标签

PostgreSQL , tpcb , pgbench , 100亿


背景

PostgreSQL 11 发布在即,以下是tpcc与tpch测试的结果:

《PostgreSQL 11 tpcc 测试(103万tpmC on ECS) - use sysbench-tpcc by Percona-Lab》

《(TPC-H测试 SF=10,SF=200) PostgreSQL 11 vs 10 vs Deepgreen》

tpcb是pgbench标准测试CASE,本文使用与以上同样的测试机器(部署方式请参考以上),测试100亿tpch的性能。

参数

1、postgresql.auto.conf

listen_addresses = '0.0.0.0'
port = 1921
max_connections = 2000
superuser_reserved_connections = 13
unix_socket_directories = '., /var/run/postgresql, /tmp'
tcp_keepalives_idle = 60
tcp_keepalives_interval = 10
tcp_keepalives_count = 10
shared_buffers = 64GB
max_prepared_transactions = 2000
work_mem = 8MB
maintenance_work_mem = 2GB
dynamic_shared_memory_type = posix
vacuum_cost_delay = 0
bgwriter_delay = 10ms
bgwriter_lru_maxpages = 1000
bgwriter_lru_multiplier = 10.0
effective_io_concurrency = 0
max_worker_processes = 128
max_parallel_maintenance_workers = 24
max_parallel_workers_per_gather = 0
parallel_leader_participation = on
min_parallel_table_scan_size=0
min_parallel_index_scan_size=0
parallel_setup_cost=0
parallel_tuple_cost=0
max_parallel_workers = 64
wal_level = minimal
synchronous_commit = off
wal_writer_delay = 10ms
checkpoint_timeout = 35min
max_wal_size = 128GB
min_wal_size = 32GB
checkpoint_completion_target = 0.1
max_wal_senders = 0
effective_cache_size = 400GB
log_destination = 'csvlog'
logging_collector = on
log_directory = 'log'
log_filename = 'postgresql-%a.log'
log_truncate_on_rotation = on
log_rotation_age = 1d
log_rotation_size = 0
log_checkpoints = on
log_connections = on
log_disconnections = on
log_error_verbosity = verbose
log_line_prefix = '%m [%p] '
log_timezone = 'PRC'
log_autovacuum_min_duration = 0
autovacuum_max_workers = 16
autovacuum_freeze_max_age = 1200000000
autovacuum_multixact_freeze_max_age = 1400000000
autovacuum_vacuum_cost_delay = 0ms
vacuum_freeze_table_age = 1150000000
vacuum_multixact_freeze_table_age = 1150000000
datestyle = 'iso, mdy'
timezone = 'PRC'
lc_messages = 'C'
lc_monetary = 'C'
lc_numeric = 'C'
lc_time = 'C'
default_text_search_config = 'pg_catalog.english'
jit = off
cpu_tuple_cost=0.00018884145574257426
cpu_index_tuple_cost = 0.00433497085216479990
cpu_operator_cost = 0.00216748542608239995
seq_page_cost=0.014329
random_page_cost = 0.016

测试

初始100亿数据

使用pgbench 初始化100亿tpcb测试数据。

man pgbench

-i
--initialize  Required to invoke initialization mode.  -I init_steps
--init-steps=init_steps  Perform just a selected set of the normal initialization steps.    init_steps specifies the initialization steps to be performed,   using one character per step. Each step is invoked in the specified order.   The default is dtgvp.   The available steps are:  d (Drop)  Drop any existing pgbench tables.  t (create Tables)  Create the tables used by the standard pgbench scenario,   namely pgbench_accounts, pgbench_branches, pgbench_history,   and pgbench_tellers.  g (Generate data)  Generate data and load it into the standard tables,   replacing any data already present.  v (Vacuum)  Invoke VACUUM on the standard tables.  p (create Primary keys)  Create primary key indexes on the standard tables.  f (create Foreign keys)  Create foreign key constraints between the standard tables.   (Note that this step is not performed by default.)

初始化

pgbench -i -s 100000 -I dtg -n --tablespace=tbs2

耗时8385秒,约119万行/s。

9998500000 of 10000000000 tuples (99%) done (elapsed 8384.23 s, remaining 1.26 s)
9998600000 of 10000000000 tuples (99%) done (elapsed 8384.31 s, remaining 1.17 s)
9998700000 of 10000000000 tuples (99%) done (elapsed 8384.40 s, remaining 1.09 s)
9998800000 of 10000000000 tuples (99%) done (elapsed 8384.48 s, remaining 1.01 s)
9998900000 of 10000000000 tuples (99%) done (elapsed 8384.56 s, remaining 0.92 s)
9999000000 of 10000000000 tuples (99%) done (elapsed 8384.65 s, remaining 0.84 s)
9999100000 of 10000000000 tuples (99%) done (elapsed 8384.73 s, remaining 0.75 s)
9999200000 of 10000000000 tuples (99%) done (elapsed 8384.82 s, remaining 0.67 s)
9999300000 of 10000000000 tuples (99%) done (elapsed 8384.90 s, remaining 0.59 s)
9999400000 of 10000000000 tuples (99%) done (elapsed 8384.98 s, remaining 0.50 s)
9999500000 of 10000000000 tuples (99%) done (elapsed 8385.07 s, remaining 0.42 s)
9999600000 of 10000000000 tuples (99%) done (elapsed 8385.15 s, remaining 0.34 s)
9999700000 of 10000000000 tuples (99%) done (elapsed 8385.24 s, remaining 0.25 s)
9999800000 of 10000000000 tuples (99%) done (elapsed 8385.33 s, remaining 0.17 s)
9999900000 of 10000000000 tuples (99%) done (elapsed 8385.41 s, remaining 0.08 s)
10000000000 of 10000000000 tuples (100%) done (elapsed 8385.49 s, remaining 0.00 s)

