HN user

west0n

279 karma

A database & K8s engineer

Posts13
Comments40
View on HN

As the founder and CEO of a two-year-old startup, seeking certainty in the direction amidst uncertainty (determining which products can bring customers and profits) should become my instinct.

See some development frameworks for local-first apps support both SQLite and PostgreSQL. The advantage of using PostgreSQL is that when you add a cloud option to your local-first app, the migration becomes much easier.

I think the codebase of Redis is too small, which makes the cost of maintaining a Redis fork relatively low. Moreover, Redis was forked too late. This puts Redis Inc. at a disadvantage in the battle against cloud vendors.

TLTR:

Understand and optimize relational databases (MySQL, PG) rather than introducing a new type of database (e.g., DynamoDB, Kafka, Redis) whenever issues arise. Common mistakes:

- Lack of indexes

- Using ORM (generating too many SQL queries or inefficient queries that return all columns)

Some suggestions:

- Use relational databases where possible, instead of NoSQL

- Use read-only replicas instead of Redis when applicable

- Keep relational databases clean and data size small

The author's message is: Hey, folks, if you don't have the ability to use relational databases properly, then your usage of Kafka and DynamoDB will certainly be a mess as well.

Recently, I learned that PostgreSQL can integrate with many languages through plugins, such as PL/Python, pgrx (Rust), pgzx (Zig), and so on. I wonder if anyone is planning to write one for Java or C#, lol.

According to this paper, GPU4FS is a file system that can run on the GPU and be accessed by applications. Since GPUs cannot make system calls, GPU4FS uses shared video memory (VRAM) and a parallel queue implementation. Applications running on the GPU can utilize GPU4FS after modifying their code, eliminating the need for a CPU-side file system when accessing the file system. The experiments are done on Optane memory.

It would be interesting to know if this approach could optimize the performance of training and inference for large models.

PostgreSQL is a very unique community compared to other database community(MySQL, Redis, MongoDB, ClickHouse et.c). It is INDEED decentralized, which means NO single company control it. This is related to the style of PostgreSQL's primary maintainers and leaders, who have ensured that the project's decisions and direction are determined collectively by the community members, rather than being controlled by a single company. Hopefully, their departure will not change this aspect.

Neal Gompa opened a discussion on the Fedora development list, noting the license change and the need to remove Redis from Fedora.

Gompa also raised the issue on openSUSE's Factory discussion list.

After Docker was phased out, various distributions have adopted the compatible Podman as a replacement for Docker. It seems that a similar story is unfolding with Redis.

I know that some companies are developing AI-based interview products. In a video conference, the AI asks questions and then evaluates the candidate's answers, giving a score. This can save a lot of time for the interviewer. Such a world is quite scary.