search

drizzle-team / drizzle-northwind-benchmarks-pg

Postgres benchmarks between Drizzle ORM and other popular ORM libraries
41 stars 6 forks source link

Can't run benchmark: Cannot connect to Postgres #1

Open mozeryansky opened 1 year ago

mozeryansky commented 1 year ago 
$ pnpm run start

> drizzle_pg_benchmarks@1.0.0 start /Users/michael/tmp/drizzle-northwind-benchmarks-pg
> pnpm run prepare:prisma && tsx src/common/benchmark

> drizzle_pg_benchmarks@1.0.0 prepare:prisma /Users/michael/tmp/drizzle-northwind-benchmarks-pg
> npx prisma generate --schema src/prisma/schema.prisma

Environment variables loaded from .env
Prisma schema loaded from src/prisma/schema.prisma
warn Preview feature "referentialIntegrity" is deprecated. The functionality can be used without specifying it as a preview feature.

✔ Generated Prisma Client (4.16.2 | library) to ./node_modules/@prisma/client in 187ms
You can now start using Prisma Client in your code. Reference: https://pris.ly/d/client
\```
import { PrismaClient } from '@prisma/client'
const prisma = new PrismaClient()
\```
(node:67457) ExperimentalWarning: Custom ESM Loaders is an experimental feature and might change at any time
(Use `node --trace-warnings ...` to show where the warning was created)
localhost postgres postgres postgres undefined
55002
node:internal/process/esm_loader:97
    internalBinding('errors').triggerUncaughtException(
                              ^

Error: connect ENOENT /var/run/docker.sock
    at __node_internal_captureLargerStackTrace (node:internal/errors:491:5)
    at __node_internal_exceptionWithHostPort (node:internal/errors:669:12)
    at PipeConnectWrap.afterConnect [as oncomplete] (node:net:1487:16) {
  errno: -2,
  code: 'ENOENT',
  syscall: 'connect',
  address: '/var/run/docker.sock'
}

Node.js v18.13.0
 ELIFECYCLE  Command failed with exit code 1.
mozeryansky commented 1 year ago

I didn't have Docker running, but I still can't get it to run. I started Docker and ran start again. On the second attempt there was a new error, which required me to stop/kill all the docker containers to try again.

Attemp 1:

~/tmp/drizzle-northwind-benchmarks-pg (main) $ pnpm run start                

> drizzle_pg_benchmarks@1.0.0 start /Users/michael/tmp/drizzle-northwind-benchmarks-pg
> pnpm run prepare:prisma && tsx src/common/benchmark

> drizzle_pg_benchmarks@1.0.0 prepare:prisma /Users/michael/tmp/drizzle-northwind-benchmarks-pg
> npx prisma generate --schema src/prisma/schema.prisma

Environment variables loaded from .env
Prisma schema loaded from src/prisma/schema.prisma
warn Preview feature "referentialIntegrity" is deprecated. The functionality can be used without specifying it as a preview feature.

✔ Generated Prisma Client (4.16.2 | library) to ./node_modules/@prisma/client in 200ms
You can now start using Prisma Client in your code. Reference: https://pris.ly/d/client
\```
import { PrismaClient } from '@prisma/client'
const prisma = new PrismaClient()
\```
(node:67871) ExperimentalWarning: Custom ESM Loaders is an experimental feature and might change at any time
(Use `node --trace-warnings ...` to show where the warning was created)
localhost postgres postgres postgres undefined
55002
Cannot connect to Postgres
/Users/michael/tmp/drizzle-northwind-benchmarks-pg/node_modules/pg/lib/client.js:132
      const error = this._ending ? new Error('Connection terminated') : new Error('Connection terminated unexpectedly')
                                                                        ^