创建索引

1、强制设置并行度为24,a,b,c,d取最小

alter table pgbench_accounts set (parallel_workers =32);  # a
max_worker_processes = 128   # b
max_parallel_maintenance_workers = 24   # c
max_parallel_workers = 64   # d  max_parallel_workers_per_gather = 0
min_parallel_table_scan_size=0
min_parallel_index_scan_size=0
parallel_setup_cost=0
parallel_tuple_cost=0

2、创建tpcb数据表索引,100亿记录,创建索引耗时44分钟。

pgbench -i -I p --index-tablespace=tbs1

IO监测

Total DISK READ :       2.77 G/s | Total DISK WRITE :       2.86 G/s
Actual DISK READ:       2.79 G/s | Actual DISK WRITE:    1795.02 M/s  TID  PRIO  USER     DISK READ  DISK WRITE  SWAPIN     IO>    COMMAND
13683 be/4 postgres  142.87 M/s  141.55 M/s  0.00 %  1.47 % postgres: parallel worker for PID 13655
13684 be/4 postgres  114.55 M/s  124.78 M/s  0.00 %  1.07 % postgres: parallel worker for PID 13655
13686 be/4 postgres  163.22 M/s  165.50 M/s  0.00 %  0.64 % postgres: parallel worker for PID 13655
13671 be/4 postgres  162.38 M/s  165.76 M/s  0.00 %  0.49 % postgres: parallel worker for PID 13655
13676 be/4 postgres  169.80 M/s  169.31 M/s  0.00 %  0.45 % postgres: parallel worker for PID 13655
13672 be/4 postgres  160.74 M/s  165.52 M/s  0.00 %  0.41 % postgres: parallel worker for PID 13655
13685 be/4 postgres  174.74 M/s  177.10 M/s  0.00 %  0.37 % postgres: parallel worker for PID 13655
13687 be/4 postgres  172.48 M/s  173.00 M/s  0.00 %  0.36 % postgres: parallel worker for PID 13655
13677 be/4 postgres  116.31 M/s  124.28 M/s  0.00 %  0.35 % postgres: parallel worker for PID 13655
13680 be/4 postgres  165.52 M/s  177.19 M/s  0.00 %  0.34 % postgres: parallel worker for PID 13655
13674 be/4 postgres  167.31 M/s  169.44 M/s  0.00 %  0.12 % postgres: parallel worker for PID 13655
13655 be/4 postgres  114.55 M/s  122.85 M/s  0.00 %  0.05 % postgres: postgres postgres [local] ALTER TABLE
13678 be/4 postgres  175.54 M/s  177.10 M/s  0.00 %  0.04 % postgres: parallel worker for PID 13655
13670 be/4 postgres  127.29 M/s  126.61 M/s  0.00 %  0.01 % postgres: parallel worker for PID 13655
13682 be/4 postgres  162.41 M/s  165.74 M/s  0.00 %  0.00 % postgres: parallel worker for PID 13655
13673 be/4 postgres  145.02 M/s  152.94 M/s  0.00 %  0.00 % postgres: parallel worker for PID 13655
13675 be/4 postgres  167.42 M/s  169.41 M/s  0.00 %  0.00 % postgres: parallel worker for PID 13655
13679 be/4 postgres  117.00 M/s  128.97 M/s  0.00 %  0.00 % postgres: parallel worker for PID 13655
13681 be/4 postgres  121.97 M/s  132.34 M/s  0.00 %  0.00 % postgres: parallel worker for PID 13655
13642 be/4 postgres    0.00 B/s    0.00 B/s  0.00 %  0.00 % postgres: logical replication launcher
13634 be/4 postgres    0.00 B/s    0.00 B/s  0.00 %  0.00 % postgres
13635 be/4 postgres    0.00 B/s    0.00 B/s  0.00 %  0.00 % postgres: logger
13637 be/4 postgres    0.00 B/s    0.00 B/s  0.00 %  0.00 % postgres: checkpointer
13638 be/4 postgres    0.00 B/s    0.00 B/s  0.00 %  0.00 % postgres: background writer
13639 be/4 postgres    0.00 B/s    0.00 B/s  0.00 %  0.00 % postgres: walwriter
13640 be/4 postgres    0.00 B/s    0.00 B/s  0.00 %  0.00 % postgres: autovacuum launcher
13641 be/4 postgres    0.00 B/s    0.00 B/s  0.00 %  0.00 % postgres: stats collector

耗时,43分50秒左右。

select now()-query_start, query from pg_stat_activity where query ~ 'primary key';  \watch 3  Sun 16 Sep 2018 05:53:13 PM CST (every 3s)  ?column?     |                                      query
-----------------+----------------------------------------------------------------------------------  00:43:49.994922 | alter table pgbench_accounts add primary key (aid) using index tablespace "tbs1"  00:00:00        | select now()-query_start,query from pg_stat_activity where query ~ 'tbs1';
(2 rows)  Sun 16 Sep 2018 05:53:16 PM CST (every 3s)  ?column? |                                   query
----------+----------------------------------------------------------------------------  00:00:00 | select now()-query_start,query from pg_stat_activity where query ~ 'tbs1';
(1 row)

tpcb 100亿,表、索引大小

postgres=# \l+ postgres  List of databases  Name   |  Owner   | Encoding  | Collate | Ctype | Access privileges |  Size   | Tablespace |                Description
----------+----------+-----------+---------+-------+-------------------+---------+------------+--------------------------------------------  postgres | postgres | SQL_ASCII | C       | C     |                   | 2599 GB | pg_default | default administrative connection database
(1 row)  postgres=# \dt+ pgbench*  List of relations  Schema |       Name       | Type  |  Owner   |  Size   | Description
--------+------------------+-------+----------+---------+-------------  public | pgbench_accounts | table | postgres | 1251 GB |   public | pgbench_branches | table | postgres | 3576 kB |   public | pgbench_history  | table | postgres | 0 bytes |   public | pgbench_tellers  | table | postgres | 42 MB   |
(4 rows)  postgres=# \di+ pgbench*  List of relations  Schema |         Name          | Type  |  Owner   |      Table       |  Size   | Description
--------+-----------------------+-------+----------+------------------+---------+-------------  public | pgbench_accounts_pkey | index | postgres | pgbench_accounts | 209 GB  |   public | pgbench_branches_pkey | index | postgres | pgbench_branches | 2208 kB |   public | pgbench_tellers_pkey  | index | postgres | pgbench_tellers  | 21 MB   |
(3 rows)

这里建议pgbench_accounts用 pg_pathman hash 分区。

tpc-b 只读性能 (tps: 118053)

测试

pgbench -M prepared -v -r -P 1 -c 64 -j 64 -T 3600 -S

结果

.........
progress: 3550.0 s, 131287.4 tps, lat 0.492 ms stddev 5.895
progress: 3551.0 s, 117855.0 tps, lat 0.543 ms stddev 7.239
progress: 3552.0 s, 102975.0 tps, lat 0.527 ms stddev 6.052
progress: 3553.0 s, 128299.1 tps, lat 0.575 ms stddev 7.191
progress: 3554.0 s, 122215.5 tps, lat 0.522 ms stddev 6.947
progress: 3555.0 s, 92727.5 tps, lat 0.689 ms stddev 9.055
progress: 3556.0 s, 144550.4 tps, lat 0.433 ms stddev 3.430
progress: 3557.0 s, 95982.1 tps, lat 0.565 ms stddev 7.052
progress: 3558.0 s, 118369.9 tps, lat 0.638 ms stddev 9.017
progress: 3559.0 s, 128757.4 tps, lat 0.497 ms stddev 5.789
progress: 3560.0 s, 99634.5 tps, lat 0.529 ms stddev 6.252
progress: 3561.0 s, 118071.4 tps, lat 0.635 ms stddev 8.541
progress: 3562.0 s, 131764.3 tps, lat 0.488 ms stddev 5.816
progress: 3563.0 s, 118408.9 tps, lat 0.540 ms stddev 7.308
progress: 3564.0 s, 97566.1 tps, lat 0.538 ms stddev 6.447
progress: 3565.0 s, 110077.0 tps, lat 0.686 ms stddev 9.656
progress: 3566.0 s, 128585.3 tps, lat 0.498 ms stddev 5.901
progress: 3567.0 s, 116302.4 tps, lat 0.514 ms stddev 6.269
progress: 3568.0 s, 111013.7 tps, lat 0.615 ms stddev 7.788
progress: 3569.0 s, 116588.9 tps, lat 0.549 ms stddev 7.378
progress: 3570.0 s, 118875.0 tps, lat 0.538 ms stddev 6.716
progress: 3571.0 s, 124222.9 tps, lat 0.514 ms stddev 6.566
progress: 3572.0 s, 93014.1 tps, lat 0.551 ms stddev 6.824
progress: 3573.0 s, 120148.7 tps, lat 0.639 ms stddev 8.537
progress: 3574.0 s, 131078.1 tps, lat 0.488 ms stddev 5.765
progress: 3575.0 s, 105341.2 tps, lat 0.509 ms stddev 5.548
progress: 3576.0 s, 121529.8 tps, lat 0.612 ms stddev 8.044
progress: 3577.0 s, 129391.4 tps, lat 0.479 ms stddev 5.173
progress: 3578.0 s, 86950.4 tps, lat 0.612 ms stddev 8.670
progress: 3579.0 s, 124735.8 tps, lat 0.616 ms stddev 8.173
progress: 3580.0 s, 120284.6 tps, lat 0.532 ms stddev 6.892
progress: 3581.0 s, 96491.2 tps, lat 0.559 ms stddev 7.261
progress: 3582.0 s, 121068.8 tps, lat 0.611 ms stddev 8.041
progress: 3583.0 s, 137346.3 tps, lat 0.466 ms stddev 4.567
progress: 3584.0 s, 104130.5 tps, lat 0.526 ms stddev 6.362
progress: 3585.0 s, 114540.8 tps, lat 0.639 ms stddev 8.500
progress: 3586.0 s, 128416.6 tps, lat 0.499 ms stddev 5.967
progress: 3587.0 s, 106762.0 tps, lat 0.527 ms stddev 6.358
progress: 3588.0 s, 102988.7 tps, lat 0.697 ms stddev 9.452
progress: 3589.0 s, 131513.3 tps, lat 0.487 ms stddev 5.648
progress: 3590.0 s, 116013.7 tps, lat 0.508 ms stddev 6.041
progress: 3591.0 s, 109295.3 tps, lat 0.632 ms stddev 9.098
progress: 3592.0 s, 113581.1 tps, lat 0.493 ms stddev 5.331
progress: 3593.0 s, 126594.3 tps, lat 0.569 ms stddev 6.812
progress: 3594.0 s, 116212.3 tps, lat 0.551 ms stddev 7.886
progress: 3595.0 s, 95710.3 tps, lat 0.532 ms stddev 6.368
progress: 3596.0 s, 125784.2 tps, lat 0.613 ms stddev 8.003
progress: 3597.0 s, 119773.1 tps, lat 0.534 ms stddev 6.934
progress: 3598.0 s, 97862.0 tps, lat 0.560 ms stddev 7.001
progress: 3599.0 s, 113585.9 tps, lat 0.644 ms stddev 8.661
progress: 3600.0 s, 128750.5 tps, lat 0.496 ms stddev 5.801
transaction type: <builtin: select only>
scaling factor: 100000
query mode: prepared
number of clients: 64
number of threads: 64
duration: 3600 s
number of transactions actually processed: 424993812
latency average = 0.542 ms
latency stddev = 6.751 ms
tps = 118053.277351 (including connections establishing)
tps = 118054.214576 (excluding connections establishing)
statement latencies in milliseconds:  0.001  \set aid random(1, 100000 * :scale)  0.542  SELECT abalance FROM pgbench_accounts WHERE aid = :aid;

IO监测

Total DISK READ :       4.90 G/s | Total DISK WRITE :       0.00 B/s
Actual DISK READ:       4.90 G/s | Actual DISK WRITE:       0.00 B/s  TID  PRIO  USER     DISK READ  DISK WRITE  SWAPIN     IO>    COMMAND
14357 be/4 postgres   74.24 M/s    0.00 B/s  0.00 % 43.98 % postgres: postgres postgres [local] SELECT
14257 be/4 postgres   88.29 M/s    0.00 B/s  0.00 % 42.37 % postgres: postgres postgres [local] idle
14318 be/4 postgres   75.55 M/s    0.00 B/s  0.00 % 41.86 % postgres: postgres postgres [local] SELECT
14322 be/4 postgres   73.42 M/s    0.00 B/s  0.00 % 41.53 % postgres: postgres postgres [local] SELECT
14356 be/4 postgres   77.37 M/s    0.00 B/s  0.00 % 41.49 % postgres: postgres postgres [local] idle
14319 be/4 postgres   90.94 M/s    0.00 B/s  0.00 % 41.41 % postgres: postgres postgres [local] SELECT
14307 be/4 postgres   83.40 M/s    0.00 B/s  0.00 % 41.40 % postgres: postgres postgres [local] SELECT
14327 be/4 postgres   84.56 M/s    0.00 B/s  0.00 % 41.33 % postgres: postgres postgres [local] SELECT
14351 be/4 postgres   92.47 M/s    0.00 B/s  0.00 % 41.14 % postgres: postgres postgres [local] SELECT
14333 be/4 postgres   84.52 M/s    0.00 B/s  0.00 % 40.96 % postgres: postgres postgres [local] SELECT
14352 be/4 postgres   85.67 M/s    0.00 B/s  0.00 % 40.85 % postgres: postgres postgres [local] SELECT
14301 be/4 postgres   74.16 M/s    0.00 B/s  0.00 % 40.82 % postgres: postgres postgres [local] SELECT
14348 be/4 postgres   78.53 M/s    0.00 B/s  0.00 % 40.52 % postgres: postgres postgres [local] SELECT
14358 be/4 postgres   75.76 M/s    0.00 B/s  0.00 % 40.50 % postgres: postgres postgres [local] SELECT
14353 be/4 postgres   81.66 M/s    0.00 B/s  0.00 % 40.47 % postgres: postgres postgres [local] SELECT
14317 be/4 postgres   82.71 M/s    0.00 B/s  0.00 % 40.28 % postgres: postgres postgres [local] SELECT
14347 be/4 postgres   86.47 M/s    0.00 B/s  0.00 % 40.25 % postgres: postgres postgres [local] SELECT
14312 be/4 postgres   82.92 M/s    0.00 B/s  0.00 % 40.20 % postgres: postgres postgres [local] SELECT
14334 be/4 postgres   77.82 M/s    0.00 B/s  0.00 % 40.19 % postgres: postgres postgres [local] SELECT
14310 be/4 postgres   72.51 M/s    0.00 B/s  0.00 % 40.08 % postgres: postgres postgres [local] SELECT
14335 be/4 postgres   75.72 M/s    0.00 B/s  0.00 % 39.64 % postgres: postgres postgres [local] SELECT
14359 be/4 postgres   77.37 M/s    0.00 B/s  0.00 % 39.57 % postgres: postgres postgres [local] SELECT
14341 be/4 postgres   81.28 M/s    0.00 B/s  0.00 % 39.49 % postgres: postgres postgres [local] SELECT
14314 be/4 postgres   99.25 M/s    0.00 B/s  0.00 % 39.06 % postgres: postgres postgres [local] SELECT
14325 be/4 postgres   92.97 M/s    0.00 B/s  0.00 % 38.86 % postgres: postgres postgres [local] SELECT
