flodgatt/benches/parse_redis.rs

70 lines
6.5 KiB
Rust
Raw Normal View History

use criterion::{black_box, criterion_group, criterion_main, Criterion};
Improve handling of large Redis input (#143) * Implement faster buffered input This commit implements a modified ring buffer for input from Redis. Specifically, Flodgatt now limits the amount of data it fetches from Redis in one syscall to 8 KiB (two pages on most systems). Flodgatt will process all complete messages it receives from Redis and then re-use the same buffer for the next time it retrieves data. If Flodgatt received a partial message, it will copy the partial message to the beginning of the buffer before its next read. This change has little effect on Flodgatt under light load (because it was rare for Redis to have more than 8 KiB of messages available at any one time). However, my hope is that this will significantly reduce memory use on the largest instances. * Improve handling of backpresure This commit alters how Flodgatt behaves if it receives enough messages for a single client to fill that clients channel. (Because the clients regularly send their messages, should only occur if a single client receives a large number of messages nearly simultaneously; this is rare, but could occur, especially on large instances). Previously, Flodgatt would drop messages in the rare case when the client's channel was full. Now, Flodgatt will pause the current Redis poll and yield control back to the client streams, allowing the clients to empty their channels; Flodgatt will then resume polling Redis/sending the messages it previously received. With the approach, Flodgatt will never drop messages. However, the risk to this approach is that, by never dropping messages, Flodgatt does not have any way to reduce the amount of work it needs to do when under heavy load – it delays the work slightly, but doesn't reduce it. What this means is that it would be *theoretically* possible for Flodgatt to fall increasingly behind, if it is continuously receiving more messages than it can process. Due to how quickly Flodgatt can process messages, though, I suspect this would only come up if an admin were running Flodgatt in a *significantly* resource constrained environment, but I wanted to mention it for the sake of completeness. This commit also adds a new /status/backpressure endpoint that displays the current length of the Redis input buffer (which should typically be low or 0). Like the other /status endpoints, this endpoint is only enabled when Flodgatt is compiled with the `stub_status` feature.
2020-04-27 22:03:05 +02:00
use flodgatt::response::{RedisMsg, RedisParseOutput};
use lru::LruCache;
use std::convert::TryFrom;
fn parse_long_redis_input<'a>(input: &'a str) -> RedisMsg<'a> {
if let RedisParseOutput::Msg(msg) = RedisParseOutput::try_from(input).unwrap() {
assert_eq!(msg.timeline_txt, "timeline:1");
msg
} else {
panic!()
}
}
Improve handling of large Redis input (#143) * Implement faster buffered input This commit implements a modified ring buffer for input from Redis. Specifically, Flodgatt now limits the amount of data it fetches from Redis in one syscall to 8 KiB (two pages on most systems). Flodgatt will process all complete messages it receives from Redis and then re-use the same buffer for the next time it retrieves data. If Flodgatt received a partial message, it will copy the partial message to the beginning of the buffer before its next read. This change has little effect on Flodgatt under light load (because it was rare for Redis to have more than 8 KiB of messages available at any one time). However, my hope is that this will significantly reduce memory use on the largest instances. * Improve handling of backpresure This commit alters how Flodgatt behaves if it receives enough messages for a single client to fill that clients channel. (Because the clients regularly send their messages, should only occur if a single client receives a large number of messages nearly simultaneously; this is rare, but could occur, especially on large instances). Previously, Flodgatt would drop messages in the rare case when the client's channel was full. Now, Flodgatt will pause the current Redis poll and yield control back to the client streams, allowing the clients to empty their channels; Flodgatt will then resume polling Redis/sending the messages it previously received. With the approach, Flodgatt will never drop messages. However, the risk to this approach is that, by never dropping messages, Flodgatt does not have any way to reduce the amount of work it needs to do when under heavy load – it delays the work slightly, but doesn't reduce it. What this means is that it would be *theoretically* possible for Flodgatt to fall increasingly behind, if it is continuously receiving more messages than it can process. Due to how quickly Flodgatt can process messages, though, I suspect this would only come up if an admin were running Flodgatt in a *significantly* resource constrained environment, but I wanted to mention it for the sake of completeness. This commit also adds a new /status/backpressure endpoint that displays the current length of the Redis input buffer (which should typically be low or 0). Like the other /status endpoints, this endpoint is only enabled when Flodgatt is compiled with the `stub_status` feature.