Error: Connection terminated unexpectedly
    at Connection.<anonymous> (/Users/michael/tmp/drizzle-northwind-benchmarks-pg/node_modules/pg/lib/client.js:132:73)
    at Object.onceWrapper (node:events:627:28)
    at Connection.emit (node:events:513:28)
    at Connection.emit (node:domain:489:12)
    at Socket.<anonymous> (/Users/michael/tmp/drizzle-northwind-benchmarks-pg/node_modules/pg/lib/connection.js:107:12)
    at Socket.emit (node:events:525:35)
    at Socket.emit (node:domain:489:12)
    at endReadableNT (node:internal/streams/readable:1359:12)
    at process.processTicksAndRejections (node:internal/process/task_queues:82:21)

Node.js v18.13.0
 ELIFECYCLE  Command failed with exit code 1.
 ```
  Attempt 2:
  ```
~/tmp/drizzle-northwind-benchmarks-pg (main) $ pnpm run start

> drizzle_pg_benchmarks@1.0.0 start /Users/michael/tmp/drizzle-northwind-benchmarks-pg
> pnpm run prepare:prisma && tsx src/common/benchmark

> drizzle_pg_benchmarks@1.0.0 prepare:prisma /Users/michael/tmp/drizzle-northwind-benchmarks-pg
> npx prisma generate --schema src/prisma/schema.prisma

Environment variables loaded from .env
Prisma schema loaded from src/prisma/schema.prisma
warn Preview feature "referentialIntegrity" is deprecated. The functionality can be used without specifying it as a preview feature.

✔ Generated Prisma Client (4.16.2 | library) to ./node_modules/@prisma/client in 194ms
You can now start using Prisma Client in your code. Reference: https://pris.ly/d/client
\```
import { PrismaClient } from '@prisma/client'
const prisma = new PrismaClient()
\```
(node:67923) ExperimentalWarning: Custom ESM Loaders is an experimental feature and might change at any time
(Use `node --trace-warnings ...` to show where the warning was created)
localhost postgres postgres postgres undefined
55002
/Users/michael/tmp/drizzle-northwind-benchmarks-pg/node_modules/docker-modem/lib/modem.js:336
      var msg = new Error(
                ^

Error: (HTTP code 500) server error - driver failed programming external connectivity on endpoint benchmarks-tests-e208aeb2-0267-486e-87ef-94d4246ddacc (8cfa84eab1f8975de4eeb512520ed50c67fb32ea011d9633ebc8345f58c2d0bd): Bind for 0.0.0.0:55001 failed: port is already allocated 
    at <anonymous> (/Users/michael/tmp/drizzle-northwind-benchmarks-pg/node_modules/docker-modem/lib/modem.js:336:17)
    at getCause (/Users/michael/tmp/drizzle-northwind-benchmarks-pg/node_modules/docker-modem/lib/modem.js:366:7)
    at Object.Modem.buildPayload (/Users/michael/tmp/drizzle-northwind-benchmarks-pg/node_modules/docker-modem/lib/modem.js:335:5)
    at IncomingMessage.<anonymous> (/Users/michael/tmp/drizzle-northwind-benchmarks-pg/node_modules/docker-modem/lib/modem.js:303:16)
    at IncomingMessage.emit (node:events:525:35)
    at IncomingMessage.emit (node:domain:489:12)
    at endReadableNT (node:internal/streams/readable:1359:12)
    at process.processTicksAndRejections (node:internal/process/task_queues:82:21) {
  reason: 'server error',
  statusCode: 500,
  json: {
    message: 'driver failed programming external connectivity on endpoint benchmarks-tests-e208aeb2-0267-486e-87ef-94d4246ddacc (8cfa84eab1f8975de4eeb512520ed50c67fb32ea011d9633ebc8345f58c2d0bd): Bind for 0.0.0.0:55001 failed: port is already allocated'
  }
}

Node.js v18.13.0
 ELIFECYCLE  Command failed with exit code 1.
ImBIOS commented 1 year ago

@mozeryansky Check my fork in rqb-patch-1 or #2, I made it works.

mozeryansky commented 1 year ago

I still have the issue when using your branch:

$ pnpm run start
...
localhost postgres postgres postgres undefined
55002
Cannot connect to Postgres
...