14272 be/4 postgres  102.70 M/s    0.00 B/s  0.00 % 38.53 % postgres: postgres postgres [local] SELECT
14306 be/4 postgres   73.62 M/s    0.00 B/s  0.00 % 38.52 % postgres: postgres postgres [local] SELECT
14321 be/4 postgres   84.34 M/s    0.00 B/s  0.00 % 37.38 % postgres: postgres postgres [local] SELECT
14344 be/4 postgres   73.31 M/s    0.00 B/s  0.00 % 37.33 % postgres: postgres postgres [local] SELECT
14349 be/4 postgres   82.12 M/s    0.00 B/s  0.00 % 36.34 % postgres: postgres postgres [local] SELECT
14350 be/4 postgres   84.84 M/s    0.00 B/s  0.00 % 36.25 % postgres: postgres postgres [local] SELECT
14354 be/4 postgres   88.81 M/s    0.00 B/s  0.00 % 36.24 % postgres: postgres postgres [local] SELECT
14329 be/4 postgres   66.78 M/s    0.00 B/s  0.00 % 34.95 % postgres: postgres postgres [local] idle
14332 be/4 postgres   77.94 M/s    0.00 B/s  0.00 % 34.89 % postgres: postgres postgres [local] SELECT
14326 be/4 postgres   73.88 M/s    0.00 B/s  0.00 % 34.81 % postgres: postgres postgres [local] SELECT
14324 be/4 postgres   72.59 M/s    0.00 B/s  0.00 % 34.74 % postgres: postgres postgres [local] SELECT
14309 be/4 postgres   70.08 M/s    0.00 B/s  0.00 % 34.62 % postgres: postgres postgres [local] SELECT
14336 be/4 postgres   76.49 M/s    0.00 B/s  0.00 % 34.60 % postgres: postgres postgres [local] SELECT
14340 be/4 postgres   78.11 M/s    0.00 B/s  0.00 % 34.54 % postgres: postgres postgres [local] SELECT
14308 be/4 postgres   67.45 M/s    0.00 B/s  0.00 % 34.48 % postgres: postgres postgres [local] SELECT
14337 be/4 postgres   70.05 M/s    0.00 B/s  0.00 % 34.35 % postgres: postgres postgres [local] idle
14305 be/4 postgres   67.95 M/s    0.00 B/s  0.00 % 34.33 % postgres: postgres postgres [local] SELECT
14311 be/4 postgres   64.25 M/s    0.00 B/s  0.00 % 34.16 % postgres: postgres postgres [local] SELECT
14313 be/4 postgres   79.62 M/s    0.00 B/s  0.00 % 34.09 % postgres: postgres postgres [local] SELECT
14316 be/4 postgres   84.10 M/s    0.00 B/s  0.00 % 34.06 % postgres: postgres postgres [local] SELECT
14355 be/4 postgres   66.73 M/s    0.00 B/s  0.00 % 34.01 % postgres: postgres postgres [local] SELECT
14345 be/4 postgres   83.61 M/s    0.00 B/s  0.00 % 33.96 % postgres: postgres postgres [local] SELECT
14343 be/4 postgres   83.96 M/s    0.00 B/s  0.00 % 33.88 % postgres: postgres postgres [local] SELECT
14323 be/4 postgres   58.07 M/s    0.00 B/s  0.00 % 33.86 % postgres: postgres postgres [local] SELECT
14282 be/4 postgres   72.93 M/s    0.00 B/s  0.00 % 33.80 % postgres: postgres postgres [local] SELECT
14320 be/4 postgres   71.65 M/s    0.00 B/s  0.00 % 33.70 % postgres: postgres postgres [local] SELECT
14328 be/4 postgres   71.67 M/s    0.00 B/s  0.00 % 33.60 % postgres: postgres postgres [local] SELECT

tpc-b 读写性能 (tps: 42058)

