Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I tried using DynamoDB. It felt really unintuitive and like I had to bend my thought process to fit it.

I misconfigured some parameters and ended up with a big bill for a database that was doing pretty much nothing.

I don't see the upside of DynamoDB.



> I don't see the upside of DynamoDB.

Wide column, no-sql data store where you pay for provisioned throughput, and storage over 25GB.

All maintence is taken care by Amazon, no sharding, scale, patching, or tuning required.

If you need a non relational data store for use inside an AWS deployment, you would be hardpressed to find a better alternative.


> inside an AWS deployment

That's the key, isn't it?

I've been trying to wrap my head around DynamoDB for something that wouldn't necessarily live inside AWS and it seems like an awful lot of trouble.

But I guess if you're in the same AWS Region and you're doing high-volume stuff then the cost savings probably get compelling real fast.


> If you need a non relational data store for use inside an AWS deployment, you would be hardpressed to find a better alternative.

RDS Postgres supports HStore and JSON types.


To utilize it in a production environment, you would need:

1. have at minimum 2 instances 2. schedule backups 3. create and manage your db schema 4. manage users/roles

Admittedly RDS makes some of those tasks easy, but Dynamodb makes it even easier.

1. Create the table, provision read/write throughput 2. Create IAM role.


You don't do backups?


Good point.

You need to back up data from Dynamodb, but as a measure to protect against user caused data loss, not AWS infrastructure failure dataloss.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: