yueyoum / bulk_create_test

Benchmark for this

The expected behavior:

Use the imported data totally. Not keep old data.

The are two way to do this:

1. override before_import and get_or_init_instance.
2. using the new option: bulk_repace

in file myapp/admin.py,

ResourceTestModel_1 is the first way.

ResourceTestModel_2 is the second way.

Benchmark

TestModel-2016-06-20.xls is the test data which contains 10K rows.

using mysql database. (pymysql as driver).

this is the runserver log:

1. First way

[TIME MEASURE] /admin/myapp/testmodel/import/: 11.0680501461
[20/Jun/2016 10:16:32] "POST /admin/myapp/testmodel/import/ HTTP/1.1" 200 5599231
[TIME MEASURE] /admin/myapp/testmodel/process_import/: 96.4507160187
[20/Jun/2016 10:18:20] "POST /admin/myapp/testmodel/process_import/ HTTP/1.1" 302 0
[TIME MEASURE] /admin/myapp/testmodel/: 0.127159118652
[20/Jun/2016 10:18:20] "GET /admin/myapp/testmodel/ HTTP/1.1" 200 46114

Total cost 107 seconds.

2. Second way

[TIME MEASURE] /admin/myapp/testmodel/import/: 1.76946806908
[20/Jun/2016 10:19:20] "POST /admin/myapp/testmodel/import/ HTTP/1.1" 302 0
[TIME MEASURE] /admin/myapp/testmodel/: 0.116591215134
[20/Jun/2016 10:19:20] "GET /admin/myapp/testmodel/ HTTP/1.1" 200 45954

Total cost 1.8 seconds.

The way 2 (using bulk_replace option) is about 60 times faster than way 1.