--- slug: 2019/09/binary-format-shootout title: "Binary format shootout" date: 2019-09-28 12:00:00 authors: [bspeice] tags: [] --- I've found that in many personal projects, [analysis paralysis](https://en.wikipedia.org/wiki/Analysis_paralysis) is particularly deadly. Making good decisions in the beginning avoids pain and suffering later; if extra research prevents future problems, I'm happy to continue ~~procrastinating~~ researching indefinitely. So let's say you're in need of a binary serialization format. Data will be going over the network, not just in memory, so having a schema document and code generation is a must. Performance is crucial, so formats that support zero-copy de/serialization are given priority. And the more languages supported, the better; I use Rust, but can't predict what other languages this could interact with. Given these requirements, the candidates I could find were: 1. [Cap'n Proto](https://capnproto.org/) has been around the longest, and is the most established 2. [Flatbuffers](https://google.github.io/flatbuffers/) is the newest, and claims to have a simpler encoding 3. [Simple Binary Encoding](https://github.com/real-logic/simple-binary-encoding) has the simplest encoding, but the Rust implementation is unmaintained Any one of these will satisfy the project requirements: easy to transmit over a network, reasonably fast, and polyglot support. But how do you actually pick one? It's impossible to know what issues will follow that choice, so I tend to avoid commitment until the last possible moment. Still, a choice must be made. Instead of worrying about which is "the best," I decided to build a small proof-of-concept system in each format and pit them against each other. All code can be found in the [repository](https://github.com/speice-io/marketdata-shootout) for this post. We'll discuss more in detail, but a quick preview of the results: - Cap'n Proto: Theoretically performs incredibly well, the implementation had issues - Flatbuffers: Has some quirks, but largely lived up to its "zero-copy" promises - SBE: Best median and worst-case performance, but the message structure has a limited feature set ## Prologue: Binary Parsing with Nom Our benchmark system will be a simple data processor; given depth-of-book market data from [IEX](https://iextrading.com/trading/market-data/#deep), serialize each message into the schema format, read it back, and calculate total size of stock traded and the lowest/highest quoted prices. This test isn't complex, but is representative of the project I need a binary format for. But before we make it to that point, we have to actually read in the market data. To do so, I'm using a library called [`nom`](https://github.com/Geal/nom). Version 5.0 was recently released and brought some big changes, so this was an opportunity to build a non-trivial program and get familiar. If you don't already know about `nom`, it's a "parser generator". By combining different smaller parsers, you can assemble a parser to handle complex structures without writing tedious code by hand. For example, when parsing [PCAP files](https://www.winpcap.org/ntar/draft/PCAP-DumpFileFormat.html#rfc.section.3.3): ``` 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +---------------------------------------------------------------+ 0 | Block Type = 0x00000006 | +---------------------------------------------------------------+ 4 | Block Total Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 8 | Interface ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 12 | Timestamp (High) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 16 | Timestamp (Low) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 20 | Captured Len | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 24 | Packet Len | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Packet Data | | ... | ``` ...you can build a parser in `nom` that looks like [this](https://github.com/speice-io/marketdata-shootout/blob/369613843d39cfdc728e1003123bf87f79422497/src/parsers.rs#L59-L93): ```rust const ENHANCED_PACKET: [u8; 4] = [0x06, 0x00, 0x00, 0x00]; pub fn enhanced_packet_block(input: &[u8]) -> IResult<&[u8], &[u8]> { let ( remaining, ( block_type, block_len, interface_id, timestamp_high, timestamp_low, captured_len, packet_len, ), ) = tuple(( tag(ENHANCED_PACKET), le_u32, le_u32, le_u32, le_u32, le_u32, le_u32, ))(input)?; let (remaining, packet_data) = take(captured_len)(remaining)?; Ok((remaining, packet_data)) } ``` While this example isn't too interesting, more complex formats (like IEX market data) are where [`nom` really shines](https://github.com/speice-io/marketdata-shootout/blob/369613843d39cfdc728e1003123bf87f79422497/src/iex.rs). Ultimately, because the `nom` code in this shootout was the same for all formats, we're not too interested in its performance. Still, it's worth mentioning that building the market data parser was actually fun; I didn't have to write tons of boring code by hand. ## Cap'n Proto Now it's time to get into the meaty part of the story. Cap'n Proto was the first format I tried because of how long it has supported Rust (thanks to [dwrensha](https://github.com/dwrensha) for maintaining the Rust port since [2014!](https://github.com/capnproto/capnproto-rust/releases/tag/rustc-0.10)). However, I had a ton of performance concerns once I started using it. To serialize new messages, Cap'n Proto uses a "builder" object. This builder allocates memory on the heap to hold the message content, but because builders [can't be re-used](https://github.com/capnproto/capnproto-rust/issues/111), we have to allocate a new buffer for every single message. I was able to work around this with a [special builder](https://github.com/speice-io/marketdata-shootout/blob/369613843d39cfdc728e1003123bf87f79422497/src/capnp_runner.rs#L17-L51) that could re-use the buffer, but it required reading through Cap'n Proto's [benchmarks](https://github.com/capnproto/capnproto-rust/blob/master/benchmark/benchmark.rs#L124-L156) to find an example, and used [`std::mem::transmute`](https://doc.rust-lang.org/std/mem/fn.transmute.html) to bypass Rust's borrow checker. The process of reading messages was better, but still had issues. Cap'n Proto has two message encodings: a ["packed"](https://capnproto.org/encoding.html#packing) representation, and an "unpacked" version. When reading "packed" messages, we need a buffer to unpack the message into before we can use it; Cap'n Proto allocates a new buffer for each message we unpack, and I wasn't able to figure out a way around that. In contrast, the unpacked message format should be where Cap'n Proto shines; its main selling point is that there's [no decoding step](https://capnproto.org/). However, accomplishing zero-copy deserialization required code in the private API ([since fixed](https://github.com/capnproto/capnproto-rust/issues/148)), and we allocate a vector on every read for the segment table. In the end, I put in significant work to make Cap'n Proto as fast as possible, but there were too many issues for me to feel comfortable using it long-term. ## Flatbuffers This is the new kid on the block. After a [first attempt](https://github.com/google/flatbuffers/pull/3894) didn't pan out, official support was [recently launched](https://github.com/google/flatbuffers/pull/4898). Flatbuffers intends to address the same problems as Cap'n Proto: high-performance, polyglot, binary messaging. The difference is that Flatbuffers claims to have a simpler wire format and [more flexibility](https://google.github.io/flatbuffers/flatbuffers_benchmarks.html). On the whole, I enjoyed using Flatbuffers; the [tooling](https://crates.io/crates/flatc-rust) is nice, and unlike Cap'n Proto, parsing messages was actually zero-copy and zero-allocation. However, there were still some issues. First, Flatbuffers (at least in Rust) can't handle nested vectors. This is a problem for formats like the following: ``` table Message { symbol: string; } table MultiMessage { messages:[Message]; } ``` We want to create a `MultiMessage` which contains a vector of `Message`, and each `Message` itself contains a vector (the `string` type). I was able to work around this by [caching `Message` elements](https://github.com/speice-io/marketdata-shootout/blob/e9d07d148bf36a211a6f86802b313c4918377d1b/src/flatbuffers_runner.rs#L83) in a `SmallVec` before building the final `MultiMessage`, but it was a painful process that I believe contributed to poor serialization performance. Second, streaming support in Flatbuffers seems to be something of an [afterthought](https://github.com/google/flatbuffers/issues/3898). Where Cap'n Proto in Rust handles