centrifuge
centrifuge copied to clipboard
centrifugal
Migrate to github.com/go-redis/redis
Codecov Report
Merging #235 (244daa6) into master (f9fe49d) will increase coverage by
1.20%. The diff coverage is87.87%.
@@ Coverage Diff @@
## master #235 +/- ##
==========================================
+ Coverage 83.06% 84.27% +1.20%
==========================================
Files 36 37 +1
Lines 8207 8017 -190
==========================================
- Hits 6817 6756 -61
+ Misses 1055 948 -107
+ Partials 335 313 -22
| Impacted Files | Coverage Δ | |
|---|---|---|
| redis_shard.go | 78.04% <77.21%> (+11.68%) |
:arrow_up: |
| broker_redis.go | 74.22% <93.10%> (+2.03%) |
:arrow_up: |
| internal/util/unsafe.go | 100.00% <100.00%> (ø) |
|
| node.go | 92.94% <100.00%> (+0.19%) |
:arrow_up: |
| presence_redis.go | 75.73% <100.00%> (+0.14%) |
:arrow_up: |
| client.go | 83.50% <0.00%> (+0.08%) |
:arrow_up: |
| hub.go | 84.92% <0.00%> (+0.26%) |
:arrow_up: |
| ... and 2 more |
:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more
![codecov[bot] avatar](https://avatars.githubusercontent.com/in/254?v=4 "Published by a pub.dev verified publisher"){kind=small}
Oh, wow, that's the hard work! And seems great overall 👍 Removing dependencies and the code that manages various connection modes is awesome.
Probably let's collaborate next time before introducing such a big changes? Just to make sure we are on the same page about things.
In general, the issue was about researching things, we need to prove the migration to go-redis/redis makes sense. So some additional understanding required here to proceed with changes.
I have not looked at code closely yet – will do at some point, for now some top-level things:
- What was the reason of bump to
go1.18in go.mod? We should not depend on features ofgo1.18for now. - The difference in timeout behavior. In the current implementation we have both read/write connection timeouts set, probably you can describe the difference in timeout behavior with new implementation. I see lot's of usages of
context.Background()- does the operations have some underlying timeouts anyway? So some description regarding working with timeouts in implementation withgo-rediswould be awesome to have in PR. - Run existing benchmarks - we have several Redis benchmarks and they should perform better to justify the migration. Could you run them with https://pkg.go.dev/golang.org/x/perf/cmd/benchstat ? Super curious to look at results!
- Performance in Redis Cluster – this is not covered by current benchmarks but we need to understand the performance difference in Cluster. Possibly we need some new benchmarks that measure how Engine performs in Cluster. Should not be worse than Redigo +
mna/redisceither. - Have you tested failover with Sentinel? I think existing tests do not cover it - only connect to current master. Probably we can automate this somehow, but at least manual testing would be awesome to have some confidence it works in general. Otherwise,
redis-goshould be stable enough I believe to rely on it. - Though I think we still need to look through
go-redisopen issues list and try to find out whether some of bugs apply to Engine implementation.
Redis Cluster can be run with sth like this locally (but I suppose to compare the performance it's better to try excluding Docker from setup eventually):
version: "3.8"
services:
cluster:
image: redis:7.0.0-alpine
entrypoint:
- /bin/sh
- -c
- |
redis-server --port 7001 --save "" --appendonly no --cluster-enabled yes --cluster-config-file 7001.conf &
redis-server --port 7002 --save "" --appendonly no --cluster-enabled yes --cluster-config-file 7002.conf &
redis-server --port 7003 --save "" --appendonly no --cluster-enabled yes --cluster-config-file 7003.conf &
while ! redis-cli --cluster create 127.0.0.1:7001 127.0.0.1:7002 127.0.0.1:7003 --cluster-yes; do sleep 1; done
wait
ports:
- "7001:7001"
- "7002:7002"
- "7003:7003"
Hi, thanks for your comment!
I'm sorry for submitting an important change without making any notes. In fact, I really wanted to explain something, but my English is so poor that it was hard to express it clearly. So I chose to submit the code first.
This PR is only a proof of concept and there is still a lot of work before it is finished. I wanted to send it first and hear your opinion, before me going too far down the wrong path.
Some to-do before it's ready for review:
- [ ] Test agasin sentinel and cluster
- [ ] Use real context (instead of context.Background)
- [ ] Verify timeout settings
- [ ] Add new cluster benchmarks
- [ ] Conclude some benchmark numbers
With the current implementation, I ran some benchmark locally, but most of the results are a bit worse than redigo (~10% slower). I'm trying to find out why and do some optimization and hopefully we can get a better result.
Hmm, interesting – I expected a better performance than with Redigo. Actually I did some measurements before opening that issue and results were quite promising. That was the biggest motivation to try the migration. So I'd say looking at benchmarks is the first step here to decide whether it's worth it or not. Since otherwise current implementation is pretty stable I believe to change it for sth new.