And a different error on a second run:


$ pnpm run start
...
localhost postgres postgres postgres undefined
55002
/Users/michael/tmp/drizzle-northwind-benchmarks-pg/node_modules/.pnpm/docker-modem@3.0.8/node_modules/docker-modem/lib/modem.js:343
      var msg = new Error(
                ^
Error: (HTTP code 500) server error - driver failed programming external connectivity on endpoint benchmarks-tests-daf95ed0-b7f8-4c2f-8b25-8ffb1a357d92 (346eee8973941cd1be05d8a7efbba413e58e42ef4e9867a7031584a018549cb0): Bind for 0.0.0.0:55000 failed: port is already allocated
...
 ```
mozeryansky commented 1 year ago

I figured out the issue, it was because my Docker was set to 1 CPU. I wrote a test script to create one docker and connect/query and it worked fine, but after 3 it was failing. Increasing the timeout worked, so when I described the issue to ChatGPT it suggested increasing the CPU.

In my opinion this kind of benchmark is very unreliable, as my production server isn't running 9 concurrent PostgreSQL instances. Instead of testing a query against all ORMs, it should use one ORM against one database to test all queries. And preferably connect to a service like PlanetScale/Supabase to get real world metrics. The benchmark also does not cleanup the docker if there's an issue so I made this function for cleanup:

clean_docker () {
    docker rm -f $(docker ps -a -q)
    docker volume rm $(docker volume ls -q)
}

Below is the result with Docker using 8 CPUs and 8 Gb. Using docker stats shows it never exceeded 20% of cpu or memory, but the script does require all docker containers to be running concurrently.

cpu: Intel(R) Core(TM) i7-6920HQ CPU @ 2.90GHz
runtime: node v20.3.0 (x64-darwin)

benchmark      time (avg)             (min … max)       p75       p99      p995
------------------------------------------------- -----------------------------
• select * from customer
------------------------------------------------- -----------------------------
pg           1.67 ms/iter     (1.27 ms … 7.73 ms)   1.68 ms   4.48 ms    5.7 ms
pg:p         1.55 ms/iter     (1.18 ms … 2.97 ms)   1.62 ms   2.49 ms   2.83 ms
drizzle      1.82 ms/iter      (1.4 ms … 2.96 ms)   1.97 ms   2.84 ms   2.95 ms
drizzle:p    1.61 ms/iter     (1.21 ms … 4.11 ms)    1.7 ms   3.12 ms   3.55 ms
knex         1.54 ms/iter      (1.24 ms … 2.6 ms)   1.63 ms   2.13 ms    2.5 ms
kysely       1.54 ms/iter       (1.2 ms … 4.9 ms)   1.58 ms   3.29 ms   3.94 ms
mikro        3.37 ms/iter    (2.24 ms … 13.35 ms)   3.64 ms   6.69 ms  13.35 ms
typeorm      2.51 ms/iter      (1.9 ms … 3.98 ms)   2.73 ms   3.51 ms   3.54 ms
prisma          3 ms/iter     (2.49 ms … 5.68 ms)   3.08 ms   5.03 ms   5.68 ms

summary for select * from customer
  kysely
   1x faster than knex
   1x faster than pg:p
   1.04x faster than drizzle:p
   1.08x faster than pg
   1.18x faster than drizzle
   1.63x faster than typeorm
   1.95x faster than prisma
   2.19x faster than mikro

• select * from customer where id = ?
