[MartinMond]

> kvpbase also allows you to program it - tell it what to do when data is read, written, or deleted - so you can offload heavy tasks from your app.

Isn't that exactly what "every database ever" calls triggers?

From the Postgres docs: http://www.postgresql.org/docs/9.4/static/trigger-definition...

> A trigger is a specification that the database should automatically execute a particular function whenever a certain type of operation is performed. Triggers can be attached to tables, views, and foreign tables.

Even NoSQL DB's like redis or riak have the ability to execute user-supplied Lua resp. Erlang.

http://docs.basho.com/riak/latest/dev/advanced/commit-hooks/

http://redis.io/commands/EVAL

[cloudgeek]

Yes, but this isn't a database, it's an object store. Block, file, and object storage platforms have yet to implement similar functionality. Block and file platforms are in particular very closed.

[KaiserPro]

Not really, Block systems are just glorified key:value databases.

Here are a few scalable open source block/object storage systems:

Lustre

Ceph

tahoe-lafs

XtreemFS

These are all opensource.

For example, if I wanted a fast global-coherent CDN, then all I need are webservers and GPFS, with some glue code to allow auth'd uploads/deletes. geographically scalable, with tiered storage levels.

If you want to run on dodgy hardware, then use XtreemFS. but you'll pay a performance price.

I'd be interested in a comparison in developer language(as in graphs code examples etc.), with a github link or similar.

[cloudgeek]

We might not be using the same definition for "block". Happy to chat offline if you'd like to email me ([e-mail]). Cheers

[MartinMond]

What about riak's pre-commit hooks?

[cloudgeek]

Appears to be only related to the write path, i.e. pre-write and post-write. Doesn't do anything for read, delete, rename, search. Also seems to be largely focused on validation. Thank you for pointing this out though.