Write concurrency is much higher than default during restoration

[beyondbill]

I use dynamo-restore-from-s3 without setting -c. So by default the write currency should be 200.
However the actual write currency is as high as 1000!

It drops down gradually only because requests got throttled.

The version of dynamo-backup-to-s3 is 0.6.1

[beyondbill]

I dig a little bit into the code base and find that request concurrency and # of capacity units seem to be treated as the same thing here.
https://github.com/markitx/dynamo-backup-to-s3/blob/master/lib/dynamo-restore.js#L303

However based on Dynamo docs they are different.

One write capacity unit represents one write per second for an item up to 1 KB in size. If you need to write an item that is larger than 1 KB, DynamoDB will need to consume additional write capacity units. The total number of write capacity units required depends on the item size.

http://docs.aws.amazon.com/amazondynamodb/latest/developerguide/HowItWorks.ProvisionedThroughput.html

What if some of individual items in my Dynamo are larger than 1KB? Then writing each of them would consume more than 1 write capacity unit, although it just counts for 1 request.

It could explain why my case consumed much more write capacity units than 200.

[beyondbill]

Find something deeper in the code base.
https://github.com/markitx/dynamo-backup-to-s3/blob/master/lib/dynamo-restore.js#L172
It actually bundles 25 items together and then make 1 request using AWS API dynamo.batchWriteItem
https://github.com/markitx/dynamo-backup-to-s3/blob/master/lib/dynamo-restore.js#L299
This will well explain why request body size can easily exceed 1KB. On average one item in our DynamoDB should be easily over 1024/25 bytes