------------------------------------------------- -----------------------------
pg         110.52 ms/iter  (97.03 ms … 166.87 ms) 110.23 ms 166.87 ms 166.87 ms
pg:p       100.04 ms/iter  (87.12 ms … 131.83 ms)  104.9 ms 131.83 ms 131.83 ms
drizzle    116.65 ms/iter (100.84 ms … 140.89 ms) 130.88 ms 140.89 ms 140.89 ms
drizzle:p   96.69 ms/iter  (85.92 ms … 131.03 ms)  99.44 ms 131.03 ms 131.03 ms
knex       113.35 ms/iter   (94.9 ms … 142.57 ms) 117.51 ms 142.57 ms 142.57 ms
kysely     112.77 ms/iter  (99.12 ms … 162.77 ms) 116.84 ms 162.77 ms 162.77 ms
mikro       142.5 ms/iter  (110.8 ms … 207.53 ms) 142.45 ms 207.53 ms 207.53 ms
typeorm    134.22 ms/iter (113.84 ms … 180.22 ms) 135.81 ms 180.22 ms 180.22 ms
prisma     155.75 ms/iter (136.19 ms … 209.53 ms) 165.55 ms 209.53 ms 209.53 ms

summary for select * from customer where id = ?
  drizzle:p
   1.03x faster than pg:p
   1.14x faster than pg
   1.17x faster than kysely
   1.17x faster than knex
   1.21x faster than drizzle
   1.39x faster than typeorm
   1.47x faster than mikro
   1.61x faster than prisma

• select * from customer where company_name ilike ?
------------------------------------------------- -----------------------------
pg          64.26 ms/iter   (58.49 ms … 81.94 ms)  65.71 ms  81.94 ms  81.94 ms
pg:p        75.85 ms/iter   (55.33 ms … 302.2 ms)  61.39 ms  302.2 ms  302.2 ms
drizzle     69.19 ms/iter   (59.81 ms … 84.04 ms)  72.95 ms  84.04 ms  84.04 ms
drizzle:p   60.71 ms/iter    (54.9 ms … 77.14 ms)  61.34 ms  77.14 ms  77.14 ms
knex        68.49 ms/iter   (62.56 ms … 94.99 ms)  69.31 ms  94.99 ms  94.99 ms
kysely      68.96 ms/iter    (59.7 ms … 82.24 ms)  75.06 ms  82.24 ms  82.24 ms
mikro       79.95 ms/iter  (64.03 ms … 101.98 ms)  91.01 ms 101.98 ms 101.98 ms
typeorm     80.71 ms/iter  (69.07 ms … 103.55 ms)  86.02 ms 103.55 ms 103.55 ms
prisma     118.26 ms/iter  (86.28 ms … 168.31 ms) 132.29 ms 168.31 ms 168.31 ms

summary for select * from customer where company_name ilike ?
  drizzle:p
   1.06x faster than pg
   1.13x faster than knex
   1.14x faster than kysely
   1.14x faster than drizzle
   1.25x faster than pg:p
   1.32x faster than mikro
   1.33x faster than typeorm
   1.95x faster than prisma

• "SELECT * FROM employee"
------------------------------------------------- -----------------------------
pg           2.63 ms/iter     (1.09 ms … 9.98 ms)   3.18 ms   9.41 ms    9.7 ms
pg:p         2.16 ms/iter    (1.16 ms … 41.51 ms)   2.18 ms   6.77 ms   7.07 ms
drizzle      4.23 ms/iter    (1.71 ms … 23.04 ms)   4.45 ms  22.93 ms  23.04 ms
drizzle:p    2.93 ms/iter    (1.24 ms … 38.54 ms)   2.93 ms  18.74 ms  38.54 ms
knex         3.14 ms/iter     (1.59 ms … 14.2 ms)   3.07 ms  13.51 ms   14.2 ms
kysely       3.69 ms/iter     (1.6 ms … 20.19 ms)   3.74 ms  18.56 ms  20.19 ms
mikro        4.24 ms/iter     (2.28 ms … 12.5 ms)   4.63 ms   9.54 ms   12.5 ms
typeorm      3.56 ms/iter    (1.68 ms … 14.61 ms)   3.89 ms  13.85 ms  14.61 ms