测试

pgbench -M prepared -v -r -P 1 -c 32 -j 32 -T 3600

结果

...............
progress: 3585.0 s, 38970.6 tps, lat 0.821 ms stddev 2.700
progress: 3586.0 s, 37586.5 tps, lat 0.851 ms stddev 3.968
progress: 3587.0 s, 42752.9 tps, lat 0.748 ms stddev 0.505
progress: 3588.0 s, 43598.0 tps, lat 0.734 ms stddev 0.459
progress: 3589.0 s, 43036.5 tps, lat 0.744 ms stddev 0.502
progress: 3590.0 s, 43520.4 tps, lat 0.735 ms stddev 0.483
progress: 3591.0 s, 37351.6 tps, lat 0.857 ms stddev 4.399
progress: 3592.0 s, 36243.5 tps, lat 0.883 ms stddev 4.897
progress: 3593.0 s, 43355.3 tps, lat 0.738 ms stddev 0.487
progress: 3594.0 s, 43592.2 tps, lat 0.734 ms stddev 0.486
progress: 3595.0 s, 44001.3 tps, lat 0.727 ms stddev 0.454
progress: 3596.0 s, 40706.8 tps, lat 0.786 ms stddev 2.690
progress: 3597.0 s, 38404.0 tps, lat 0.833 ms stddev 3.124
progress: 3598.0 s, 43741.3 tps, lat 0.732 ms stddev 0.470
progress: 3599.0 s, 39648.9 tps, lat 0.807 ms stddev 3.401
progress: 3600.0 s, 43696.8 tps, lat 0.731 ms stddev 0.458
transaction type: <builtin: TPC-B (sort of)>
scaling factor: 100000
query mode: prepared
number of clients: 32
number of threads: 32
duration: 3600 s
number of transactions actually processed: 151412741
latency average = 0.761 ms
latency stddev = 1.873 ms
tps = 42058.973764 (including connections establishing)
tps = 42059.153928 (excluding connections establishing)
statement latencies in milliseconds:  0.002  \set aid random(1, 100000 * :scale)  0.001  \set bid random(1, 1 * :scale)  0.001  \set tid random(1, 10 * :scale)  0.001  \set delta random(-5000, 5000)  0.029  BEGIN;  0.425  UPDATE pgbench_accounts SET abalance = abalance + :delta WHERE aid = :aid;  0.062  SELECT abalance FROM pgbench_accounts WHERE aid = :aid;  0.075  UPDATE pgbench_tellers SET tbalance = tbalance + :delta WHERE tid = :tid;  0.065  UPDATE pgbench_branches SET bbalance = bbalance + :delta WHERE bid = :bid;  0.052  INSERT INTO pgbench_history (tid, bid, aid, delta, mtime) VALUES (:tid, :bid, :aid, :delta, CURRENT_TIMESTAMP);  0.049  END;

