Skip to content

Latest commit

 

History

History
146 lines (127 loc) · 6.57 KB

README.md

File metadata and controls

146 lines (127 loc) · 6.57 KB
Raw

HTTP server

In this project I experiment with various techniques, architectures and infrastructure (on Linux) to make fast HTTP 1.1 servers. I'm using the Rust may-minihttp microframework as a baseline and trying to make something faster. may-minihttp ended up on 2nd place in the plaintext Techempower benchmarks during round 21.

Here I'll simply loadtest the servers using wrk2 by giltene on my desktop machine to observe latency distribution and throughput and use that as a tool to learn more about Linux, HTTP etc etc...

This will always be a work in progress/playground for stuff I want to try.

Current projects under testing

This is never going to be a fair comparison, as my implementation is not a real server. Just an excuse to try out new things and theories.

My hardware

$ uname -a
Linux desktop 6.2.6-76060206-generic #202303130630~1689015125~22.04~ab2190e SMP PREEMPT_DYNAMIC Mon J x86_64 x86_64 x86_64 GNU/Linux
$ lspci | egrep -i --color 'network|ethernet'
08:00.0 Ethernet controller: Intel Corporation I211 Gigabit Network Connection (rev 03)
$ lscpu
Architecture:            x86_64
  CPU op-mode(s):        32-bit, 64-bit
  Address sizes:         48 bits physical, 48 bits virtual
  Byte Order:            Little Endian
CPU(s):                  12
  On-line CPU(s) list:   0-11
Vendor ID:               AuthenticAMD
  Model name:            AMD Ryzen 5 5600X 6-Core Processor
    CPU family:          25
    Model:               33
    Thread(s) per core:  2
    Core(s) per socket:  6
    Socket(s):           1
    Stepping:            0
    Frequency boost:     enabled
    CPU max MHz:         4650.2920
    CPU min MHz:         2200.0000
    BogoMIPS:            7385.58
    Flags:               fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt
                         pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16
                         sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefet
                         ch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate ssbd mba ibrs ib
                         pb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xget
                         bv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr rdpru wbnoinvd arat npt lbrv svm_lock nrip
                         _save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif v_spec_ctrl umip pku ospk
                         e vaes vpclmulqdq rdpid overflow_recov succor smca fsrm
Virtualization features:
  Virtualization:        AMD-V
Caches (sum of all):
  L1d:                   192 KiB (6 instances)
  L1i:                   192 KiB (6 instances)
  L2:                    3 MiB (6 instances)
  L3:                    32 MiB (1 instance)

Methodology

It's certainly harder to run these kinds of load test as opposed to microbenchmarks. I try to run my hardware as idle as possible, but there is alawys going to be some noise. I pin wrk to 4 specific threads, and I also pin the servers to specific threads (if I'm able). There is no other tuning done. NIC, kernel etc uses whatever CPUs it wants and scheduling priority are all defaults.

Latency

Baseline

$ taskset -c 8-11 wrk -t4 -c1000 -d15s -R100000 --latency http://127.0.0.1:8080
Running 15s test @ http://127.0.0.1:8080
  4 threads and 1000 connections
  Thread calibration: mean lat.: 1.418ms, rate sampling interval: 10ms
  Thread calibration: mean lat.: 1.424ms, rate sampling interval: 10ms
  Thread calibration: mean lat.: 1.406ms, rate sampling interval: 10ms
  Thread calibration: mean lat.: 1.428ms, rate sampling interval: 10ms
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.41ms  582.55us   9.78ms   69.90%
    Req/Sec    26.38k     3.78k   40.33k    82.41%
  Latency Distribution (HdrHistogram - Recorded Latency)
 50.000%    1.35ms
 75.000%    1.78ms
 90.000%    2.15ms
 99.000%    2.79ms
 99.900%    4.70ms
 99.990%    8.27ms
 99.999%    8.94ms
100.000%    9.78ms

#[Mean    =        1.405, StdDeviation   =        0.583]
#[Max     =        9.776, Total count    =       375000]
#[Buckets =           27, SubBuckets     =         2048]
----------------------------------------------------------
  1438076 requests in 15.00s, 172.80MB read
Requests/sec:  95861.78
Transfer/sec:     11.52MB

My server

$ taskset -c 8-11 wrk -t4 -c1000 -d15s -R100000 --latency http://127.0.0.1:8081
Running 15s test @ http://127.0.0.1:8081
  4 threads and 1000 connections
  Thread calibration: mean lat.: 1.372ms, rate sampling interval: 10ms
  Thread calibration: mean lat.: 1.353ms, rate sampling interval: 10ms
  Thread calibration: mean lat.: 1.410ms, rate sampling interval: 10ms
  Thread calibration: mean lat.: 1.339ms, rate sampling interval: 10ms
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.38ms  575.84us   4.62ms   66.34%
    Req/Sec    26.31k     3.47k   42.44k    83.40%
  Latency Distribution (HdrHistogram - Recorded Latency)
 50.000%    1.34ms
 75.000%    1.78ms
 90.000%    2.13ms
 99.000%    2.83ms
 99.900%    3.27ms
 99.990%    3.90ms
 99.999%    4.48ms
100.000%    4.62ms

#[Mean    =        1.377, StdDeviation   =        0.576]
#[Max     =        4.616, Total count    =       374995]
#[Buckets =           27, SubBuckets     =         2048]
----------------------------------------------------------
  1427928 requests in 15.00s, 100.77MB read
Requests/sec:  95186.12
Transfer/sec:      6.72MB

Topics to explore