prisma          3 ms/iter        (1.62 ms … 8 ms)    3.6 ms   6.22 ms      8 ms

summary for "SELECT * FROM employee"
  pg:p
   1.21x faster than pg
   1.36x faster than drizzle:p
   1.39x faster than prisma
   1.45x faster than knex
   1.64x faster than typeorm
   1.71x faster than kysely
   1.96x faster than drizzle
   1.96x faster than mikro

• select * from employee where id = ? left join reportee
------------------------------------------------- -----------------------------
pg          20.61 ms/iter  (10.86 ms … 201.78 ms)  17.64 ms 201.78 ms 201.78 ms
pg:p        15.03 ms/iter    (9.17 ms … 32.84 ms)  16.44 ms  32.84 ms  32.84 ms
drizzle     25.88 ms/iter   (15.55 ms … 51.81 ms)  29.75 ms  51.81 ms  51.81 ms
drizzle:p   14.11 ms/iter    (9.35 ms … 31.65 ms)   15.4 ms  31.65 ms  31.65 ms
knex        37.65 ms/iter   (21.24 ms … 70.64 ms)  42.44 ms  70.64 ms  70.64 ms
kysely      21.37 ms/iter      (14.66 ms … 41 ms)  24.06 ms     41 ms     41 ms
mikro       20.21 ms/iter    (14.7 ms … 30.04 ms)  21.61 ms  30.04 ms  30.04 ms
typeorm     36.17 ms/iter   (29.89 ms … 48.37 ms)  36.85 ms  48.37 ms  48.37 ms
prisma      33.81 ms/iter   (25.45 ms … 54.22 ms)  37.61 ms  54.22 ms  54.22 ms

summary for select * from employee where id = ? left join reportee
  drizzle:p
   1.07x faster than pg:p
   1.43x faster than mikro
   1.46x faster than pg
   1.52x faster than kysely
   1.83x faster than drizzle
   2.4x faster than prisma
   2.56x faster than typeorm
   2.67x faster than knex

• SELECT * FROM supplier
------------------------------------------------- -----------------------------
pg           1.58 ms/iter     (1.03 ms … 5.76 ms)   1.79 ms   3.44 ms   4.34 ms
pg:p         1.21 ms/iter   (927.62 µs … 2.12 ms)   1.28 ms    1.8 ms   1.87 ms
drizzle      1.69 ms/iter     (1.02 ms … 5.14 ms)   1.92 ms   3.54 ms   4.19 ms
drizzle:p    1.41 ms/iter    (994.6 µs … 3.37 ms)   1.51 ms   2.86 ms   2.92 ms
knex         1.82 ms/iter      (1.07 ms … 3.8 ms)   2.03 ms   3.13 ms   3.56 ms
kysely       1.49 ms/iter     (1.09 ms … 3.16 ms)   1.63 ms   2.71 ms   2.92 ms
mikro        2.72 ms/iter     (1.66 ms … 6.72 ms)   3.05 ms   5.87 ms   6.72 ms
typeorm       2.3 ms/iter    (1.45 ms … 17.63 ms)   2.37 ms  13.88 ms  16.33 ms
prisma        2.1 ms/iter     (1.65 ms … 3.34 ms)   2.19 ms   3.08 ms   3.33 ms

summary for SELECT * FROM supplier
  pg:p
   1.17x faster than drizzle:p
   1.23x faster than kysely
   1.3x faster than pg
   1.4x faster than drizzle
   1.5x faster than knex
   1.74x faster than prisma
   1.9x faster than typeorm
   2.25x faster than mikro

• select * from supplier where id = ?
------------------------------------------------- -----------------------------
pg          47.05 ms/iter   (40.22 ms … 56.69 ms)  49.12 ms  56.69 ms  56.69 ms
pg:p        37.91 ms/iter   (30.36 ms … 63.37 ms)  39.15 ms  63.37 ms  63.37 ms
drizzle     43.07 ms/iter      (34 ms … 59.64 ms)  50.42 ms  59.64 ms  59.64 ms