2020-04-27 22:03:05 +02:00
// fn parse_to_timeline(msg: RedisMsg) -> Timeline {
// let trimmed_tl_txt = &msg.timeline_txt["timeline:".len()..];
// let tl = Timeline::from_redis_text(trimmed_tl_txt, &mut LruCache::new(1000)).unwrap();
// assert_eq!(tl, Timeline(User(Id(1)), Federated, All));
// tl
// }
// fn parse_to_checked_event(msg: RedisMsg) -> EventKind {
// EventKind::TypeSafe(serde_json::from_str(msg.event_txt).unwrap())
// }
Improve handling of large Redis input (#143) * Implement faster buffered input This commit implements a modified ring buffer for input from Redis. Specifically, Flodgatt now limits the amount of data it fetches from Redis in one syscall to 8 KiB (two pages on most systems). Flodgatt will process all complete messages it receives from Redis and then re-use the same buffer for the next time it retrieves data. If Flodgatt received a partial message, it will copy the partial message to the beginning of the buffer before its next read. This change has little effect on Flodgatt under light load (because it was rare for Redis to have more than 8 KiB of messages available at any one time). However, my hope is that this will significantly reduce memory use on the largest instances. * Improve handling of backpresure This commit alters how Flodgatt behaves if it receives enough messages for a single client to fill that clients channel. (Because the clients regularly send their messages, should only occur if a single client receives a large number of messages nearly simultaneously; this is rare, but could occur, especially on large instances). Previously, Flodgatt would drop messages in the rare case when the client's channel was full. Now, Flodgatt will pause the current Redis poll and yield control back to the client streams, allowing the clients to empty their channels; Flodgatt will then resume polling Redis/sending the messages it previously received. With the approach, Flodgatt will never drop messages. However, the risk to this approach is that, by never dropping messages, Flodgatt does not have any way to reduce the amount of work it needs to do when under heavy load – it delays the work slightly, but doesn't reduce it. What this means is that it would be *theoretically* possible for Flodgatt to fall increasingly behind, if it is continuously receiving more messages than it can process. Due to how quickly Flodgatt can process messages, though, I suspect this would only come up if an admin were running Flodgatt in a *significantly* resource constrained environment, but I wanted to mention it for the sake of completeness. This commit also adds a new /status/backpressure endpoint that displays the current length of the Redis input buffer (which should typically be low or 0). Like the other /status endpoints, this endpoint is only enabled when Flodgatt is compiled with the `stub_status` feature.
2020-04-27 22:03:05 +02:00
// fn parse_to_dyn_event(msg: RedisMsg) -> EventKind {
// EventKind::Dynamic(serde_json::from_str(msg.event_txt).unwrap())
// }
Improve handling of large Redis input (#143) * Implement faster buffered input This commit implements a modified ring buffer for input from Redis. Specifically, Flodgatt now limits the amount of data it fetches from Redis in one syscall to 8 KiB (two pages on most systems). Flodgatt will process all complete messages it receives from Redis and then re-use the same buffer for the next time it retrieves data. If Flodgatt received a partial message, it will copy the partial message to the beginning of the buffer before its next read. This change has little effect on Flodgatt under light load (because it was rare for Redis to have more than 8 KiB of messages available at any one time). However, my hope is that this will significantly reduce memory use on the largest instances. * Improve handling of backpresure This commit alters how Flodgatt behaves if it receives enough messages for a single client to fill that clients channel. (Because the clients regularly send their messages, should only occur if a single client receives a large number of messages nearly simultaneously; this is rare, but could occur, especially on large instances). Previously, Flodgatt would drop messages in the rare case when the client's channel was full. Now, Flodgatt will pause the current Redis poll and yield control back to the client streams, allowing the clients to empty their channels; Flodgatt will then resume polling Redis/sending the messages it previously received. With the approach, Flodgatt will never drop messages. However, the risk to this approach is that, by never dropping messages, Flodgatt does not have any way to reduce the amount of work it needs to do when under heavy load – it delays the work slightly, but doesn't reduce it. What this means is that it would be *theoretically* possible for Flodgatt to fall increasingly behind, if it is continuously receiving more messages than it can process. Due to how quickly Flodgatt can process messages, though, I suspect this would only come up if an admin were running Flodgatt in a *significantly* resource constrained environment, but I wanted to mention it for the sake of completeness. This commit also adds a new /status/backpressure endpoint that displays the current length of the Redis input buffer (which should typically be low or 0). Like the other /status endpoints, this endpoint is only enabled when Flodgatt is compiled with the `stub_status` feature.