Here are some benchmark results ran in my local machine:
goos: darwin goarch: amd64 cpu: Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
Most numbers are fine, except for RedisAddPresence_1Ch, RedisPresence_1Ch and RedisPresence_ManyCh, especially RedisAddPresence_1Ch became incredibly slower. No idea why, still investigating.
name old time/op new time/op delta
RedisExtractPushData-12 54.9ns ± 0% 38.3ns ± 5% -30.12% (p=0.008 n=5+5)
name old alloc/op new alloc/op delta
RedisExtractPushData-12 16.0B ± 0% 0.0B -100.00% (p=0.008 n=5+5)
name old allocs/op new allocs/op delta
RedisExtractPushData-12 1.00 ± 0% 0.00 -100.00% (p=0.008 n=5+5)
// 100 nodes always stuck, it's wired, just skipped it for now
name old time/op new time/op delta
RedisSurvey/1_node_512B-12 1.54µs ±11% 1.30µs ± 6% -15.77% (p=0.008 n=5+5)
RedisSurvey/2_nodes_512B-12 49.7µs ± 2% 29.4µs ± 4% -40.94% (p=0.008 n=5+5)
RedisSurvey/3_nodes_512B-12 66.0µs ±15% 45.0µs ± 4% -31.85% (p=0.008 n=5+5)
RedisSurvey/4_nodes_512B-12 86.6µs ±11% 61.5µs ± 5% -28.94% (p=0.008 n=5+5)
RedisSurvey/5_nodes_512B-12 103µs ± 5% 82µs ± 3% -20.66% (p=0.008 n=5+5)
RedisSurvey/10_nodes_512B-12 246µs ±16% 200µs ±14% -18.98% (p=0.032 n=5+5)
RedisSurvey/1_node_4KB-12 1.86µs ± 8% 1.68µs ± 4% -9.29% (p=0.016 n=5+5)
RedisSurvey/2_nodes_4KB-12 77.3µs ±13% 57.4µs ±11% -25.71% (p=0.008 n=5+5)
RedisSurvey/3_nodes_4KB-12 180µs ±31% 173µs ±17% ~ (p=1.000 n=5+5)
RedisSurvey/4_nodes_4KB-12 297µs ±55% 256µs ±29% ~ (p=0.421 n=5+5)
RedisSurvey/5_nodes_4KB-12 393µs ±42% 328µs ±12% ~ (p=0.151 n=5+5)
RedisSurvey/10_nodes_4KB-12 835µs ± 8% 473µs ±42% -43.36% (p=0.008 n=5+5)
name old time/op new time/op delta
RedisConsistentIndex-12 55.7ns ± 1% 51.5ns ± 0% -7.53% (p=0.016 n=5+4)
name old time/op new time/op delta
RedisIndex-12 43.3ns ± 1% 39.5ns ± 2% -8.70% (p=0.008 n=5+5)
name old time/op new time/op delta
RedisPublish_1Ch-12 2.67µs ±12% 2.81µs ± 2% ~ (p=0.222 n=5+5)
name old time/op new time/op delta
RedisPublish_ManyCh-12 2.42µs ± 1% 2.76µs ± 1% +14.15% (p=0.008 n=5+5)
name old time/op new time/op delta
RedisPublish_History_1Ch/lists-12 27.7µs ± 2% 27.2µs ± 1% -1.97% (p=0.008 n=5+5)
RedisPublish_History_1Ch/streams-12 34.1µs ± 1% 33.8µs ± 1% ~ (p=0.056 n=5+5)
name old time/op new time/op delta
RedisPub_History_ManyCh/lists-12 28.4µs ± 2% 28.0µs ± 5% ~ (p=0.310 n=5+5)
RedisPub_History_ManyCh/streams-12 28.7µs ± 2% 28.6µs ± 5% ~ (p=0.690 n=5+5)
name old time/op new time/op delta
RedisSubscribe-12 2.20µs ± 7% 2.41µs ± 9% ~ (p=0.056 n=5+5)
name old time/op new time/op delta
RedisHistory_1Ch/lists-12 23.0µs ± 8% 20.7µs ± 1% -10.09% (p=0.008 n=5+5)
RedisHistory_1Ch/streams-12 37.6µs ± 1% 36.8µs ± 0% -2.17% (p=0.008 n=5+5)
name old time/op new time/op delta
RedisRecover_1Ch/lists-12 388µs ± 3% 386µs ± 3% ~ (p=0.421 n=5+5)
RedisRecover_1Ch/streams-12 48.8µs ± 2% 47.3µs ± 1% -2.98% (p=0.008 n=5+5)
name old time/op new time/op delta
RedisHistoryIteration/lists-12 591ms ± 2% 524ms ± 2% -11.40% (p=0.008 n=5+5)
RedisHistoryIteration/streams-12 57.1ms ± 2% 52.5ms ± 4% -8.13% (p=0.008 n=5+5)
// RedisAddPresence is awfully slower, still investigaing
name old time/op new time/op delta
RedisAddPresence_1Ch-12 15.8µs ± 0% 4680.2µs ± 1% +29449.22% (p=0.016 n=4+5)
name old time/op new time/op delta
RedisPresence_1Ch-12 13.8µs ± 1% 18.4µs ± 6% +33.55% (p=0.008 n=5+5)
name old time/op new time/op delta
RedisPresence_ManyCh-12 11.7µs ± 0% 15.7µs ± 4% +33.82% (p=0.008 n=5+5)
After fixed the redis.Nil problem, here is the updated benchmark results about RedisPresence:
name old time/op new time/op delta
RedisAddPresence_1Ch-12 16.2µs ± 2% 16.6µs ±12% ~ (p=0.690 n=5+5)
name old time/op new time/op delta
RedisPresence_1Ch-12 13.9µs ± 1% 12.9µs ± 1% -7.44% (p=0.008 n=5+5)
name old time/op new time/op delta
RedisPresence_ManyCh-12 11.9µs ± 1% 11.0µs ± 1% -7.44% (p=0.008 n=5+5)
Hi, this is an Interesting work, I just started my implementation using https://github.com/rueian/rueidis which supports sharded pubsub feature of redis 7 for my project.
Think bench results look promising, @j178 actually we are interested not only in latency improvements but also in the number of allocations since that directly affects CPU usage. I suppose that in go-redis allocs are consistently less.
As the next step I think comparing the performance of go-redis with Redigo in Redis Cluster case would be great.