drizzle:p   33.25 ms/iter   (28.91 ms … 44.66 ms)  36.28 ms  44.66 ms  44.66 ms
knex        40.31 ms/iter   (33.98 ms … 59.64 ms)   44.5 ms  59.64 ms  59.64 ms
kysely      43.19 ms/iter   (34.95 ms … 62.04 ms)  47.75 ms  62.04 ms  62.04 ms
mikro       54.98 ms/iter   (43.59 ms … 74.89 ms)  59.37 ms  74.89 ms  74.89 ms
typeorm     52.61 ms/iter   (40.28 ms … 77.23 ms)  55.34 ms  77.23 ms  77.23 ms
prisma      53.17 ms/iter   (48.11 ms … 72.49 ms)  56.24 ms  72.49 ms  72.49 ms

summary for select * from supplier where id = ?
  drizzle:p
   1.14x faster than pg:p
   1.21x faster than knex
   1.3x faster than drizzle
   1.3x faster than kysely
   1.41x faster than pg
   1.58x faster than typeorm
   1.6x faster than prisma
   1.65x faster than mikro

• SELECT * FROM product
------------------------------------------------- -----------------------------
pg           1.57 ms/iter     (1.17 ms … 2.61 ms)   1.67 ms   2.47 ms   2.53 ms
pg:p         1.89 ms/iter       (1.2 ms … 5.1 ms)   2.04 ms   3.92 ms   3.95 ms
drizzle      2.06 ms/iter     (1.35 ms … 3.57 ms)   2.34 ms   3.27 ms   3.44 ms
drizzle:p    1.81 ms/iter     (1.24 ms … 6.35 ms)   1.95 ms   4.73 ms   5.69 ms
knex         1.71 ms/iter     (1.31 ms … 3.18 ms)   1.83 ms   2.92 ms   3.03 ms
kysely        1.7 ms/iter     (1.15 ms … 4.45 ms)   1.87 ms   2.92 ms   3.06 ms
mikro        3.52 ms/iter     (2.71 ms … 8.77 ms)    3.6 ms   8.32 ms   8.77 ms
typeorm      1.94 ms/iter     (1.67 ms … 3.92 ms)   1.98 ms   3.04 ms   3.46 ms
prisma          3 ms/iter     (2.37 ms … 7.09 ms)   3.24 ms   4.82 ms   7.09 ms

summary for SELECT * FROM product
  pg
   1.08x faster than kysely
   1.09x faster than knex
   1.15x faster than drizzle:p
   1.21x faster than pg:p
   1.24x faster than typeorm
   1.32x faster than drizzle
   1.92x faster than prisma
   2.25x faster than mikro

• SELECT * FROM product LEFT JOIN supplier WHERE product.id = ?
------------------------------------------------- -----------------------------
pg         106.66 ms/iter  (93.68 ms … 132.87 ms) 110.18 ms 132.87 ms 132.87 ms
pg:p        82.11 ms/iter  (71.34 ms … 104.12 ms)  82.54 ms 104.12 ms 104.12 ms
drizzle    131.79 ms/iter (105.26 ms … 227.54 ms) 135.11 ms 227.54 ms 227.54 ms
drizzle:p   95.85 ms/iter  (80.99 ms … 229.67 ms)  88.87 ms 229.67 ms 229.67 ms
knex       110.56 ms/iter (100.48 ms … 129.08 ms) 120.56 ms 129.08 ms 129.08 ms
kysely     112.98 ms/iter (104.18 ms … 136.33 ms) 114.82 ms 136.33 ms 136.33 ms
mikro      169.56 ms/iter (150.86 ms … 213.16 ms) 176.32 ms 213.16 ms 213.16 ms
typeorm    319.25 ms/iter  (289.33 ms … 348.8 ms) 329.27 ms  348.8 ms  348.8 ms
prisma     227.01 ms/iter (192.04 ms … 268.14 ms) 262.87 ms 268.14 ms 268.14 ms

summary for SELECT * FROM product LEFT JOIN supplier WHERE product.id = ?
  pg:p
   1.17x faster than drizzle:p
   1.3x faster than pg
   1.35x faster than knex
   1.38x faster than kysely
   1.6x faster than drizzle
   2.06x faster than mikro
   2.76x faster than prisma
