top | item 15808317

Amazon DynamoDB Update – Global Tables and On-Demand Backup

102 points| polmolea | 8 years ago |aws.amazon.com

10 comments

order

johnvega|8 years ago

Looks like a direct competition to Microsoft Azure Cosmos DB. Anyone care to point out the top 3 differences?

sheraz|8 years ago

Maybe sorta kinda -- just a heads-up that CosmosDB can talk multiple API protocols:

  - MongoDB
  - Cassandra
  - Graph API
  - Azure tables
  - DocumentDB
I've used the MongoDB version, and it works ok for the basic CRUD stuff. I cant comment on aggregations, etc.

Not sure from a quick peek if DynamoDB does the same...

uji|8 years ago

It seems this service is targeted towards specific applications where write traffic is somewhat independent between 2 regions. However, you can shoot yourself in foot, if 2 regions are trying to update the same row, because of eventual consistency.

ramshanker|8 years ago

>> You can now create tables that are automatically replicated across two or more AWS Regions, with full support for multi-master writes, with a couple of clicks. <<

Hamm, If I understand it correctly, REGIONS are no longer heavily isolated.

seniorsassycat|8 years ago

Both features, global tables and automatic backups, seem like automation of features you could build yourself.

Global tables use DynamoDB Streams put items from Table A to Table B using conditional put (A Timestamp > B Timestamp).

Automated backup may be DynamoDB Streams or an automated setup of Datapipelines export to s3.

http://docs.aws.amazon.com/datapipeline/latest/DeveloperGuid...

dastbe|8 years ago

I would say this offers global capabilities while preserving regional isolation. you're composing the global table out of regional tables (which replicate between each other), and you're still using regional endpoints. this means that a failure with one region's table does not have impact on the other.

this model does mean that certain things, like conditional expressions, are less useful/useless because you could very well execute a conditional expression against stale date.