@tony-pang I used rueidis to implement Redis engine for Centrifugo PRO. Unfortunately we can't replace Centrifuge library builtin Redis engine with it since we need to maintain compatibility with Redis instances which work over RESP2. Probably at some point in the future it will be possible to make rueidis-based implementation default in the core of Centrifuge too. But for now go-redis seems the only reasonable alternative to redigo for builtin Engine implementation.
As the next step I think comparing the performance of go-redis with Redigo would be great.
Sorry, I mean performance comparison in Redis Cluster case, updated the comment :)
Sorry, I mean performance comparison in Redis Cluster case, updated the comment :)
Will do it 😄 But I want to figure out why the BenchmarkRedisSurvey hang when running with 100 nodes on my machine first. Could you run this benchmark successfully? Maybe there is something wrong with my local environment.
But I want to figure out why the BenchmarkRedisSurvey hang when running with 100 nodes on my machine first. Could you run this benchmark successfully? Maybe there is something wrong with my local environment.
Tried, works fine for both redigo and go-redis implementations. Is it hanging for both redigo and go-redis on your machine?
Is it hanging for both redigo and go-redis on your machine?
That's right. Seems like there is something wrong with my setup.
@FZambia Hi, have you tried run the benchamrk multiple times? like go test -tags integration -bench=BenchmarkRedisSurvey/100_nodes -run=NONE -count=5. In my case, it always hang in the second or third run.
have you tried run the benchamrk multiple times? like go test -tags integration -bench=BenchmarkRedisSurvey/100_nodes -run=NONE -count=5. In my case, it always hang in the second or third run.
Yep - reproduced! Could you try https://github.com/centrifugal/centrifuge/pull/237 - this fixed the issue for me.
Latest benchstat results against the master branch(including cluster mode stats):
name old time/op new time/op delta
RedisExtractPushData-16 71.9ns ± 1% 32.0ns ± 1% -55.50% (p=0.008 n=5+5)
RedisSurvey/1_node_512B-16 1.57µs ± 2% 1.60µs ± 1% +1.93% (p=0.032 n=5+5)
RedisSurvey/2_nodes_512B-16 276µs ± 6% 165µs ± 3% -40.45% (p=0.008 n=5+5)
RedisSurvey/3_nodes_512B-16 308µs ± 0% 177µs ± 3% -42.54% (p=0.008 n=5+5)
RedisSurvey/4_nodes_512B-16 331µs ± 3% 190µs ± 3% -42.50% (p=0.008 n=5+5)
RedisSurvey/5_nodes_512B-16 349µs ± 2% 204µs ± 1% -41.48% (p=0.008 n=5+5)
RedisSurvey/10_nodes_512B-16 433µs ± 2% 273µs ± 1% -36.97% (p=0.008 n=5+5)
RedisSurvey/100_nodes_512B-16 2.54ms ± 5% 2.06ms ± 2% -18.77% (p=0.008 n=5+5)
RedisSurvey/1_node_4096B-16 1.53µs ± 1% 1.52µs ± 1% -0.88% (p=0.024 n=5+5)
RedisSurvey/2_nodes_4096B-16 312µs ± 3% 188µs ± 2% -39.74% (p=0.008 n=5+5)
RedisSurvey/3_nodes_4096B-16 371µs ± 2% 231µs ± 2% -37.70% (p=0.008 n=5+5)
RedisSurvey/4_nodes_4096B-16 423µs ± 3% 275µs ± 1% -34.86% (p=0.008 n=5+5)
RedisSurvey/5_nodes_4096B-16 478µs ± 2% 333µs ± 1% -30.22% (p=0.008 n=5+5)
RedisSurvey/10_nodes_4096B-16 758µs ± 4% 681µs ± 2% -10.08% (p=0.008 n=5+5)
RedisSurvey/100_nodes_4096B-16 10.3ms ± 3% 9.3ms ± 2% -9.60% (p=0.008 n=5+5)
RedisSurvey/1_node_512B_cluster-16 1.45µs ± 1% 1.44µs ± 1% -0.88% (p=0.032 n=5+5)
RedisSurvey/2_nodes_512B_cluster-16 285µs ± 7% 162µs ± 1% -43.16% (p=0.008 n=5+5)
RedisSurvey/3_nodes_512B_cluster-16 310µs ± 3% 177µs ± 1% -42.98% (p=0.008 n=5+5)
RedisSurvey/4_nodes_512B_cluster-16 333µs ± 2% 193µs ± 1% -41.95% (p=0.008 n=5+5)
RedisSurvey/5_nodes_512B_cluster-16 354µs ± 2% 206µs ± 1% -41.91% (p=0.008 n=5+5)
RedisSurvey/10_nodes_512B_cluster-16 442µs ± 1% 277µs ± 2% -37.45% (p=0.008 n=5+5)
RedisSurvey/100_nodes_512B_cluster-16 2.51ms ± 2% 2.10ms ± 5% -16.23% (p=0.008 n=5+5)
RedisSurvey/1_node_4096B_cluster-16 1.44µs ± 1% 1.45µs ± 2% ~ (p=0.889 n=5+5)
RedisSurvey/2_nodes_4096B_cluster-16 307µs ± 2% 191µs ± 1% -37.85% (p=0.008 n=5+5)
RedisSurvey/3_nodes_4096B_cluster-16 364µs ± 2% 231µs ± 1% -36.53% (p=0.008 n=5+5)
RedisSurvey/4_nodes_4096B_cluster-16 416µs ± 1% 277µs ± 1% -33.37% (p=0.008 n=5+5)
RedisSurvey/5_nodes_4096B_cluster-16 465µs ± 0% 332µs ± 2% -28.57% (p=0.008 n=5+5)
RedisSurvey/10_nodes_4096B_cluster-16 751µs ± 3% 688µs ± 1% -8.45% (p=0.008 n=5+5)
RedisSurvey/100_nodes_4096B_cluster-16 10.4ms ± 3% 9.4ms ± 2% -9.55% (p=0.008 n=5+5)
RedisConsistentIndex-16 54.9ns ± 0% 52.6ns ± 0% -4.25% (p=0.008 n=5+5)
RedisIndex-16 44.5ns ± 2% 39.4ns ± 1% -11.37% (p=0.008 n=5+5)
RedisPublish_1Ch/lists-16 9.91µs ± 1% 9.70µs ± 2% -2.14% (p=0.032 n=5+5)
RedisPublish_1Ch/streams-16 10.0µs ± 3% 9.4µs ± 1% -5.90% (p=0.008 n=5+5)
RedisPublish_1Ch/lists_cluster-16 9.00µs ± 2% 10.91µs ± 2% +21.19% (p=0.008 n=5+5)
RedisPublish_1Ch/streams_cluster-16 8.87µs ± 0% 10.93µs ± 4% +23.26% (p=0.016 n=5+4)
RedisPublish_ManyCh/lists-16 9.80µs ± 2% 9.48µs ± 1% -3.27% (p=0.008 n=5+5)
RedisPublish_ManyCh/streams-16 9.87µs ± 1% 9.49µs ± 2% -3.81% (p=0.008 n=5+5)
RedisPublish_ManyCh/lists_cluster-16 9.10µs ± 1% 10.07µs ± 3% +10.66% (p=0.008 n=5+5)
RedisPublish_ManyCh/streams_cluster-16 9.03µs ± 1% 9.99µs ± 1% +10.62% (p=0.008 n=5+5)
RedisPublish_History_1Ch/lists-16 43.2µs ± 1% 33.4µs ± 2% -22.69% (p=0.008 n=5+5)
RedisPublish_History_1Ch/streams-16 48.5µs ± 1% 38.7µs ± 2% -20.30% (p=0.008 n=5+5)
RedisPublish_History_1Ch/lists_cluster-16 45.6µs ± 2% 45.7µs ± 1% ~ (p=0.841 n=5+5)
RedisPublish_History_1Ch/streams_cluster-16 50.7µs ± 2% 51.2µs ± 1% ~ (p=0.548 n=5+5)
RedisPub_History_ManyCh/lists-16 42.9µs ± 1% 33.5µs ± 2% -22.01% (p=0.008 n=5+5)
RedisPub_History_ManyCh/streams-16 43.0µs ± 2% 33.3µs ± 2% -22.62% (p=0.008 n=5+5)
RedisPub_History_ManyCh/lists_cluster-16 23.1µs ± 4% 25.5µs ± 5% +10.43% (p=0.008 n=5+5)
RedisPub_History_ManyCh/streams_cluster-16 23.0µs ± 4% 25.2µs ± 4% +9.43% (p=0.008 n=5+5)
RedisSubscribe/non_cluster-16 6.33µs ± 2% 6.39µs ± 1% ~ (p=0.151 n=5+5)
RedisSubscribe/with_cluster-16 6.34µs ± 2% 6.38µs ± 2% ~ (p=0.841 n=5+5)
RedisHistory_1Ch/lists-16 38.1µs ± 1% 28.5µs ± 0% -25.26% (p=0.008 n=5+5)
RedisHistory_1Ch/streams-16 49.3µs ± 1% 39.8µs ± 1% -19.16% (p=0.008 n=5+5)
RedisHistory_1Ch/lists_cluster-16 37.3µs ± 1% 39.5µs ± 3% +5.91% (p=0.008 n=5+5)
RedisHistory_1Ch/streams_cluster-16 49.5µs ± 1% 51.2µs ± 2% +3.46% (p=0.008 n=5+5)
RedisRecover_1Ch/lists-16 675µs ± 1% 661µs ± 2% -2.03% (p=0.032 n=5+5)
RedisRecover_1Ch/streams-16 57.9µs ± 1% 49.4µs ± 1% -14.60% (p=0.008 n=5+5)
RedisRecover_1Ch/lists_cluster-16 399µs ± 2% 399µs ± 3% ~ (p=1.000 n=5+5)
RedisRecover_1Ch/streams_cluster-16 56.2µs ± 5% 58.8µs ± 2% +4.78% (p=0.032 n=5+5)
RedisHistoryIteration/lists-16 1.03s ± 1% 0.93s ± 1% -10.00% (p=0.008 n=5+5)
RedisHistoryIteration/streams-16 163ms ± 1% 105ms ± 2% -35.48% (p=0.008 n=5+5)
RedisHistoryIteration/lists_cluster-16 925ms ± 3% 914ms ± 1% ~ (p=0.421 n=5+5)
