Performance.md

Main Docs

Performance

Some basic performance tests were done against bucket operations both remotely (North San Diego going to us-west-2) and within a Lambda function in the same region as the S3 bucket. The test code is available for anyone who is curious. Each test was run twice. It worth noting that the document size could dramatically affect the performance you encounter. These tests are admittedly primitive and aimed more to determine the performance of the s3-db code itself. More comprehensive tests will be done in the future for loading and saving of larger documents.

25 delete requests, of invalid records. bucket.delete()

Environment	Average	Median	Total
Remote	298.92	218	7486.5
Lambda	37.82	29	945.5

25 load requests, of the same document. bucket.load()

Environment	Average	Median	Total
Remote	221.5	215	5537.5
Lambda	28.4	22	710

25 save requests, of unique documents each time. bucket.save()

Environment	Average	Median	Total
Remote	221.5	215	6140.5
Lambda	59.14	40.5	1478.5

25 list requests, # of record pointers returned was between 70 and 100

Environment	Average	Median	Total
Remote	309.5	306.5	7747
Lambda	58.8	50	1472

Load every document on a bucket.list() request via .get() on each record

Environment	# of docs	Total Time	Average
Remote	85.5	21701.5	256.5
Lambda	98	3280.5	33.6

CRUD chain save(), modify, save(), reload(), delete()

Environment	Average Total Time
Remote	1369.5
Lambda	226