AWS Postgres RDS Proxy

Posted on | Edited on | In , | Comments: | Views:
Symbols count in article: 9.3k | Reading time ≈ 17 mins.

前言

在 Postgresql 世界裡每建立一個 Connection，都會有一個 process 來服務該請求，Postgresql 沒有 ThreadPool 概念所以在 Connection 使用量大時會發生一些問題

  • 資源浪費
  • 超過同時間超過 max_connection 後面連接會無法連接

一個 open source 插件 PgBouncer 就是幫我們解決上面問題，不過如果你是使用 AWS RDS 來當作 DB 就很適合使用 RDS Proxy

本次使用免費帳號範圍提供的機器 db.t2.micro

另外因為 RDS Proxy 在 EC2 內網才能連接，所以我也利用免費帳號準備一台 EC2 並安裝 pgbench 進行壓力測試

壓測前準備 & 說明

我把 max_connections 最大連接數量調整成 550

在進行壓力測試，我分別會針對 100,200,300,400,500,600 這幾個 connection 數量

針對有使用 RDS Proxy 跟沒有使用 Proxy 來比較效能

這邊我使用 pgbench 來進行壓力測試

這邊我選用 xridge/pgbench pgbench 來方便我們壓力測試

一開始我們先建立 pgbench 初始資料

1
docker run xridge/pgbench -i -s 4 --unlogged-tables --foreign-keys postgresql://{{postgres info}}

-s 可以依照 RDS 規格來改變 (代表有幾個用戶資料)

準備完後就進入 EC2 虛擬機準備進行壓力測試

沒有 ThreadPool 測試資料

1
docker run  xridge/pgbench -S -c 100 -j 20 -t 10 postgresql://{{postgres info}}

-c 代表測試併發 client 數量

使用 100,200,300,400,500,600 測試結果如下

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 100
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 1000/1000
latency average = 32.723 ms
initial connection time = 667.011 ms
tps = 3056.001222 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 200
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 2000/2000
latency average = 164.399 ms
initial connection time = 1441.067 ms
tps = 1216.553152 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 300
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 3000/3000
latency average = 261.908 ms
initial connection time = 2369.158 ms
tps = 1145.441258 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 400
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 4000/4000
latency average = 647.347 ms
initial connection time = 3203.360 ms
tps = 617.906720 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 500
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 5000/5000
latency average = 1396.268 ms
initial connection time = 5780.255 ms
tps = 358.097443 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
pgbench: error: connection to server at xxxxDB (172.31.28.185), port 5432 failed: FATAL:  remaining connection slots are reserved for non-replication superuser connections
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 600
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 0/6000
pgbench: fatal: Run was aborted; the above results are incomplete.

沒有使用 ThreadPool 情況超過 max_connection 就會出現異常導致無法完成請求

有 ThreadPool 測試資料 (RDS-Proxy)

我個人覺得 RDS-proxy 官網使用簡介蠻詳細的可以參考就可以建立這邊就不多贅述

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 100
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 1000/1000
latency average = 97.921 ms
initial connection time = 562.785 ms
tps = 1021.229315 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 200
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 2000/2000
latency average = 110.732 ms
initial connection time = 1095.188 ms
tps = 1806.162627 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 300
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 3000/3000
latency average = 205.738 ms
initial connection time = 1474.857 ms
tps = 1458.164531 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 400
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 4000/4000
latency average = 288.763 ms
initial connection time = 1798.422 ms
tps = 1385.217102 (without initial connection time)


pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 500
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 5000/5000
latency average = 288.763 ms
initial connection time = 1798.422 ms
tps = 1342.151131 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 600
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 6000/6000
latency average = 525.494 ms
initial connection time = 2539.418 ms
tps = 1141.782562 (without initial connection time)

pgbench (14.1 (Ubuntu 14.1-1.pgdg20.04+1), server 11.10)
starting vacuum...end.
transaction type: <builtin: select only>
scaling factor: 4
query mode: simple
number of clients: 700
number of threads: 20
number of transactions per client: 10
number of transactions actually processed: 7000/7000
latency average = 452.461 ms
initial connection time = 3174.766 ms
tps = 1547.095351 (without initial connection time)

RDS Proxy 使用前後比較

在 max_Connection = 550 壓力測試結果如下

發現在 connection = 100 時沒有 proxy 效率比有proxy好，但併發 coonnection 上升後效率就急速下降,反而 Proxy 就很穩定 rps 一直在 1000 以上

另外一般情況如果 connection 數量大於 max_Connection 就會噴出錯誤，這時有使用 ThreadPool 就是一個很好的保護機制，會進入 queue 消化請求

小結

在 SqlServer 預設有 ConnectionPool 但在 Postgresql 就要自己注意，我個人是建議要在 postgresql DB 前面使用 ThreadPool 一來更有穩定效率，二來在大量請求來時不會出 error.

此文作者：Daniel Shih(石頭)
此文地址： https://isdaniel.github.io/aws-rds-proxy/
版權聲明：本博客所有文章除特別聲明外，均採用 CC BY-NC-SA 3.0 TW 許可協議。轉載請註明出處！

Related Posts

如果本文對您幫助很大，可街口支付斗內鼓勵石頭^^