RedisHistoryIteration/streams_cluster-16 101ms ± 1% 101ms ± 1% ~ (p=0.310 n=5+5)
RedisAddPresence_1Ch/lists-16 32.7µs ± 1% 23.9µs ± 1% -26.76% (p=0.008 n=5+5)
RedisAddPresence_1Ch/streams-16 32.5µs ± 1% 24.2µs ± 3% -25.61% (p=0.008 n=5+5)
RedisAddPresence_1Ch/lists_cluster-16 30.1µs ± 2% 31.3µs ± 1% +4.08% (p=0.008 n=5+5)
RedisAddPresence_1Ch/streams_cluster-16 31.8µs ± 4% 31.2µs ± 1% ~ (p=0.310 n=5+5)
RedisPresence_1Ch/lists-16 32.4µs ± 1% 22.4µs ± 2% -31.01% (p=0.008 n=5+5)
RedisPresence_1Ch/streams-16 31.8µs ± 2% 22.4µs ± 2% -29.66% (p=0.008 n=5+5)
RedisPresence_1Ch/lists_cluster-16 29.9µs ± 2% 30.9µs ± 2% +3.29% (p=0.032 n=5+5)
RedisPresence_1Ch/streams_cluster-16 30.4µs ± 3% 30.7µs ± 3% ~ (p=0.421 n=5+5)
RedisPresence_ManyCh/lists-16 31.6µs ±11% 21.4µs ± 2% -32.20% (p=0.008 n=5+5)
RedisPresence_ManyCh/streams-16 31.3µs ±11% 21.4µs ± 2% -31.57% (p=0.008 n=5+5)
RedisPresence_ManyCh/lists_cluster-16 17.0µs ± 4% 17.0µs ± 2% ~ (p=0.690 n=5+5)
RedisPresence_ManyCh/streams_cluster-16 16.7µs ± 4% 16.9µs ± 3% ~ (p=0.690 n=5+5)
name old alloc/op new alloc/op delta
RedisExtractPushData-16 16.0B ± 0% 0.0B -100.00% (p=0.008 n=5+5)
RedisSurvey/1_node_512B-16 1.14kB ± 0% 1.14kB ± 0% ~ (all equal)
RedisSurvey/2_nodes_512B-16 7.52kB ± 0% 7.14kB ± 0% -5.12% (p=0.008 n=5+5)
RedisSurvey/3_nodes_512B-16 12.6kB ± 0% 12.0kB ± 0% -5.28% (p=0.008 n=5+5)
RedisSurvey/4_nodes_512B-16 17.7kB ± 0% 16.8kB ± 0% -5.42% (p=0.008 n=5+5)
RedisSurvey/5_nodes_512B-16 22.8kB ± 0% 21.6kB ± 0% -5.45% (p=0.008 n=5+5)
RedisSurvey/10_nodes_512B-16 49.2kB ± 0% 46.5kB ± 0% -5.47% (p=0.008 n=5+5)
RedisSurvey/100_nodes_512B-16 544kB ± 0% 505kB ± 0% -7.05% (p=0.008 n=5+5)
RedisSurvey/1_node_4096B-16 1.15kB ± 0% 1.15kB ± 0% ~ (p=0.643 n=5+5)
RedisSurvey/2_nodes_4096B-16 28.8kB ± 0% 28.1kB ± 0% -2.68% (p=0.008 n=5+5)
RedisSurvey/3_nodes_4096B-16 54.7kB ± 1% 53.7kB ± 1% -1.91% (p=0.008 n=5+5)
RedisSurvey/4_nodes_4096B-16 82.1kB ± 4% 80.0kB ± 1% -2.47% (p=0.008 n=5+5)
RedisSurvey/5_nodes_4096B-16 109kB ± 3% 107kB ± 2% ~ (p=0.151 n=5+5)
RedisSurvey/10_nodes_4096B-16 243kB ± 2% 240kB ± 1% ~ (p=0.222 n=5+5)
RedisSurvey/100_nodes_4096B-16 2.68MB ± 0% 2.66MB ± 3% ~ (p=0.730 n=4+5)
RedisSurvey/1_node_512B_cluster-16 1.15kB ± 0% 1.15kB ± 0% ~ (p=0.238 n=4+5)
RedisSurvey/2_nodes_512B_cluster-16 10.6kB ±10% 7.8kB ± 5% -25.95% (p=0.008 n=5+5)
RedisSurvey/3_nodes_512B_cluster-16 15.4kB ±15% 13.2kB ± 4% -14.00% (p=0.008 n=5+5)
RedisSurvey/4_nodes_512B_cluster-16 22.8kB ± 7% 19.4kB ± 7% -14.87% (p=0.008 n=5+5)
RedisSurvey/5_nodes_512B_cluster-16 27.0kB ±12% 25.1kB ±11% ~ (p=0.222 n=5+5)
RedisSurvey/10_nodes_512B_cluster-16 53.4kB ± 4% 50.3kB ± 2% -5.91% (p=0.008 n=5+5)
RedisSurvey/100_nodes_512B_cluster-16 562kB ± 3% 515kB ± 1% -8.35% (p=0.008 n=5+5)
RedisSurvey/1_node_4096B_cluster-16 1.16kB ± 1% 1.15kB ± 1% ~ (p=0.087 n=5+5)
RedisSurvey/2_nodes_4096B_cluster-16 33.1kB ± 6% 29.9kB ± 4% -9.79% (p=0.008 n=5+5)
RedisSurvey/3_nodes_4096B_cluster-16 57.9kB ± 3% 58.2kB ± 3% ~ (p=0.690 n=5+5)
RedisSurvey/4_nodes_4096B_cluster-16 86.5kB ± 6% 83.7kB ± 4% ~ (p=0.222 n=5+5)
RedisSurvey/5_nodes_4096B_cluster-16 113kB ± 2% 110kB ± 3% ~ (p=0.222 n=5+5)
RedisSurvey/10_nodes_4096B_cluster-16 246kB ± 2% 243kB ± 3% ~ (p=0.690 n=5+5)
RedisSurvey/100_nodes_4096B_cluster-16 2.85MB ± 3% 2.74MB ± 4% -3.80% (p=0.032 n=5+5)
RedisConsistentIndex-16 8.00B ± 0% 8.40B ± 7% ~ (p=0.444 n=5+5)
RedisIndex-16 8.00B ± 0% 8.00B ± 0% ~ (all equal)
