Skip to content

rest_bench.md

Latest commit

 

History

History
147 lines (111 loc) · 3.69 KB

rest_bench.md

File metadata and controls

147 lines (111 loc) · 3.69 KB

REST API benchmark

Here are some results of benchmarking GCache performance via REST API using wrk utility.

Setup

  • Test machine: Macbook Pro, Intel Core i5-4278U CPU @ 2.60GHz, 8Gb RAM;
  • loopback networking, ulimit 7168;
  • GCache settings: master, 64 shards, file dumps turned off.

Results

SET

Insert unique keys (worst time complexity):

wrk -s post_unique.lua -t10 -c30 -d60s http://localhost:8080/cache/v1/item
Running 1m test @ http://localhost:8080/cache/v1/item
  10 threads and 30 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     2.31ms    4.25ms  168.95ms  94.55%
    Req/Sec     1.92k     437.33  4.48k     69.70%
  1148875 requests in 1.00m, 204.92MB read
Requests/sec:  19135.48
Transfer/sec:      3.41MB

Update single key (best time complexity)

wrk -s post_update.lua -t10 -c30 -d60s http://localhost:8080/cache/v1/item
Running 1m test @ http://localhost:8080/cache/v1/item
  10 threads and 30 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.39ms  2.75ms    95.43ms   95.18%
    Req/Sec     3.21k   356.97     4.39k    78.52%
  1914267 requests in 1.00m, 337.73MB read
Requests/sec:   31895.16
Transfer/sec:       5.63MB

GET

Retrieve different non-existing keys:

wrk -s get_var.lua -t10 -c30 -d60s http://localhost:8080/cache/v1/item
Running 1m test @ http://localhost:8080/cache/v1/item
  10 threads and 30 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.38ms   2.31ms   78.11ms   94.79%
    Req/Sec     3.02k    311.03     4.08k    75.68%
  1806133 requests in 1.00m, 284.21MB read
  Non-2xx or 3xx responses: 1806133
Requests/sec:   30091.40
Transfer/sec:       4.74MB

Retrieve different stored keys:

wrk -s get_var.lua -t10 -c30 -d60s http://localhost:8080/cache/v1/item
Running 1m test @ http://localhost:8080/cache/v1/item
  10 threads and 30 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.93ms   4.48ms   140.80ms   95.13%
    Req/Sec     2.74k    445.75     4.95k    69.57%
  1639081 requests in 1.00m, 265.68MB read
  Non-2xx or 3xx responses: 483652
Requests/sec:  27300.32
Transfer/sec:      4.43MB

REMOVE

Delete unique keys from storage:

wrk -s remove_var.lua -t10 -c30 -d60s http://localhost:8080/cache/v1/item
Running 1m test @ http://localhost:8080/cache/v1/item
  10 threads and 30 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     1.34ms   2.58ms   97.11ms   95.16%
    Req/Sec     3.22k    336.55    5.47k    76.98%
  1923526 requests in 1.00m, 117.40MB read
Requests/sec:  32044.70
Transfer/sec:      1.96MB

Test scripts

Lua scripts used for benchmarking:

get_var.lua

keyCount = 0

request = function()
    wrk.body = string.format("{\"key\": \"testkey%d\"}", keyCount)
    wrk.headers["Content-Type"] = "application/json"
    keyCount = keyCount + 1
    return wrk.format("GET",nil)
end

post_update.lua

wrk.method = "POST"
wrk.body   = "{\"key\": \"sometestkey\", \"value\": \"some test string here\", \"ttl\": 60}"
wrk.headers["Content-Type"] = "application/json"

post_unique.lua

keyCount = 0

request = function()
    wrk.body = string.format("{\"key\": \"testkey%d\", \"value\": \"some test string here\", \"ttl\": 600}", keyCount)
    wrk.headers["Content-Type"] = "application/json"
    keyCount = keyCount + 1
    return wrk.format("POST",nil)
end

remove_var.lua

keyCount = 0

request = function()
    wrk.body   = string.format("{\"key\": \"testkey%d\"}", keyCount)
    wrk.headers["Content-Type"] = "application/json"
    keyCount = keyCount + 1
    return wrk.format("DELETE",nil)
end