Bulk inserting performance

[paradis-A]

OS: Windows 10 64bit
RAM: 8gb
STORAGE: 128gb SSD M.2
CPU: Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz 1.99 GHz
---------------------------------First run---------------------------------
Insert Non-indexed
√ (767ms)

Insert Indexed
√ (4727ms)

Get count of child (non-indexed)
Childs: 2000
√ (345ms)

Get count of child (indexed)
Childs: 2000
√ (39ms)

Non-Indexed instance.ref Foreach performance
√ (800ms)

Indexed instance.ref Foreach performance
√ (740ms)

Non-Indexed Query Foreach performance
√ (924ms)

Indexed Query Foreach performance
√ (676ms)

8 passing (9s)

---------------------------------Second run---------------------------------
Insert Non-indexed
√ (13152ms)

Insert Indexed
√ (23780ms)

Get count of child (non-indexed)
Childs: 4000
√ (667ms)

Get count of child (indexed)
Childs: 4000
√ (87ms)

Non-Indexed instance.ref Foreach performance
√ (1434ms)

Indexed instance.ref Foreach performance
√ (1369ms)

Non-Indexed Query Foreach performance
√ (1644ms)

Indexed Query Foreach performance
√ (1261ms)

8 passing (43s)

---------------------------------Thrid run---------------------------------
Insert Non-indexed
√ (15832ms)

Insert Indexed
√ (29385ms)

Get count of child (non-indexed)
Childs: 6000
√ (888ms)

Get count of child (indexed)
Childs: 6000
√ (100ms)

Non-Indexed instance.ref Foreach performance
√ (2238ms)

Indexed instance.ref Foreach performance
√ (2083ms)

Non-Indexed Query Foreach performance
√ (2672ms)

Indexed Query Foreach performance
√ (2007ms)

8 passing (55s)

Does update works like transaction where it loads all props then does the update stuff?

See Test Code

[appy-one]

I've been working on improving the performance past week, almost ready to commit - one issue still needs to be ironed out. I wonder what your stats will look after those updates!

[paradis-A]

Just wow! I wish I can donate. Just am still seeking fortune out of this database hopefully more people your work when i'm ready to sell my products (sitting on top of this -Acebase-) in the market. Goodluck us!

[appy-one]

I've updated to version 1.15.0, you should notice significant performance improvements on these tests. Can you update and run them again? Would love to see new results 😄

[paradis-A]

you rock! it actually does

-----------first run------------
Insert Non-indexed
√ (770ms)

Insert Indexed
[perf] tree rebuild took 5ms
[perf] getLeafStartKeys: 1ms
[perf] getEntries: 1ms
[perf] tree.create: 4ms
[perf] tree rebuild took 5ms
[perf] getLeafStartKeys: 1ms
[perf] getEntries: 0ms
[perf] tree.create: 5ms
[perf] tree rebuild took 12ms
[perf] getLeafStartKeys: 0ms
[perf] getEntries: 3ms
[perf] tree.create: 12ms
[perf] tree rebuild took 13ms
[perf] getLeafStartKeys: 0ms
[perf] getEntries: 4ms
[perf] tree.create: 13ms
[perf] tree rebuild took 22ms
[perf] getLeafStartKeys: 0ms
[perf] getEntries: 6ms
[perf] tree.create: 22ms
√ (3783ms)

Get count of child (non-indexed)
Childs: 2000
√ (331ms)

Get count of child (indexed)
Childs: 2000
√

Non-Indexed instance.ref Foreach performance
√ (737ms)

Indexed instance.ref Foreach performance
√ (776ms)

Non-Indexed Query Foreach performance
√ (818ms)

Indexed Query Foreach performance
√ (657ms)

8 passing (8s)

---------second run-----------
Insert Non-indexed
√ (936ms)

Insert Indexed
[perf] tree rebuild took 32ms
[perf] getLeafStartKeys: 0ms
[perf] getEntries: 7ms
[perf] tree.create: 32ms
[perf] tree rebuild took 46ms
[perf] getLeafStartKeys: 0ms
[perf] getEntries: 10ms
[perf] tree.create: 46ms
√ (3908ms)

Get count of child (non-indexed)
Childs: 3994
√ (591ms)

Get count of child (indexed)
Childs: 4000
√ (71ms)

Non-Indexed instance.ref Foreach performance
√ (1383ms)

Indexed instance.ref Foreach performance
√ (1428ms)

Non-Indexed Query Foreach performance
√ (1763ms)

Indexed Query Foreach performance
√ (1327ms)

8 passing (11s)

---------third run-------------
Insert Non-indexed
√ (1021ms)

Insert Indexed
[perf] tree rebuild took 97ms
[perf] getLeafStartKeys: 1ms
[perf] getEntries: 17ms
[perf] tree.create: 97ms
√ (4600ms)

Get count of child (non-indexed)
Childs: 5990
√ (871ms)

Get count of child (indexed)
Childs: 6000
√ (98ms)

Non-Indexed instance.ref Foreach performance
√ (2072ms)

Indexed instance.ref Foreach performance
√ (2123ms)

Non-Indexed Query Foreach performance
√ (2649ms)

Indexed Query Foreach performance
√ (1985ms)

8 passing (15s)

[paradis-A]

Excluding the query perfs. The bulk insert with or without index improved. Unlike before that it was like proportional based on the stored collection + inserts. Repetition of Bulk inserting now with the same amount (2000 in my case) improved.

Bulk Insert Before * 3 reps
1st Run

• Non-indexed (767ms)
• Indexed (4727ms)
  2nd Run
• Non-indexed (13152ms)
• Indexed (23780ms)
  3rd Run
• Non-indexed (15832ms)
• Indexed (29385ms)

Bulk Inser @v1.15.0 * 3 reps
1st Run

• Non-indexed (770ms)
• Indexed (3783ms)
  2nd Run
• Non-indexed (936ms)
• Indexed (3908ms)
  3rd Run
• Non-indexed (1021ms)
• Indexed (4600ms)