RedisPublish_1Ch/lists-16 648B ±10% 670B ± 6% ~ (p=0.690 n=5+5)
RedisPublish_1Ch/streams-16 643B ±12% 689B ± 8% ~ (p=0.278 n=5+5)
RedisPublish_1Ch/lists_cluster-16 760B ±14% 730B ±18% ~ (p=0.548 n=5+5)
RedisPublish_1Ch/streams_cluster-16 735B ±14% 718B ±19% ~ (p=0.730 n=5+5)
RedisPublish_ManyCh/lists-16 697B ± 5% 663B ±14% ~ (p=0.222 n=5+5)
RedisPublish_ManyCh/streams-16 719B ±12% 692B ±10% ~ (p=0.690 n=5+5)
RedisPublish_ManyCh/lists_cluster-16 709B ± 6% 780B ±24% ~ (p=0.151 n=5+5)
RedisPublish_ManyCh/streams_cluster-16 667B ±10% 734B ±10% ~ (p=0.310 n=5+5)
RedisPublish_History_1Ch/lists-16 2.10kB ±13% 1.68kB ±16% -20.16% (p=0.016 n=5+5)
RedisPublish_History_1Ch/streams-16 2.03kB ±13% 1.75kB ±20% ~ (p=0.222 n=5+5)
RedisPublish_History_1Ch/lists_cluster-16 2.60kB ± 2% 1.73kB ±20% -33.61% (p=0.008 n=5+5)
RedisPublish_History_1Ch/streams_cluster-16 2.45kB ±23% 2.06kB ±26% ~ (p=0.222 n=5+5)
RedisPub_History_ManyCh/lists-16 2.27kB ±11% 1.74kB ± 9% -23.29% (p=0.008 n=5+5)
RedisPub_History_ManyCh/streams-16 2.02kB ±10% 1.85kB ± 9% ~ (p=0.095 n=5+5)
RedisPub_History_ManyCh/lists_cluster-16 1.84kB ± 8% 1.57kB ±20% -14.52% (p=0.032 n=5+5)
RedisPub_History_ManyCh/streams_cluster-16 1.91kB ± 3% 1.67kB ±17% ~ (p=0.095 n=5+5)
RedisSubscribe/non_cluster-16 1.44kB ± 4% 1.20kB ± 4% -16.75% (p=0.008 n=5+5)
RedisSubscribe/with_cluster-16 1.53kB ± 5% 1.16kB ± 6% -24.23% (p=0.008 n=5+5)
RedisHistory_1Ch/lists-16 2.53kB ±17% 2.02kB ±17% -20.20% (p=0.016 n=5+5)
RedisHistory_1Ch/streams-16 3.52kB ± 5% 2.70kB ±12% -23.49% (p=0.008 n=5+5)
RedisHistory_1Ch/lists_cluster-16 2.80kB ± 9% 2.36kB ± 4% -15.54% (p=0.008 n=5+5)
RedisHistory_1Ch/streams_cluster-16 3.45kB ± 7% 3.12kB ±22% ~ (p=0.151 n=5+5)
RedisRecover_1Ch/lists-16 173kB ± 3% 142kB ± 3% -18.29% (p=0.008 n=5+5)
RedisRecover_1Ch/streams-16 3.82kB ±19% 3.21kB ±18% ~ (p=0.222 n=5+5)
RedisRecover_1Ch/lists_cluster-16 167kB ± 2% 140kB ± 0% -16.51% (p=0.016 n=5+4)
RedisRecover_1Ch/streams_cluster-16 4.22kB ± 9% 3.51kB ±15% -16.89% (p=0.008 n=5+5)
RedisHistoryIteration/lists-16 184MB ± 4% 148MB ± 2% -19.44% (p=0.008 n=5+5)
RedisHistoryIteration/streams-16 5.41MB ±17% 4.66MB ±25% ~ (p=0.222 n=5+5)
RedisHistoryIteration/lists_cluster-16 178MB ± 4% 145MB ± 3% -18.65% (p=0.008 n=5+5)
RedisHistoryIteration/streams_cluster-16 5.25MB ±13% 4.68MB ±23% ~ (p=0.548 n=5+5)
RedisAddPresence_1Ch/lists-16 1.57kB ± 5% 1.14kB ±17% -27.45% (p=0.008 n=5+5)
RedisAddPresence_1Ch/streams-16 1.29kB ±17% 1.32kB ±17% ~ (p=0.548 n=5+5)
RedisAddPresence_1Ch/lists_cluster-16 1.80kB ±12% 1.46kB ±25% ~ (p=0.056 n=5+5)
RedisAddPresence_1Ch/streams_cluster-16 1.75kB ±15% 1.48kB ±20% -15.53% (p=0.032 n=5+5)
RedisPresence_1Ch/lists-16 1.97kB ±16% 1.49kB ±12% -24.24% (p=0.008 n=5+5)
RedisPresence_1Ch/streams-16 2.02kB ±12% 1.59kB ± 9% -21.28% (p=0.008 n=5+5)
RedisPresence_1Ch/lists_cluster-16 2.06kB ±14% 1.61kB ±15% -21.66% (p=0.032 n=5+5)
RedisPresence_1Ch/streams_cluster-16 2.22kB ± 5% 1.61kB ±27% -27.28% (p=0.008 n=5+5)
RedisPresence_ManyCh/lists-16 1.39kB ± 8% 1.24kB ±18% ~ (p=0.095 n=5+5)
RedisPresence_ManyCh/streams-16 1.45kB ±14% 1.19kB ±21% -17.79% (p=0.032 n=5+5)
RedisPresence_ManyCh/lists_cluster-16 1.49kB ± 8% 1.16kB ±11% -22.00% (p=0.008 n=5+5)
RedisPresence_ManyCh/streams_cluster-16 1.53kB ± 3% 1.17kB ±14% -23.43% (p=0.008 n=5+5)
name old allocs/op new allocs/op delta
RedisExtractPushData-16 1.00 ± 0% 0.00 -100.00% (p=0.008 n=5+5)