2020-04-27 22:03:05 +02:00
// fn redis_msg_to_event_string(msg: RedisMsg) -> String {
// msg.event_txt.to_string()
// }
Improve handling of large Redis input (#143) * Implement faster buffered input This commit implements a modified ring buffer for input from Redis. Specifically, Flodgatt now limits the amount of data it fetches from Redis in one syscall to 8 KiB (two pages on most systems). Flodgatt will process all complete messages it receives from Redis and then re-use the same buffer for the next time it retrieves data. If Flodgatt received a partial message, it will copy the partial message to the beginning of the buffer before its next read. This change has little effect on Flodgatt under light load (because it was rare for Redis to have more than 8 KiB of messages available at any one time). However, my hope is that this will significantly reduce memory use on the largest instances. * Improve handling of backpresure This commit alters how Flodgatt behaves if it receives enough messages for a single client to fill that clients channel. (Because the clients regularly send their messages, should only occur if a single client receives a large number of messages nearly simultaneously; this is rare, but could occur, especially on large instances). Previously, Flodgatt would drop messages in the rare case when the client's channel was full. Now, Flodgatt will pause the current Redis poll and yield control back to the client streams, allowing the clients to empty their channels; Flodgatt will then resume polling Redis/sending the messages it previously received. With the approach, Flodgatt will never drop messages. However, the risk to this approach is that, by never dropping messages, Flodgatt does not have any way to reduce the amount of work it needs to do when under heavy load – it delays the work slightly, but doesn't reduce it. What this means is that it would be *theoretically* possible for Flodgatt to fall increasingly behind, if it is continuously receiving more messages than it can process. Due to how quickly Flodgatt can process messages, though, I suspect this would only come up if an admin were running Flodgatt in a *significantly* resource constrained environment, but I wanted to mention it for the sake of completeness. This commit also adds a new /status/backpressure endpoint that displays the current length of the Redis input buffer (which should typically be low or 0). Like the other /status endpoints, this endpoint is only enabled when Flodgatt is compiled with the `stub_status` feature.
2020-04-27 22:03:05 +02:00
// fn string_to_checked_event(event_txt: &String) -> EventKind {
// EventKind::TypeSafe(serde_json::from_str(event_txt).unwrap())
// }
2019-09-11 23:28:27 +02:00
fn criterion_benchmark(c: &mut Criterion) {
let input = ONE_MESSAGE_FOR_THE_USER_TIMLINE_FROM_REDIS;
2019-09-11 23:28:27 +02:00
let mut group = c.benchmark_group("Parse redis RESP array");
group.bench_function("parse redis input to RedisMsg", |b| {
b.iter(|| black_box(parse_long_redis_input(input)))
});
Improve handling of large Redis input (#143) * Implement faster buffered input This commit implements a modified ring buffer for input from Redis. Specifically, Flodgatt now limits the amount of data it fetches from Redis in one syscall to 8 KiB (two pages on most systems). Flodgatt will process all complete messages it receives from Redis and then re-use the same buffer for the next time it retrieves data. If Flodgatt received a partial message, it will copy the partial message to the beginning of the buffer before its next read. This change has little effect on Flodgatt under light load (because it was rare for Redis to have more than 8 KiB of messages available at any one time). However, my hope is that this will significantly reduce memory use on the largest instances. * Improve handling of backpresure This commit alters how Flodgatt behaves if it receives enough messages for a single client to fill that clients channel. (Because the clients regularly send their messages, should only occur if a single client receives a large number of messages nearly simultaneously; this is rare, but could occur, especially on large instances). Previously, Flodgatt would drop messages in the rare case when the client's channel was full. Now, Flodgatt will pause the current Redis poll and yield control back to the client streams, allowing the clients to empty their channels; Flodgatt will then resume polling Redis/sending the messages it previously received. With the approach, Flodgatt will never drop messages. However, the risk to this approach is that, by never dropping messages, Flodgatt does not have any way to reduce the amount of work it needs to do when under heavy load – it delays the work slightly, but doesn't reduce it. What this means is that it would be *theoretically* possible for Flodgatt to fall increasingly behind, if it is continuously receiving more messages than it can process. Due to how quickly Flodgatt can process messages, though, I suspect this would only come up if an admin were running Flodgatt in a *significantly* resource constrained environment, but I wanted to mention it for the sake of completeness. This commit also adds a new /status/backpressure endpoint that displays the current length of the Redis input buffer (which should typically be low or 0). Like the other /status endpoints, this endpoint is only enabled when Flodgatt is compiled with the `stub_status` feature.
