Estimated time for using DDB Scans to fetch a non partition key

Question

I have a DDB where data is stored in the format:
{
 "GpId": "XXXXXX",
 "AnciIds": [
  {
   "Domain": "FCOpsOrderId",
   "Id": "XXXXXXXX"
  }
 ],
 "GpIdDom": "FCOpsShipmentId",
 "InDate": "XXXXXXXX",
 "LineIts": [
  {
   "Amt": {
    "amt": 293.33,
    "unt": "INR"
   },
   "ByrId": {
    "Domain": "MerchantId",
    "Id": "XXXXXXX"
   },
   "SlrId": {
    "Domain": "MerchantId",
    "Id": "XXXXXXX"
   }
  }
 ]
}

I want to delete MerchantId which is contained under LineIts which is not the partition key. If we use DDB scan to filter the data by selecting the attribute name as LineIts, Condition as Contains and Value as the Id I want to delete, then could you tell me what would be the approximate time taken by this operation considering the table size is 344.8 gigabytes? By how much would the efficiency increase if I use parallel Scans?

Answer

That question is borderline impossible to answer, as it depends on factors such as your operating system, host CPU/RAM etc....

32MB/s is the rate at which DynamoDB can return data approx. So about 3 hours for a sequential Scan.

What I can answer is that it will be significantly faster using parallel scan, but again that assumes you are running from a machine thats capable of spanning many threads.

Estimated time for using DDB Scans to fetch a non partition key

Relevant content