IO监测

Total DISK READ :    2027.55 M/s | Total DISK WRITE :     661.01 M/s
Actual DISK READ:    2028.33 M/s | Actual DISK WRITE:     653.42 M/s  TID  PRIO  USER     DISK READ  DISK WRITE  SWAPIN     IO>    COMMAND
14545 be/4 postgres   59.30 M/s    9.82 M/s  0.00 % 26.14 % postgres: postgres postgres [local] UPDATE
14588 be/4 postgres   58.05 M/s   11.29 M/s  0.00 % 26.07 % postgres: postgres postgres [local] BINDTE
14574 be/4 postgres   62.22 M/s   10.33 M/s  0.00 % 25.95 % postgres: postgres postgres [local] UPDATE
14582 be/4 postgres   60.69 M/s   10.70 M/s  0.00 % 25.94 % postgres: postgres postgres [local] UPDATE
14561 be/4 postgres   56.00 M/s    9.60 M/s  0.00 % 25.77 % postgres: postgres postgres [local] idle
14567 be/4 postgres   70.98 M/s    9.82 M/s  0.00 % 25.70 % postgres: postgres postgres [local] UPDATE
14589 be/4 postgres   67.12 M/s    9.80 M/s  0.00 % 25.68 % postgres: postgres postgres [local] INSERT
14563 be/4 postgres   60.19 M/s    9.95 M/s  0.00 % 25.68 % postgres: postgres postgres [local] UPDATE
14583 be/4 postgres   69.52 M/s    9.87 M/s  0.00 % 25.60 % postgres: postgres postgres [local] UPDATE
14566 be/4 postgres   58.28 M/s    9.92 M/s  0.00 % 25.59 % postgres: postgres postgres [local] UPDATE
14587 be/4 postgres   54.09 M/s    9.86 M/s  0.00 % 25.57 % postgres: postgres postgres [local] UPDATE
14565 be/4 postgres   59.45 M/s    9.79 M/s  0.00 % 25.43 % postgres: postgres postgres [local] BINDRT
14580 be/4 postgres   71.12 M/s    9.96 M/s  0.00 % 25.32 % postgres: postgres postgres [local] SELECTn transaction
14576 be/4 postgres   64.08 M/s    9.54 M/s  0.00 % 25.20 % postgres: postgres postgres [local] UPDATE
14568 be/4 postgres   57.60 M/s   10.07 M/s  0.00 % 25.16 % postgres: postgres postgres [local] UPDATE
14584 be/4 postgres   63.87 M/s   10.05 M/s  0.00 % 25.13 % postgres: postgres postgres [local] UPDATE
14564 be/4 postgres   68.96 M/s    9.78 M/s  0.00 % 25.00 % postgres: postgres postgres [local] UPDATE
14550 be/4 postgres   68.82 M/s    9.85 M/s  0.00 % 24.94 % postgres: postgres postgres [local] UPDATE
14562 be/4 postgres   80.99 M/s   10.59 M/s  0.00 % 24.88 % postgres: postgres postgres [local] UPDATE
14581 be/4 postgres   59.95 M/s   11.24 M/s  0.00 % 24.85 % postgres: postgres postgres [local] UPDATE
14578 be/4 postgres   65.43 M/s   10.52 M/s  0.00 % 24.83 % postgres: postgres postgres [local] UPDATE
14570 be/4 postgres   61.71 M/s   10.06 M/s  0.00 % 24.51 % postgres: postgres postgres [local] UPDATE
14573 be/4 postgres   64.77 M/s   10.00 M/s  0.00 % 24.50 % postgres: postgres postgres [local] UPDATE
14575 be/4 postgres   61.35 M/s   10.01 M/s  0.00 % 24.40 % postgres: postgres postgres [local] UPDATE
14585 be/4 postgres   59.80 M/s   10.36 M/s  0.00 % 24.29 % postgres: postgres postgres [local] UPDATE
14560 be/4 postgres   74.76 M/s   11.13 M/s  0.00 % 24.09 % postgres: postgres postgres [local] idle in transaction
14577 be/4 postgres   58.25 M/s    9.75 M/s  0.00 % 23.98 % postgres: postgres postgres [local] idle in transaction
14579 be/4 postgres   72.02 M/s   10.09 M/s  0.00 % 23.94 % postgres: postgres postgres [local] UPDATE
14571 be/4 postgres   57.67 M/s   10.01 M/s  0.00 % 23.63 % postgres: postgres postgres [local] UPDATE
14572 be/4 postgres   55.61 M/s   10.20 M/s  0.00 % 23.36 % postgres: postgres postgres [local] UPDATE
14569 be/4 postgres   64.92 M/s   12.51 M/s  0.00 % 23.32 % postgres: postgres postgres [local] UPDATE
14586 be/4 postgres   59.95 M/s    9.79 M/s  0.00 % 23.29 % postgres: postgres postgres [local] UPDATE
13639 be/4 postgres   27.33 K/s   23.99 M/s  0.00 %  1.14 % postgres: walwriter
13638 be/4 postgres    0.00 B/s  214.99 M/s  0.00 %  0.00 % postgres: background writer
13637 be/4 postgres    0.00 B/s   95.76 M/s  0.00 %  0.00 % postgres: checkpointer

写测试后的数据大小

postgres=# \dt+ pgbench*  List of relations  Schema |       Name       | Type  |  Owner   |  Size   | Description
--------+------------------+-------+----------+---------+-------------  public | pgbench_accounts | table | postgres | 1264 GB |   public | pgbench_branches | table | postgres | 16 MB   |   public | pgbench_history  | table | postgres | 8701 MB |   public | pgbench_tellers  | table | postgres | 66 MB   |
(4 rows)  postgres=# \di+ pgbench*  List of relations  Schema |         Name          | Type  |  Owner   |      Table       |  Size   | Description
--------+-----------------------+-------+----------+------------------+---------+-------------  public | pgbench_accounts_pkey | index | postgres | pgbench_accounts | 209 GB  |   public | pgbench_branches_pkey | index | postgres | pgbench_branches | 8776 kB |   public | pgbench_tellers_pkey  | index | postgres | pgbench_tellers  | 43 MB   |
(3 rows)

100亿 tpbc 性能数据小结

1、100亿TPCB 初始化

耗时:8385秒。

速度:约119万行/s。

2、100亿TPCB 创建索引(24并行)

耗时:43分50秒。

速度:约380万行/s。

3、100亿TPCB 空间占用

表:1.251 TB

索引:209 GB

4、100亿TPCB 只读3600秒

TPS: 118053

QPS: 118053

5、100亿TPCB 读写3600秒

TPS: 42058

QPS: 210290

小结

PostgreSQL 在oltp, olap混合场景下有着出色的表现(100亿 TPCB 只读qps 11.8万,读写qps 21万 ; 1000万 TPCC:103万 tpmC; SF=10 TPCH 150秒,SF=200 TPCH 39 min;),再加上对Oracle兼容性的加持,以及阿里、平安集团、邮储、国家电网、中兴、华为、铁总、苏宁、去哪儿、探探等PG企业用户的背书,已成为市场去O的标志性替代产品,帮助企业实现几乎无痛的去O。

1、阿里云ADAM 去O 专版 PPAS(PostgreSQL高级版) 提供免费去O评估。

2、Ora2pg 开源去O产品.