2020-04-27 22:03:05 +02:00
// let msg = parse_long_redis_input(input);
// group.bench_function("parse RedisMsg to Timeline", |b| {
// b.iter(|| black_box(parse_to_timeline(msg.clone())))
// });
Improve handling of large Redis input (#143) * Implement faster buffered input This commit implements a modified ring buffer for input from Redis. Specifically, Flodgatt now limits the amount of data it fetches from Redis in one syscall to 8 KiB (two pages on most systems). Flodgatt will process all complete messages it receives from Redis and then re-use the same buffer for the next time it retrieves data. If Flodgatt received a partial message, it will copy the partial message to the beginning of the buffer before its next read. This change has little effect on Flodgatt under light load (because it was rare for Redis to have more than 8 KiB of messages available at any one time). However, my hope is that this will significantly reduce memory use on the largest instances. * Improve handling of backpresure This commit alters how Flodgatt behaves if it receives enough messages for a single client to fill that clients channel. (Because the clients regularly send their messages, should only occur if a single client receives a large number of messages nearly simultaneously; this is rare, but could occur, especially on large instances). Previously, Flodgatt would drop messages in the rare case when the client's channel was full. Now, Flodgatt will pause the current Redis poll and yield control back to the client streams, allowing the clients to empty their channels; Flodgatt will then resume polling Redis/sending the messages it previously received. With the approach, Flodgatt will never drop messages. However, the risk to this approach is that, by never dropping messages, Flodgatt does not have any way to reduce the amount of work it needs to do when under heavy load – it delays the work slightly, but doesn't reduce it. What this means is that it would be *theoretically* possible for Flodgatt to fall increasingly behind, if it is continuously receiving more messages than it can process. Due to how quickly Flodgatt can process messages, though, I suspect this would only come up if an admin were running Flodgatt in a *significantly* resource constrained environment, but I wanted to mention it for the sake of completeness. This commit also adds a new /status/backpressure endpoint that displays the current length of the Redis input buffer (which should typically be low or 0). Like the other /status endpoints, this endpoint is only enabled when Flodgatt is compiled with the `stub_status` feature.
2020-04-27 22:03:05 +02:00
// group.bench_function("parse RedisMsg -> DynamicEvent", |b| {
// b.iter(|| black_box(parse_to_dyn_event(msg.clone())))
// });
Improve handling of large Redis input (#143) * Implement faster buffered input This commit implements a modified ring buffer for input from Redis. Specifically, Flodgatt now limits the amount of data it fetches from Redis in one syscall to 8 KiB (two pages on most systems). Flodgatt will process all complete messages it receives from Redis and then re-use the same buffer for the next time it retrieves data. If Flodgatt received a partial message, it will copy the partial message to the beginning of the buffer before its next read. This change has little effect on Flodgatt under light load (because it was rare for Redis to have more than 8 KiB of messages available at any one time). However, my hope is that this will significantly reduce memory use on the largest instances. * Improve handling of backpresure This commit alters how Flodgatt behaves if it receives enough messages for a single client to fill that clients channel. (Because the clients regularly send their messages, should only occur if a single client receives a large number of messages nearly simultaneously; this is rare, but could occur, especially on large instances). Previously, Flodgatt would drop messages in the rare case when the client's channel was full. Now, Flodgatt will pause the current Redis poll and yield control back to the client streams, allowing the clients to empty their channels; Flodgatt will then resume polling Redis/sending the messages it previously received. With the approach, Flodgatt will never drop messages. However, the risk to this approach is that, by never dropping messages, Flodgatt does not have any way to reduce the amount of work it needs to do when under heavy load – it delays the work slightly, but doesn't reduce it. What this means is that it would be *theoretically* possible for Flodgatt to fall increasingly behind, if it is continuously receiving more messages than it can process. Due to how quickly Flodgatt can process messages, though, I suspect this would only come up if an admin were running Flodgatt in a *significantly* resource constrained environment, but I wanted to mention it for the sake of completeness. This commit also adds a new /status/backpressure endpoint that displays the current length of the Redis input buffer (which should typically be low or 0). Like the other /status endpoints, this endpoint is only enabled when Flodgatt is compiled with the `stub_status` feature.
