Blog post author here (I did not post the link above to HN). I also wrote a short follow-up blog post at https://philipotoole.com/proving-to-fable-i-maintain-the-rep...
5,064 karma · joined April 3, 2014
https://www.philipotoole.com
[1] https://www.rqlite.io
One application that comes up over and over again -- wanting relational modeling, along with HA, but with low operational costs. People sometimes start with Postgres, then need to set up HA, and find it's a lot of work. They realize that their data set requirements are not huge, don't need fancy features, so start to think Postgres is overkill. That's what Replicated found.
In Europe the individual has almost no legal reason to use force, and force by individuals is considered illegitimate. The "Sovereign" has all the coercive power in European states. In the US, however, a certain amount of legitimate force explicitly remains with the individual i.e. the 2nd amendment. (I am not making a value judgement here).
Of course, Europe has government with the consent of the governed, so is Lockean in that sense. But the balance of force between the "Sovereign" and the people in Europe is all Hobbes. You only notice it when you move to the US and compare it to Europe.
Europe had centuries of religious and civil war. It's not surprising Hobbes won out.
As for reliability - it's a fault-tolerant, highly available system. Reliability is the reason it exists. :-) If you're asking about quality and test coverage, you might like to check out these resources:
- https://rqlite.io/docs/design/
This is the only part of your response that doesn't quite sit right with me. There could be many "highly scrutinized and bureaucratic functions" out there that are working very well, you just don't notice because they work so well. There could be a selection-effect here.
Quality is a big deal for me[1]. But I think you're defining "quality" too narrowly in this context. "Quality" could also mean "allows everyone, at scale, reliably, to do what they need to do." The US Tax Filing system (and its associated software) meets that goal.
[1] https://philipotoole.com/always-thinking-of-the-next-guy/
Otherwise, if you try with more modern networks and disks, let me know what you see.
I did introduce Queued Writes[1] since that talk, allowing you to trade off performance versus immediate durability. It may interest you -- network is much less of a factor then, and you should get a 10-100x increase in throughput.
Yes I do have practical experience to share, I wrote a blog post on rqlite and FTS5: https://philipotoole.com/building-a-highly-available-search-...
Will it allow you to reach the same scale in terms of data set size that Elasticsearch supports? Almost certainly no, but it might be enough depending on your use case.
- https://www.philipotoole.com/2021-rqlite-cmu-tech-talk - see slide 33.
- There is also a recording that goes with the talk: https://www.youtube.com/watch?v=JLlIAWjvHxM
You can also read about Performance in the docs at: https://rqlite.io/docs/guides/performance/
An important thing to note: this testing was done 4+ years ago, on moderately-powerful hardware for the time. With higher-end, more modern hardware you may get even better results.
Nit: dqlite is a library, it is not a network-exposed database like rqlite is. Sure, it requires connecting to other nodes over the network, but local access is via in-process. In contrast one connects with rqlite over the network - HTTP specifically.
Exactly.
You are right that when someone (a human) submits a PR it didn't cost me anything (short of my time to review it). But those folks are not a team, not someone I could rely on or direct. Open-source projects -- successful ones -- often turn into a company, and then hire a dev team. We all know this.
I have no plans to commercialize rqlite, and I certainly couldn't afford a team of human developers. But I've got Copilot (and Gemini when I use it) now. So, in a sense, I now do have a team. And it's allowed me to fix bugs and add small features I wouldn't have bothered to in the past. It's definitely faster (20 mins to fire up my computer, write the code, push the PR vs. 5 mins to create the GitHub issue, assign to Copilot, review, and merge).
Case in point: I'm currently adding change-data-capture to rqlite. Development is going faster, but it's also more erratic because I'm reviewing more, and coding less. It reminds me of when I've been a TL of a software team.
From my perspective I didn't have a development team before. I have one now. I guess I am a member of that team now. But I hadn't thought of it like that -- another strange dimension to working Copilot (and its ilk).
What you're describing sounds like "none", but adding the "limit read staleness" requirement. That is described in the docs.
https://rqlite.io/docs/api/read-consistency/#limiting-read-s...
If you're interested in writing Raft-backed databases you might be interested in my talk at GopherCon 2023. It walks through doing exactly that, step by step: https://www.youtube.com/watch?v=8XbxQ1Epi5w
Agreed, this has been my experience too.
The section of the talk at the timestamp above is titled "Should I use Raft"?