Repo containing some experiments in azure table storage, mainly testing batch insert performance using various methods.
elapsed | eps | function | partitionSize | partitionCount |
---|---|---|---|---|
2.91963 | 342.509 | batch_upsert | n/a | n/a |
15.9524 | 62.6867 | batch_upsert_partitioned | 100 | 100 |
28.7108 | 34.83 | batch_upsert_partitioned | 100 | 200 |
66.311 | 15.0805 | batch_upsert_partitioned | 100 | 500 |
134.702 | 7.42379 | batch_upsert_partitioned | 100 | 1000 |
82.2025 | 12.1651 | batch_upsert_partitioned | 100 | 2000 |
87.4439 | 11.4359 | batch_upsert_partitioned | 100 | 2500 |
83.6264 | 11.9579 | batch_upsert_partitioned | 100 | 5000 |
3.55084 | 281.623 | batch_upsert_partitioned_async | 100 | 100 |
5.05056 | 197.998 | batch_upsert_partitioned_async | 100 | 200 |
8.34641 | 119.812 | batch_upsert_partitioned_async | 100 | 500 |
14.7585 | 67.7576 | batch_upsert_partitioned_async | 100 | 1000 |
14.1735 | 70.5541 | batch_upsert_partitioned_async | 100 | 2000 |
14.6752 | 68.1423 | batch_upsert_partitioned_async | 100 | 2500 |
13.6486 | 73.2674 | batch_upsert_partitioned_async | 100 | 5000 |
0.145516 | 6872.1 | batch_upsert_partitioned_parallel | 100 | 100 |
0.151052 | 6620.24 | batch_upsert_partitioned_parallel | 100 | 200 |
0.408113 | 2450.3 | batch_upsert_partitioned_parallel | 100 | 500 |
0.583888 | 1712.66 | batch_upsert_partitioned_parallel | 100 | 1000 |
0.679692 | 1471.25 | batch_upsert_partitioned_parallel | 100 | 2000 |
1.09833 | 910.472 | batch_upsert_partitioned_parallel | 100 | 2500 |
1.08817 | 918.976 | batch_upsert_partitioned_parallel | 100 | 5000 |
Standard batch insert, no partitioning.
Batch insert, partitioned into partitionCount
partitions.
Async batch insert, partitioned into partitionCount
partitions. Theoretically this should be faster than the synchronous batch_upsert_partitioned
method.
Parallelised batch insert, one process per partition.