2020-04-27 22:03:05 +02:00
// group.bench_function("parse RedisMsg -> CheckedEvent", |b| {
// b.iter(|| black_box(parse_to_checked_event(msg.clone())))
// });
Improve handling of large Redis input (#143) * Implement faster buffered input This commit implements a modified ring buffer for input from Redis. Specifically, Flodgatt now limits the amount of data it fetches from Redis in one syscall to 8 KiB (two pages on most systems). Flodgatt will process all complete messages it receives from Redis and then re-use the same buffer for the next time it retrieves data. If Flodgatt received a partial message, it will copy the partial message to the beginning of the buffer before its next read. This change has little effect on Flodgatt under light load (because it was rare for Redis to have more than 8 KiB of messages available at any one time). However, my hope is that this will significantly reduce memory use on the largest instances. * Improve handling of backpresure This commit alters how Flodgatt behaves if it receives enough messages for a single client to fill that clients channel. (Because the clients regularly send their messages, should only occur if a single client receives a large number of messages nearly simultaneously; this is rare, but could occur, especially on large instances). Previously, Flodgatt would drop messages in the rare case when the client's channel was full. Now, Flodgatt will pause the current Redis poll and yield control back to the client streams, allowing the clients to empty their channels; Flodgatt will then resume polling Redis/sending the messages it previously received. With the approach, Flodgatt will never drop messages. However, the risk to this approach is that, by never dropping messages, Flodgatt does not have any way to reduce the amount of work it needs to do when under heavy load – it delays the work slightly, but doesn't reduce it. What this means is that it would be *theoretically* possible for Flodgatt to fall increasingly behind, if it is continuously receiving more messages than it can process. Due to how quickly Flodgatt can process messages, though, I suspect this would only come up if an admin were running Flodgatt in a *significantly* resource constrained environment, but I wanted to mention it for the sake of completeness. This commit also adds a new /status/backpressure endpoint that displays the current length of the Redis input buffer (which should typically be low or 0). Like the other /status endpoints, this endpoint is only enabled when Flodgatt is compiled with the `stub_status` feature.
2020-04-27 22:03:05 +02:00
// group.bench_function("parse RedisMsg -> String -> CheckedEvent", |b| {
// b.iter(|| {
// let txt = black_box(redis_msg_to_event_string(msg.clone()));
// black_box(string_to_checked_event(&txt));
// })
// });
2019-09-11 23:28:27 +02:00
}
criterion_group!(benches, criterion_benchmark);
criterion_main!(benches);
const ONE_MESSAGE_FOR_THE_USER_TIMLINE_FROM_REDIS: &str = "*3\r\n$7\r\nmessage\r\n$10\r\ntimeline:1\r\n$3790\r\n{\"event\":\"update\",\"payload\":{\"id\":\"102775370117886890\",\"created_at\":\"2019-09-11T18:42:19.000Z\",\"in_reply_to_id\":null,\"in_reply_to_account_id\":null,\"sensitive\":false,\"spoiler_text\":\"\",\"visibility\":\"unlisted\",\"language\":\"en\",\"uri\":\"https://mastodon.host/users/federationbot/statuses/102775346916917099\",\"url\":\"https://mastodon.host/@federationbot/102775346916917099\",\"replies_count\":0,\"reblogs_count\":0,\"favourites_count\":0,\"favourited\":false,\"reblogged\":false,\"muted\":false,\"content\":\"<p>Trending