   3.89x faster than typeorm

• SELECT * FROM product WHERE product.name ILIKE ?
------------------------------------------------- -----------------------------
pg          70.14 ms/iter    (61.23 ms … 95.7 ms)  74.26 ms   95.7 ms   95.7 ms
pg:p        58.24 ms/iter   (52.68 ms … 70.28 ms)  60.08 ms  70.28 ms  70.28 ms
drizzle     74.32 ms/iter   (65.93 ms … 93.31 ms)  76.15 ms  93.31 ms  93.31 ms
drizzle:p   68.69 ms/iter   (56.68 ms … 94.16 ms)  77.28 ms  94.16 ms  94.16 ms
knex       121.17 ms/iter  (75.15 ms … 217.72 ms) 144.33 ms 217.72 ms 217.72 ms
kysely     121.98 ms/iter  (79.12 ms … 185.13 ms) 137.16 ms 185.13 ms 185.13 ms
mikro       117.7 ms/iter   (91.6 ms … 183.02 ms) 131.39 ms 183.02 ms 183.02 ms
typeorm    101.25 ms/iter  (77.52 ms … 135.57 ms) 116.96 ms 135.57 ms 135.57 ms
prisma     120.28 ms/iter   (94.5 ms … 211.39 ms)  127.1 ms 211.39 ms 211.39 ms

summary for SELECT * FROM product WHERE product.name ILIKE ?
  pg:p
   1.18x faster than drizzle:p
   1.2x faster than pg
   1.28x faster than drizzle
   1.74x faster than typeorm
   2.02x faster than mikro
   2.07x faster than prisma
   2.08x faster than knex
   2.09x faster than kysely

• select all order with sum and count
------------------------------------------------- -----------------------------
pg          10.88 ms/iter    (6.43 ms … 29.24 ms)  12.27 ms  29.24 ms  29.24 ms
pg:p         8.27 ms/iter    (6.13 ms … 14.17 ms)   8.86 ms  14.17 ms  14.17 ms
drizzle      8.95 ms/iter    (7.42 ms … 12.62 ms)   9.77 ms  12.62 ms  12.62 ms
drizzle:p   10.07 ms/iter    (7.32 ms … 14.91 ms)  11.42 ms  14.91 ms  14.91 ms
knex         9.24 ms/iter    (7.12 ms … 19.49 ms)   9.67 ms  19.49 ms  19.49 ms
kysely       9.38 ms/iter    (6.58 ms … 23.09 ms)  10.28 ms  23.09 ms  23.09 ms
mikro      274.13 ms/iter (239.21 ms … 292.57 ms) 290.49 ms 292.57 ms 292.57 ms
typeorm     38.94 ms/iter   (31.35 ms … 63.78 ms)  39.04 ms  63.78 ms  63.78 ms
prisma      75.38 ms/iter     (64.5 ms … 94.8 ms)  79.84 ms   94.8 ms   94.8 ms

summary for select all order with sum and count
  pg:p
   1.08x faster than drizzle
   1.12x faster than knex
   1.13x faster than kysely
   1.22x faster than drizzle:p
   1.31x faster than pg
   4.71x faster than typeorm
   9.11x faster than prisma
   33.13x faster than mikro

• select order with sum and count using limit with offset
------------------------------------------------- -----------------------------
pg         124.35 ms/iter (100.55 ms … 209.31 ms) 125.77 ms 209.31 ms 209.31 ms
pg:p        94.21 ms/iter  (87.36 ms … 118.15 ms)  92.23 ms 118.15 ms 118.15 ms
drizzle     100.7 ms/iter  (95.27 ms … 119.61 ms) 101.18 ms 119.61 ms 119.61 ms
drizle:p    91.89 ms/iter  (82.76 ms … 150.41 ms)  86.71 ms 150.41 ms 150.41 ms
knex        99.79 ms/iter  (92.62 ms … 112.78 ms) 105.24 ms 112.78 ms 112.78 ms
kysely      97.81 ms/iter  (91.21 ms … 109.63 ms)  99.76 ms 109.63 ms 109.63 ms
mikro       177.4 ms/iter (170.34 ms … 188.28 ms) 180.38 ms 188.28 ms 188.28 ms
