. Give it a ⭐! github.com/agavra/yuzud...

github.com

GitHub - agavra/yuzudraw: ASCII diagramming tool for both humans and agents

ASCII diagramming tool for both humans and agents. Contribute to agavra/yuzudraw development by creating an account on GitHub.

030

Almog @almog.xyz · 02/03/2026

“Gradually, then suddenly.” That’s how adoption works when you’re building something new. Opendata is still "gradually" but 100 stars with $0 spent on marketing is a good start. Back to building! github.com/opendata-oss...

091

Almog @almog.xyz · 12/02/2026

Yes! With larger payloads compression is almost always worth it. Thanks for sharing

000

Almog @almog.xyz · 10/02/2026

I implemented prefix compression for SlateDB & noticed benchmarks looked "worse". Fell down a rabbit hole. Turns out I was thinking about compression backwards... Wrote up my learning: www.bitsxpages.com/p/the-mathem...

bitsxpages.com

the mathematics of compression in database systems

why compression is (almost) always worthwhile

1101

Almog @almog.xyz · 28/01/2026

Can you beat 180KB? I created a challenge to reduce a dataset as much as possible 🏆 my approach uses delta encoding & prefix compression + zstd(22) compression to reduce 25MB -> 180KB github.com/agavra/bit-g...

github.com

GitHub - agavra/bit-golf: a compression golf challenge for GitHub event data

a compression golf challenge for GitHub event data - agavra/bit-golf

082

Almog @almog.xyz · 09/01/2026

by the way, it's pronounced like "tweaker"

010

Almog @almog.xyz · 09/01/2026

I fixed the problem with reviewing code written by Claude in "accept edits on" mode: github.com/agavra/tuicr would love to know what you think, and if you want to contribute to an OSS rust project there's a bunch of open issues to pick up!

github.com

GitHub - agavra/tuicr: Review AI-generated diffs like a GitHub pull request, right from your terminal.

Review AI-generated diffs like a GitHub pull request, right from your terminal. - agavra/tuicr

120

Almog @almog.xyz · 05/01/2026

I'll die on this hill: Sorted String Tables (SSTs) are the single most important data structure for modern DBs. They lean in to the limitations of SSDs and Object Storage, making them (and similar layouts) the best choice for many databases. Blogged in detail: www.bitsxpages.com/p/sorted-str...

bitsxpages.com

sorted string tables (SST) from first principles

why sorted string tables are the swiss army knife for data systems and how they are implemented

060

Almog @almog.xyz · 10/12/2025

thanks for the kind words! I use monodraw.helftone.com -- not sure if planetscale/turbopuffer use the same but I was definitely inspired by their design style

monodraw.helftone.com

Monodraw for macOS — Helftone

010

Almog @almog.xyz · 09/12/2025

The inner join between sets of people who build databases, write, and draw? Low cardinality. I'm in that set, so I'm starting a blog! Here's my first post: www.bitsxpages.com/p/frameworks...

bitsxpages.com

frameworks for understanding databases

building mental models for tradeoffs in performance, availability and durability in data systems

2163

Almog @almog.xyz · 25/11/2025

Sometimes the best solution is "do nothing", but it's always more fun to play with tools.

040

Almog @almog.xyz · 21/10/2025

Calling database nerds in SF! I'm covering SlateDB at the systems meetup next Wednesday (10/29). If you're around, I'd love to meet you in person (that way you'll have proof I'm not just an AI bot). 👉 luma.com/e7feg2i6

0153

Almog @almog.xyz · 08/08/2025

I wonder why it hasn't made it's way to the US! The only time I get it is when I make it at home.

120

Almog @almog.xyz · 08/08/2025

Beans on toast is underrated.

210

Almog @almog.xyz · 07/08/2025

I recently implemented Gorilla encoding (www.vldb.org/pvldb/vol8/p...) for a SlateDB PR. Pretty cool stuff - easy to understand but really powerful. Here it is, explained by a gorilla.

091

Almog @almog.xyz · 05/08/2025

I guess the answer to that is “technically, yes.”

000

Almog @almog.xyz · 05/08/2025

It's true. Every company eventually becomes a database company.

0101

Almog @almog.xyz · 15/07/2025

Despite using so many new technologies, I somehow never learn my lesson: read the docs sooner and read the docs thoroughly.

092

Almog @almog.xyz · 08/07/2025

Reading the tokio-rs async documentation makes me feel like...

161

Almog @almog.xyz · 02/07/2025

that's so cool - I love contraptions that are just there to show that we can do something!

000

Almog @almog.xyz · 01/07/2025

No shame in appreciating bona fide nerd humor 😉

010

Almog @almog.xyz · 01/07/2025

One day I'll open a coffee shop dedicated to the not-insignificant intersection between database nerds and coffee snobs. Until then, enjoy this comic.

1395

Almog @almog.xyz · 28/04/2025

Maybe... just maybe, adding more features and complexity into stream processors is NOT what we need?