tags:<br><a href=\\\"https://mastodon.host/tags/neverforget\\\" class=\\\"mention hashtag\\\" rel=\\\"nofollow noopener\\\" target=\\\"_blank\\\">#<span>neverforget</span></a><br><a href=\\\"https://mastodon.host/tags/4styles\\\" class=\\\"mention hashtag\\\" rel=\\\"nofollow noopener\\\" target=\\\"_blank\\\">#<span>4styles</span></a><br><a href=\\\"https://mastodon.host/tags/newpipe\\\" class=\\\"mention hashtag\\\" rel=\\\"nofollow noopener\\\" target=\\\"_blank\\\">#<span>newpipe</span></a><br><a href=\\\"https://mastodon.host/tags/uber\\\" class=\\\"mention hashtag\\\" rel=\\\"nofollow noopener\\\" target=\\\"_blank\\\">#<span>uber</span></a><br><a href=\\\"https://mastodon.host/tags/mercredifiction\\\" class=\\\"mention hashtag\\\" rel=\\\"nofollow noopener\\\" target=\\\"_blank\\\">#<span>mercredifiction</span></a></p>\",\"reblog\":null,\"account\":{\"id\":\"78\",\"username\":\"federationbot\",\"acct\":\"federationbot@mastodon.host\",\"display_name\":\"Federation Bot\",\"locked\":false,\"bot\":false,\"created_at\":\"2019-09-10T15:04:25.559Z\",\"note\":\"<p>Hello, I am mastodon.host official semi bot.</p><p>Follow me if you want to have some updates on the view of the fediverse from here ( I only post unlisted ). </p><p>I also randomly boost one of my followers toot every hour !</p><p>If you don\'t feel confortable with me following you, tell me: unfollow and I\'ll do it :)</p><p>If you want me to follow you, just tell me follow ! </p><p>If you want automatic follow for new users on your instance and you are an instance admin, contact me !</p><p>Other commands are private :)</p>\",\"url\":\"https://mastodon.host/@federationbot\",\"avatar\":\"https://instance.codesections.com/system/accounts/avatars/000/000/078/original/d9e2be5398629cf8.jpeg?1568127863\",\"avatar_static\":\"https://instance.codesections.com/system/accounts/avatars/000/000/078/original/d9e2be5398629cf8.jpeg?1568127863\",\"header\":\"https://instance.codesections.com/headers/original/missing.png\",\"header_static\":\"https://instance.codesections.com/headers/original/missing.png\",\"followers_count\":16636,\"following_count\":179532,\"statuses_count\":50554,\"emojis\":[],\"fields\":[{\"name\":\"More stats\",\"value\":\"<a href=\\\"https://mastodon.host/stats.html\\\" rel=\\\"nofollow noopener\\\" target=\\\"_blank\\\"><span class=\\\"invisible\\\">https://</span><span class=\\\"\\\">mastodon.host/stats.html</span><span class=\\\"invisible\\\"></span></a>\",\"verified_at\":null},{\"name\":\"More infos\",\"value\":\"<a href=\\\"https://mastodon.host/about/more\\\" rel=\\\"nofollow noopener\\\" target=\\\"_blank\\\"><span class=\\\"invisible\\\">https://</span><span class=\\\"\\\">mastodon.host/about/more</span><span class=\\\"invisible\\\"></span></a>\",\"verified_at\":null},{\"name\":\"Owner/Friend\",\"value\":\"<span class=\\\"h-card\\\"><a href=\\\"https://mastodon.host/@gled\\\" class=\\\"u-url mention\\\" rel=\\\"nofollow noopener\\\" target=\\\"_blank\\\">@<span>gled</span></a></span>\",\"verified_at\":null}]},\"media_attachments\":[],\"mentions\":[],\"tags\":[{\"name\":\"4styles\",\"url\":\"https://instance.codesections.com/tags/4styles\"},{\"name\":\"neverforget\",\"url\":\"https://instance.codesections.com/tags/neverforget\"},{\"name\":\"mercredifiction\",\"url\":\"https://instance.codesections.com/tags/mercredifiction\"},{\"name\":\"uber\",\"url\":\"https://instance.codesections.com/tags/ube