typeorm    158.91 ms/iter (148.38 ms … 192.19 ms) 160.47 ms 192.19 ms 192.19 ms
prisma     125.63 ms/iter (113.01 ms … 210.64 ms) 118.05 ms 210.64 ms 210.64 ms

summary for select order with sum and count using limit with offset
  drizle:p
   1.03x faster than pg:p
   1.06x faster than kysely
   1.09x faster than knex
   1.1x faster than drizzle
   1.35x faster than pg
   1.37x faster than prisma
   1.73x faster than typeorm
   1.93x faster than mikro

• select order where order.id = ? with sum and count
------------------------------------------------- -----------------------------
pg         156.77 ms/iter     (82.99 ms … 1.12 s)  96.39 ms    1.12 s    1.12 s
pg:p        85.93 ms/iter  (70.47 ms … 165.75 ms)  86.69 ms 165.75 ms 165.75 ms
drizzle     118.6 ms/iter (100.03 ms … 146.16 ms)  127.3 ms 146.16 ms 146.16 ms
drizzle:p  117.49 ms/iter     (91 ms … 152.79 ms) 135.43 ms 152.79 ms 152.79 ms
knex       152.52 ms/iter   (94.64 ms … 241.5 ms) 168.67 ms  241.5 ms  241.5 ms
kysely     159.76 ms/iter  (105.5 ms … 265.44 ms) 173.05 ms 265.44 ms 265.44 ms
mikro      320.03 ms/iter (216.58 ms … 436.04 ms) 348.77 ms 436.04 ms 436.04 ms
prisma     251.59 ms/iter    (160.39 ms … 1.12 s) 190.84 ms    1.12 s    1.12 s
typeorm    360.08 ms/iter    (217.89 ms … 1.25 s) 341.25 ms    1.25 s    1.25 s

summary for select order where order.id = ? with sum and count
  pg:p
   1.37x faster than drizzle:p
   1.38x faster than drizzle
   1.77x faster than knex
   1.82x faster than pg
   1.86x faster than kysely
   2.93x faster than prisma
   3.72x faster than mikro
   4.19x faster than typeorm

• SELECT * FROM order_detail WHERE order_id = ?
------------------------------------------------- -----------------------------
pg         407.22 ms/iter (352.35 ms … 486.95 ms) 418.14 ms 486.95 ms 486.95 ms
pg:p       330.17 ms/iter  (281.75 ms … 381.4 ms) 351.11 ms  381.4 ms  381.4 ms
drizzle    447.76 ms/iter  (400.6 ms … 505.14 ms) 481.09 ms 505.14 ms 505.14 ms
drizzle:p  416.55 ms/iter    (266.15 ms … 1.28 s) 365.39 ms    1.28 s    1.28 s
knex       456.09 ms/iter  (397.44 ms … 535.1 ms) 508.03 ms  535.1 ms  535.1 ms
kysely     437.95 ms/iter (382.76 ms … 594.61 ms) 440.22 ms 594.61 ms 594.61 ms
mikro      885.97 ms/iter (841.21 ms … 957.26 ms) 917.52 ms 957.26 ms 957.26 ms
typeorm    482.95 ms/iter (454.86 ms … 527.15 ms) 495.79 ms 527.15 ms 527.15 ms
prisma     943.06 ms/iter    (870.56 ms … 1.02 s) 992.96 ms    1.02 s    1.02 s

summary for SELECT * FROM order_detail WHERE order_id = ?
  pg:p
   1.23x faster than pg
   1.26x faster than drizzle:p
   1.33x faster than kysely
   1.36x faster than drizzle
   1.38x faster than knex
   1.46x faster than typeorm
   2.68x faster than mikro
   2.86x faster than prisma

I would like to keep this issue open, as I suggest there should be at least a readme change for the minimum CPU requirements, and a way to use a remote url instead of docker. Its fine if someone from drizzle would like to close it.