reading messages from a stream as part of the API, Flatbuffers just sticks a `u32` at the front of each message to indicate the size. Not specifically a problem, but calculating message size without that tag is nigh on impossible. Ultimately, I enjoyed using Flatbuffers, and had to do significantly less work to make it perform well. ## Simple Binary Encoding Support for SBE was added by the author of one of my favorite [Rust blog posts](https://web.archive.org/web/20190427124806/https://polysync.io/blog/session-types-for-hearty-codecs/). I've [talked previously](/2019/06/high-performance-systems) about how important variance is in high-performance systems, so it was encouraging to read about a format that [directly addressed](https://github.com/real-logic/simple-binary-encoding/wiki/Why-Low-Latency) my concerns. SBE has by far the simplest binary format, but it does make some tradeoffs. Both Cap'n Proto and Flatbuffers use [message offsets](https://capnproto.org/encoding.html#structs) to handle variable-length data, [unions](https://capnproto.org/language.html#unions), and various other features. In contrast, messages in SBE are essentially [just structs](https://github.com/real-logic/simple-binary-encoding/blob/master/sbe-samples/src/main/resources/example-schema.xml); variable-length data is supported, but there's no union type. As mentioned in the beginning, the Rust port of SBE works well, but is [essentially unmaintained](https://users.rust-lang.org/t/zero-cost-abstraction-frontier-no-copy-low-allocation-ordered-decoding/11515/9). However, if you don't need union types, and can accept that schemas are XML documents, it's still worth using. SBE's implementation had the best streaming support of all formats I tested, and doesn't trigger allocation during de/serialization. ## Results After building a test harness [for](https://github.com/speice-io/marketdata-shootout/blob/master/src/capnp_runner.rs) [each](https://github.com/speice-io/marketdata-shootout/blob/master/src/flatbuffers_runner.rs) [format](https://github.com/speice-io/marketdata-shootout/blob/master/src/sbe_runner.rs), it was time to actually take them for a spin. I used [this script](https://github.com/speice-io/marketdata-shootout/blob/master/run_shootout.sh) to run the benchmarks, and the raw results are [here](https://github.com/speice-io/marketdata-shootout/blob/master/shootout.csv). All data reported below is the average of 10 runs on a single day of IEX data. Results were validated to make sure that each format parsed the data correctly. ### Serialization This test measures, on a [per-message basis](https://github.com/speice-io/marketdata-shootout/blob/master/src/main.rs#L268-L272), how long it takes to serialize the IEX message into the desired format and write to a pre-allocated buffer. | Schema | Median | 99th Pctl | 99.9th Pctl | Total | | :------------------- | :----- | :-------- | :---------- | :----- | | Cap'n Proto Packed | 413ns | 1751ns | 2943ns | 14.80s | | Cap'n Proto Unpacked | 273ns | 1828ns | 2836ns | 10.65s | | Flatbuffers | 355ns | 2185ns | 3497ns | 14.31s | | SBE | 91ns | 1535ns | 2423ns | 3.91s | ### Deserialization This test measures, on a [per-message basis](https://github.com/speice-io/marketdata-shootout/blob/master/src/main.rs#L294-L298), how long it takes to read the previously-serialized message and perform some basic aggregation. The aggregation code is the same for each format, so any performance differences are due solely to the format implementation. | Schema | Median | 99th Pctl | 99.9th Pctl | Total | | :------------------- | :----- | :-------- | :---------- | :----- | | Cap'n Proto Packed | 539ns | 1216ns | 2599ns | 18.92s | | Cap'n Proto Unpacked | 366ns | 737ns | 1583ns | 12.32s | | Flatbuffers | 173ns | 421ns | 1007ns | 6.00s | | SBE | 116ns | 286ns | 659ns | 4.05s | ## Conclusion Building a benchmark turned out to be incredibly helpful in making a decision; because a "union" type isn't important to me, I can be confident that SBE best addresses my needs. While SBE was the fastest in terms of both median and worst-case performance, its worst case performance was proportionately far higher than any other format. It seems to be that de/serialization time scales with message size, but I'll need to do some more research to understand what exactly is going on.