061

Almog @almog.xyz · 24/04/2025

The new electric Caltrain cars have WiFi. 🙏 SF bay area has finally entered the 21st century (on this dimension of public transit only).

010

Reposted by Almog

Chris @chris.blue · 22/04/2025

Today marks SlateDB’s one year anniversary! It’s been a lot of fun. Thanks to @rohanpd.bsky.social @flaneur2024.bsky.social @almog.ai @vigneshc.bsky.social @paulbutler.org Jason Gustafson, David Moravek, and many others for joining the project. 😀

slatedb.io

SlateDB - An embedded storage engine built on object storage | SlateDB

Description will go into a meta tag in <head />

0165

Almog @almog.xyz · 21/04/2025

This got me thinking: Should MCP itself formally distinguish between request vs response context? It could let LLMs be more intelligent about when to pull extra info. 🤔 [6/6]

000

Almog @almog.xyz · 21/04/2025

I built a demo agent for hotel booking. My setup: Inject request context into the prompt (user prefs, budget, pulled from account) Fetch response context live using MCP tools (available hotels that match search params) Result: the agent felt way smarter. [5/N]

100

Almog @almog.xyz · 21/04/2025

Back when we only had prompts, you could frame the task (request context). Today, tools like MCP change the game. But should you use them? [4/N]

100

Almog @almog.xyz · 21/04/2025

Example: a hotel booking agent 🛌 Request context: "Book a hotel in Barcelona on April 28 under $400" Response context: A list of hotels available in Barcelona that day, within budget. [3/N]

100

Almog @almog.xyz · 21/04/2025

⚖️ You need to balance two types of context: Request Context = frames the task Response Context = info needed to complete it [2/N]

100

Almog @almog.xyz · 21/04/2025

Prompt engineering was v0. Context engineering is v1.0. How should you think about supplying LLMs with the right information? 📖 [1/N]

140

Almog @almog.xyz · 10/02/2025

"Kafka Configs: A Rube Goldberg Machine?" - an actual quote from a customer after discussing cleanup.policy and Kafka Streams changelogs.

030

Almog @almog.xyz · 04/02/2025

This is really neat, thanks for sharing. Little performance hacks like this require understanding how things work underneath the hood, and tend to have the most impact on overall performance! Would love to know other similar tips you have.

020

Reposted by Almog

Chris @chris.blue · 03/02/2025

SlateDB now has clones 🤯 Users can clone an existing DB's data to a new location. It's nearly instantaneous since it references the data from the old bucket rather than copying. Writes to the clone update the new location. Compaction lazily merges old data into the new directory./ht @responsive.dev

github.com

Initial clone implementation by hachikuji · Pull Request #430 · slatedb/slatedb

Fixes #315. This patch contains the logic to create and initialize clones as outlined in RFC-0004.

1332

Reposted by Almog

Apurva Mehta @apurvamehta.com · 04/02/2025

Should you use Kafka? If so, when? And what are the tradeoffs presented by the dizzying variety of Kafka-adjacent technologies? I hope that my latest blog post provides unique and useful answers to these questions. Let me know what you all think! 👉 www.responsive.dev/blog/why-whe...

0155

Almog @almog.xyz · 31/01/2025

I often have days where GPT is immensely helpful, today I had one where it had led me down multiple rabbit holes and I ended up solving my problem with good old Google search. Nice reminder of when it’s helpful and when it’s not!

030

Almog @almog.xyz · 30/01/2025

Username checks out 😆

010

Almog @almog.xyz · 30/01/2025

As part of my post-Elden Ring hunt for a game, I tried Balatro last night. I'm starting to think I may have made a mistake.

130

Reposted by Almog

ScyllaDB @scylladb.com · 29/01/2025

How did Responsive simplify its architecture and achieve monstrously high availability with unlimited scale potential? Co-founder @almog.ai will share how replacing RocksDB with #ScyllaDB in #Kafka Streams led to big wins for their team at Monster Scale Summit. www.scylladb.com/monster-scal...

084

Almog @almog.xyz · 29/01/2025

My hobbies all have the ability to "reclaim" waste: 🪴 for ceramics, wasted clay from throwing goes right back in to your next pot 🪵 for wood working, reusing 2x4s is second nature 🥘 for cooking, leftover food is tossed into salads 💻 what's for software? bin-packing containers? 😆

040

Almog @almog.xyz · 28/01/2025

Kafka Streams 101: Windows & Time! 🕰️ What's the difference between event, stream and wall clock time? 🪟 What are the four different types of windows? ⁉️ What are the important error messages and metrics, and what do they mean? 👉 Read the full lesson here: www.responsive.dev/blog/windows...

The different types of windows in Kafka Streams

084

Almog @almog.xyz · 28/01/2025

Haha I often do! My last jam was www.dishoom.com/store/produc... from the Dishoom Cookbook. Would recommend! Sweet and a touch of spicy.

110

Almog @almog.xyz · 28/01/2025

I did see your other post (which is what inspired this one!) but I actually haven't noticed that same problem recently. Maybe we have a smaller notion space?

100