Apache Arrow VS FlatBuffers

Apache Arrow : It contains a set of technologies that enable big data systems to process and move data fast

One of the main selling points of Polars over similar solutions such as Pandas is performance. Polars is written in highly optimized Rust and uses the Apache Arrow container format.

Kotlin DataFrame v0.14 comes with improvements for reading Apache Arrow format, especially loading a DataFrame from any ArrowReader. This improvement can be used to easily load results from analytical databases (such as DuckDB, ClickHouse) directly into Kotlin DataFrame.

Apache Arrow

It's this kind of certainty that underscores the vital role of the Apache Software Foundation (ASF). Many first encounter Apache through its pioneering project, the open-source web server framework that remains ubiquitous in web operations today. The ASF was initially created to hold the intellectual property and assets of the Apache project, and it has since evolved into a cornerstone for open-source projects worldwide. The ASF enforces strict standards for diverse contributions, independence, and activity in its projects, ensuring they can withstand the test of time as standards in software development. Many open-source projects strive to become Apache projects to gain the community credibility necessary for adoption as standard software building blocks, such as Apache Tomcat for Java web applications, Apache Arrow for in-memory data representation, and Apache Parquet for data file formatting, among others.

Apache Doris 2.1 has a data transmission channel built on Arrow Flight SQL. (Apache Arrow is a software development platform designed for high data movement efficiency across systems and languages, and the Arrow format aims for high-performance, lossless data exchange.) It allows high-speed, large-scale data reading from Doris via SQL in various mainstream programming languages. For target clients that also support the Arrow format, the whole process will be free of serialization/deserialization, thus no performance loss. Another upside is, Arrow Flight can make full use of multi-node and multi-core architecture and implement parallel data transfer, which is another enabler of high data throughput.

In comes Polars: a brand new dataframe library, or how the author Ritchie Vink describes it... a query engine with a dataframe frontend. Polars is built on top of the Arrow memory format and is written in Rust, which is a modern performant and memory-safe systems programming language similar to C/C++.

I learned yesterday about GoLang's assembler https://go.dev/doc/asm - after browsing how arrow is implemented for different languages (my experience is mainly C/C++) - https://github.com/apache/arrow/tree/main/go/arrow/math - there are bunch of .S ("asm" files) and I'm still not able to comprehend how these work exactly (I guess it'll take more reading) - it seems very peculiar.

The last time I've used inlined assembly was back in Turbo/Borland Pascal, then bit in Visual Studio (32-bit), until they got disabled. Then did very little gcc with their more strict specification (while the former you had to know how the ABI worked, the latter too - but it was specced out).

Anyway - I wasn't expecting to find this in "Go" :) But I guess you can always start with .go code then produce assembly (-S) then optimize it, or find/hire someone to do it.

According to the official website:

> Protobuf is intentionally designed to NOT require any parsing at all.

As others have mentioned, this is simply not the case, and the VARINT encoding is a trivial counterexample.

It is this required decoding/parsing that (largely) distinguishes protobuf from Google's flatbuffers:

https://github.com/google/flatbuffers

https://flatbuffers.dev/

Cap'n Proto (developed by Kenton Varda, the former Google engineer who, while at Google, re-wrote/refactored Google's protobuf to later open source it as the library we all know today) is another example of zero-copy (de)serialization.

I don't work at Cloudflare but follow their work and occasionally work on performance sensitive projects.

If I had to guess, they looked at the landscape a bit like I do and regarded Cap'n Proto, flatbuffers, SBE, etc. as being in one category apart from other data formats like Avro, protobuf, and the like.

So once you're committed to record'ish shaped (rather than columnar like Parquet) data that has an upfront parse time of zero (nominally, there could be marshalling if you transmogrify the field values on read), the list gets pretty short.

https://capnproto.org/news/2014-06-17-capnproto-flatbuffers-... goes into some of the trade-offs here.

Cap'n Proto was originally made for https://sandstorm.io/. That work (which Kenton has presumably done at Cloudflare since he's been employed there) eventually turned into Cloudflare workers.

Another consideration: https://github.com/google/flatbuffers/issues/2#issuecomment-...

Much more in the discussion of this particular issue onGitHub: flatbuffers:Reverse engineering #4258

Related but not Rust-specific: FlatBuffers, Cap'n Proto.

If json files are a concern because of space, you can always look into something like protobuffers or flatbuffers. But whatever you use, you should try to find a solution where you don't have to think about the actual serialization/deserialization of your objects, and can just concentrate on the data.