3、《Oracle migration to Greenplum - (含 Ora2pg)》

参考

man pgbench

pgbench --help

《PostgreSQL 11 tpcc 测试(103万tpmC on ECS) - use sysbench-tpcc by Percona-Lab》

《(TPC-H测试 SF=10,SF=200) PostgreSQL 11 vs 10 vs Deepgreen》

大量PostgreSQL pgbench测试场景

PostgreSQL 11 100亿 tpcb 性能测试 on ECS相关推荐

  1. PostgreSQL 11 1000亿 tpcb、1000W tpcc 性能测试 - on 阿里云ECS + ESSD (含quorum based 0丢失多副本配置与性能测试)...

    标签 PostgreSQL , pgbench , tpcb , tpcc , tpch 背景 https://help.aliyun.com/knowledge_detail/64950.html ...

  2. PostgreSQL 11 1Kw TPCC , 1亿 TPCB 7*24 强压耐久测试

    标签 PostgreSQL , tpcc , tpcb 背景 TPCC, TPCB是工业标准的OLTP类型业务的数据库测试,包含大量的读.写.更新.删除操作. 7*24小时强压耐久测试,主要看数据库在 ...

  3. PostgreSQL 11 1Kw TPCC , 1亿 TPCB 7*24 强压耐久测试 1

    标签 PostgreSQL , tpcc , tpcb 背景 TPCC, TPCB是工业标准的OLTP类型业务的数据库测试,包含大量的读.写.更新.删除操作. 7*24小时强压耐久测试,主要看数据库在 ...

  4. PostgreSQL 11 tpcc 测试(103万tpmC on ECS) - use sysbench-tpcc by Percona-Lab

    标签 PostgreSQL , tpcc 背景 环境 阿里云虚拟机 [root@pg11-test ~]# lscpu Architecture: x86_64 CPU op-mode(s): 32- ...

  5. 腾讯云100亿元目标达成,发阳光普照奖iPhone 11 Pro,你酸了吗?

    12 月 19 日,鹅厂腾讯发钱的消息又像往年一样引来了一群柠檬精.除了微信支付团队获得 2 亿元奖金的消息之外,一张腾讯云团队每个员工奖励一部 iPhone 11 Pro 的 H5 页面截图也不断地 ...

  6. 96秒100亿!如何抗住双11高并发流量?

    点击上方"朱小厮的博客",选择"设为星标" 后台回复"加群"加入公众号专属技术群 来源:uee.me/c9UsN 今年双 11 全民购物狂欢 ...

  7. 第一百一十二期:96秒100亿!如何抗住双11高并发流量?

    今年双 11 全民购物狂欢节进入第十一个年头,1 分 36 秒,交易额冲到 100 亿 !比 2018 年快了近 30 秒,比 2017 年快了近 1 分半!这个速度再次刷新天猫双 11 成交总额破 ...

  8. 万字长文丨1分36秒,100亿,支付宝技术双11答卷:没有不可能

    2019年双11来了.1分36秒100亿,5分25秒超过300亿,12分49秒超500亿--如果没有双11,中国的互联网技术要发展到今天的水平,或许要再多花20年. 从双11诞生至今的11年里,有一个 ...

  9. 备战双 11!96秒,100亿,支付宝万级规模 K8s 集群管理系统如何设计?

    2019年双11来了.1分36秒100亿,5分25秒超过300亿,12分49秒超500亿--全球最大流量洪峰54.4万笔,阿里巴巴核心系统全部上云... 背后的技术支撑到底是怎么样的呢? 导读:Kub ...

最新文章

  1. 移动端框架如雨后春笋,你该如何选择?
  2. 技术社区,你真的会混吗?
  3. MFC VS2005 添加Override 和 Message
  4. 信息学奥赛一本通(C++)在线评测系统——基础(一)C++语言—— 1114:白细胞计数
  5. 菜鸟教程-Javascript学习笔记-JS函数之前
  6. Databricks:2015 Spark调查报告(部分译文)
  7. 多进程与多线程通信同步机制
  8. CRecordset类
  9. CPropertySheet 与CPropertyPage 的基本使用
  10. jQuery实现购物车多物品数量的加减+总价计算
  11. 华为捐赠欧拉 共建数字基础设施开源操作系统
  12. SSH登陆的原理分析和免登陆设置!
  13. sqlite dev破解
  14. 工行u盾显示316_工行U盾只显示金融@家怎么显密码
  15. matlab coder 安装,MATLAB Coder
  16. iptables之forward转发
  17. java freemarker转PDF和Word
  18. mac电脑的vue问题
  19. 用Linux开发板制作智能音箱,【工程师实战】只要几步,普通音箱秒变小度智能音箱...
  20. 【珍藏版】 2012Java开发工程师必备精品资料(115个)

热门文章

  1. SparkSQL(Spark-1.4.0)实战系列(一)——DataFrames基础
  2. Jetty开发指导:框架
  3. Stack(栈)和Heap(堆)的区别
  4. 如何记录网内用户访问了哪些网站
  5. mysql 社区版 阿里云_Mysql各版本介绍及下载
  6. java编写之jpg图片与base64编码之间的转换
  7. 20145309信息安全系统设计基础第3周学习总结
  8. 转载---虚拟机类加载机制
  9. spring文件下载记录
  10. data类型的Url格式及Base64编码:把小数据直接嵌入到Url中