Hi Owen
AWS keyspaces based on Apache Cassandra is ment for huge dataset
in terrabytes.
Strat ML initial dataset is about 250MB. It is kind of overkill
to go for AWS keyspaces.
Regarding DynamoDB, it is based on throughout coming out of
database.
For provisioned IO and 1MB throughput / sec of writing, about
$0.65 and 3MB throughput / sec of reading about $0.65
In nutshell, if we are using indexes on specific columns, then
cost will be miniscule.
But index are limited per table.
If you want to search each and every field, then it will scan
whole database which means too many read IOs needed
To avoid whole database scan, indexes is only way out and loading
same data into multiple tables in the only solution.
So we will have query multiple tables to find the data from
indexes to save cost.
It is sort of manipulation.
Biggest advantage of Dynamo DB is that you can scale like
anything and it might turn out cost to be near zero for low
volumes of searches. So your application is up all the time
without paying any money if volumes are low.
I have not used Dynamo DB in real life situation to give justice
if Dynamo DB will charge a bomb or is very cost effective. So you
might look for alternate opinion.
Theortically I know most of the Dynamo DB concepts as I am
certified as AWS certified developer associate and AWS certified
sys ops associate.
Naval
On 23/09/22 1:41 am, Owen Ambur wrote: