问题
I have a user object with these attributes.
id (key), name and email
And I am trying to make sure these attributes are unique in the DB.
How do I prevent a put/create/save operation from succeeding, in case of either of the non-key attributes, email and name, already exists in the DB?
I have a table, tblUsers with one key-attribute being the id.
I then have two globally secondary indexes, each with also one key-attribute, being the email for the first index-table, and name for the second.
I am using microsoft .net identity framework, which itself checks for existing users with a given name or email, before creating a user.
The problem I forsee, is the delay between checking for existing users and creating a new. There is no safety, that multiple threads wont end up creating two users with the same name or email.
回答1:
dynamodb can force uniqueness only for hash-range table keys (not for global secondary index keys)
do in your case there are 2 options:
1) force it on application level - if your problem is safety, then use locks (cache locks)
2) dont use dynamodb (maybe its not answer your requirements )
回答2:
I am using ehcache locally to check for duplicates . Adding one more check if ehcache is empty (for some reason ,cache has been reset ) .repopulate cache by making query to dynamoDb .
来源：https://stackoverflow.com/questions/35387768/dynamodb-avoid-duplicate-non-key-attributes