RedisSurvey/1_node_512B-16 15.0 ± 0% 15.0 ± 0% ~ (all equal)
RedisSurvey/2_nodes_512B-16 90.0 ± 0% 81.0 ± 0% -10.00% (p=0.008 n=5+5)
RedisSurvey/3_nodes_512B-16 140 ± 0% 124 ± 0% -11.43% (p=0.008 n=5+5)
RedisSurvey/4_nodes_512B-16 190 ± 0% 167 ± 0% -12.11% (p=0.008 n=5+5)
RedisSurvey/5_nodes_512B-16 240 ± 0% 210 ± 0% ~ (p=0.079 n=4+5)
RedisSurvey/10_nodes_512B-16 491 ± 0% 426 ± 0% -13.16% (p=0.000 n=4+5)
RedisSurvey/100_nodes_512B-16 5.63k ± 1% 4.66k ± 0% -17.19% (p=0.008 n=5+5)
RedisSurvey/1_node_4096B-16 15.0 ± 0% 15.0 ± 0% ~ (all equal)
RedisSurvey/2_nodes_4096B-16 96.8 ± 1% 82.8 ± 1% -14.46% (p=0.008 n=5+5)
RedisSurvey/3_nodes_4096B-16 147 ± 4% 127 ± 3% -13.99% (p=0.008 n=5+5)
RedisSurvey/4_nodes_4096B-16 214 ±16% 180 ± 3% -16.14% (p=0.008 n=5+5)
RedisSurvey/5_nodes_4096B-16 279 ±14% 243 ±12% ~ (p=0.056 n=5+5)
RedisSurvey/10_nodes_4096B-16 575 ± 7% 530 ± 7% ~ (p=0.190 n=5+5)
RedisSurvey/100_nodes_4096B-16 7.85k ± 1% 6.94k ±11% -11.62% (p=0.016 n=4+5)
RedisSurvey/1_node_512B_cluster-16 15.0 ± 0% 15.0 ± 0% ~ (all equal)
RedisSurvey/2_nodes_512B_cluster-16 129 ±10% 89 ± 6% -30.75% (p=0.008 n=5+5)
RedisSurvey/3_nodes_512B_cluster-16 175 ±17% 140 ± 6% -19.79% (p=0.008 n=5+5)
RedisSurvey/4_nodes_512B_cluster-16 252 ± 7% 200 ± 8% -20.84% (p=0.008 n=5+5)
RedisSurvey/5_nodes_512B_cluster-16 290 ±13% 253 ±14% ~ (p=0.063 n=5+5)
RedisSurvey/10_nodes_512B_cluster-16 541 ± 5% 472 ± 3% -12.82% (p=0.008 n=5+5)
RedisSurvey/100_nodes_512B_cluster-16 5.85k ± 3% 4.79k ± 2% -18.09% (p=0.008 n=5+5)
RedisSurvey/1_node_4096B_cluster-16 15.0 ± 0% 15.0 ± 0% ~ (all equal)
RedisSurvey/2_nodes_4096B_cluster-16 151 ±15% 106 ±13% -30.25% (p=0.008 n=5+5)
RedisSurvey/3_nodes_4096B_cluster-16 188 ±12% 184 ±12% ~ (p=0.722 n=5+5)
RedisSurvey/4_nodes_4096B_cluster-16 271 ±22% 226 ±19% ~ (p=0.222 n=5+5)
RedisSurvey/5_nodes_4096B_cluster-16 329 ± 7% 282 ±14% -14.17% (p=0.008 n=5+5)
RedisSurvey/10_nodes_4096B_cluster-16 612 ±11% 571 ±17% ~ (p=0.222 n=5+5)
RedisSurvey/100_nodes_4096B_cluster-16 9.93k ±10% 8.00k ±16% -19.36% (p=0.016 n=5+5)
RedisConsistentIndex-16 1.00 ± 0% 1.00 ± 0% ~ (all equal)
RedisIndex-16 1.00 ± 0% 1.00 ± 0% ~ (all equal)
RedisPublish_1Ch/lists-16 11.0 ± 0% 9.6 ± 6% -12.73% (p=0.000 n=4+5)
RedisPublish_1Ch/streams-16 10.8 ±11% 10.0 ± 0% ~ (p=0.095 n=5+4)
RedisPublish_1Ch/lists_cluster-16 12.2 ±10% 9.8 ±18% -19.67% (p=0.024 n=5+5)
RedisPublish_1Ch/streams_cluster-16 12.2 ±10% 10.0 ±20% -18.03% (p=0.032 n=5+5)
RedisPublish_ManyCh/lists-16 12.0 ± 0% 10.6 ±13% -11.67% (p=0.048 n=5+5)
RedisPublish_ManyCh/streams-16 12.4 ± 5% 10.8 ±11% -12.90% (p=0.032 n=5+5)
RedisPublish_ManyCh/lists_cluster-16 12.4 ± 5% 12.0 ± 0% ~ (p=0.333 n=5+4)
RedisPublish_ManyCh/streams_cluster-16 12.0 ± 8% 11.0 ± 9% ~ (p=0.246 n=5+5)
RedisPublish_History_1Ch/lists-16 38.6 ± 9% 31.4 ±11% -18.65% (p=0.008 n=5+5)
RedisPublish_History_1Ch/streams-16 37.8 ± 8% 32.4 ±14% ~ (p=0.056 n=5+5)
RedisPublish_History_1Ch/lists_cluster-16 47.4 ± 1% 33.4 ±14% -29.54% (p=0.008 n=5+5)
RedisPublish_History_1Ch/streams_cluster-16 45.6 ±16% 37.6 ±18% ~ (p=0.056 n=5+5)
RedisPub_History_ManyCh/lists-16 41.0 ± 7% 32.2 ± 6% -21.46% (p=0.008 n=5+5)
RedisPub_History_ManyCh/streams-16 37.6 ± 6% 33.8 ± 7% -10.11% (p=0.016 n=5+5)
RedisPub_History_ManyCh/lists_cluster-16 37.6 ± 4% 31.2 ±13% -17.02% (p=0.008 n=5+5)
RedisPub_History_ManyCh/streams_cluster-16 38.2 ± 3% 32.4 ±11% -15.18% (p=0.008 n=5+5)
RedisSubscribe/non_cluster-16 28.2 ± 8% 20.4 ± 3% -27.66% (p=0.008 n=5+5)
RedisSubscribe/with_cluster-16 31.0 ± 3% 19.8 ± 6% -36.13% (p=0.008 n=5+5)
RedisHistory_1Ch/lists-16 59.4 ± 9% 48.8 ± 8% -17.85% (p=0.008 n=5+5)
RedisHistory_1Ch/streams-16 96.2 ± 2% 81.2 ± 5% -15.59% (p=0.008 n=5+5)
RedisHistory_1Ch/lists_cluster-16 65.4 ± 5% 54.4 ± 1% -16.82% (p=0.008 n=5+5)
RedisHistory_1Ch/streams_cluster-16 97.4 ± 3% 88.0 ±10% -9.65% (p=0.008 n=5+5)
RedisRecover_1Ch/lists-16 5.20k ± 1% 4.18k ± 1% -19.60% (p=0.008 n=5+5)
RedisRecover_1Ch/streams-16 110 ± 8% 96 ± 8% -12.52% (p=0.008 n=5+5)
RedisRecover_1Ch/lists_cluster-16 5.12k ± 1% 4.15k ± 0% -18.89% (p=0.016 n=5+4)
RedisRecover_1Ch/streams_cluster-16 117 ± 4% 101 ± 6% -13.65% (p=0.008 n=5+5)
RedisHistoryIteration/lists-16 5.33M ± 2% 4.27M ± 1% -19.80% (p=0.008 n=5+5)
RedisHistoryIteration/streams-16 164k ± 7% 147k ±10% -9.83% (p=0.016 n=5+5)
RedisHistoryIteration/lists_cluster-16 5.26M ± 2% 4.23M ± 1% -19.48% (p=0.008 n=5+5)
RedisHistoryIteration/streams_cluster-16 162k ± 5% 147k ± 9% ~ (p=0.056 n=5+5)
RedisAddPresence_1Ch/lists-16 26.0 ± 4% 19.8 ±11% -23.85% (p=0.008 n=5+5)
RedisAddPresence_1Ch/streams-16 22.2 ±10% 22.2 ±10% ~ (p=1.000 n=5+5)
RedisAddPresence_1Ch/lists_cluster-16 31.0 ±10% 25.4 ±17% -18.06% (p=0.008 n=5+5)
RedisAddPresence_1Ch/streams_cluster-16 31.2 ± 2% 25.4 ±14% -18.72% (p=0.016 n=4+5)
RedisPresence_1Ch/lists-16 34.0 ±12% 26.8 ± 8% -21.18% (p=0.008 n=5+5)
RedisPresence_1Ch/streams-16 34.8 ± 8% 28.0 ± 7% -19.54% (p=0.008 n=5+5)
RedisPresence_1Ch/lists_cluster-16 37.0 ±11% 29.4 ±12% -20.54% (p=0.008 n=5+5)
RedisPresence_1Ch/streams_cluster-16 39.4 ± 4% 29.6 ±18% -24.87% (p=0.008 n=5+5)
RedisPresence_ManyCh/lists-16 24.8 ± 5% 22.8 ±12% ~ (p=0.063 n=5+5)
RedisPresence_ManyCh/streams-16 25.8 ±11% 22.0 ±14% -14.73% (p=0.040 n=5+5)
RedisPresence_ManyCh/lists_cluster-16 27.8 ± 6% 22.6 ± 7% -18.71% (p=0.008 n=5+5)
RedisPresence_ManyCh/streams_cluster-16 28.6 ± 2% 23.0 ± 9% -19.58% (p=0.008 n=5+5)
I think overall looks good and worth moving forward 👍
As the next step - let's concentrate on timeouts? In general the goal is to avoid calls without timeout. We need to make sure we don't have infinite waiting for sending/waiting indefinitely. In Redigo case we achieve this by having both WriteTimeout and ReadTimeout set for the pool.
Possibly go-redis also have a way to set global timeouts to use. Possibly not - and we will have to use per-call context-based timeouts then. Using per-call context timeouts will definitely hit the performance so I'd prefer mimicking Redigo as first step if possible.
From time to time I am thinking that all our Engine methods should have context.Context as first argument (and possibly Node methods that call Engine), but that's sth we can consider later. Whether using context cancellation/timeouts from the outside is good or not is not obvious for me at this point. Most Go libraries tend to propagate Context in every call where possible, but since we can't really cancel operation sent to Redis - I still think global timeouts have sense and pretty practical.
I will try to look at timeouts too and code here more closely during next days - for now a bit busy with various home tasks :) Many thanks for pushing this forward @j178.
Some interesting links I just found: https://redis.uptrace.dev/guide/go-redis-debugging.html#timeouts and https://uptrace.dev/blog/posts/go-context-timeout.html