i would like to hear your ideas about couchdb, and wouldn't it handle my use situation.
Things i is going to do, i'll have database where i store documents in dimensions about 20kb with attachment of just one-10MB for every.
will couch handle database 10TB or even more per server with my schema?(in 4u situation place the 24 2TB drives is an excessive amount of per couch node?, you will see very less reads, and so i lower need speed)
will couch have the ability replicate all documents with accessories
what about splitting all data to multiple servers (for instance to 4 nodes)? does it handle much accessories?
what problems would you see here?
require more info please request :)
I do not think you'll hit an actual limitation having a 10TB file, that's I do not think couch has some built-in "can't use files larger than X" with X being < 10TB.
The greatest problem may be the file compaction. To be able to reclaim space, Couch really wants to compress the file. This effectively means copying the file. So, for many point a minimum of, 10TB must be 20TB because it replicates the live data within the new copy.
If you're mostly appending towards the file, that's you're simply adding new data and never upgrading or overwriting old data, then this is a lesser problem, as compaction will not gain you quite much. In case your information is essentially static, i quickly would build the file little it your final some time and be doe by using it.
You will find "third partyInch sharding solution for Couch, Lounge is popular.
After I approach a couch solution the main factor to think about is exactly what your query criteria is. Couch is about the sights, really. What type of sights are you currently searching at? If you are simply storing data by some simple key (file title, the date, or whatever), you could be best simply utilizing a file system, as well as an appropriate directory structure, frankly.
So Let me learn more regarding your sights you intend to make use of because you don't plan to perform a large amount of reading through.
You'll still haven't pointed out what type of queries you are searching for. The queries are, effectively, The look component, specifically for a Couch DB because it will get increasingly more hard to add new queries on large datasets.
Whenever you stated accessories, I believed you meant accessories towards the Couch DB payload (because it are designed for accessories).
So, all nevertheless, you can easily create meta-data document taking all the whatever information you need to capture, and included in that document give a path title towards the actual file saved around the file system. This can lessen the overall size the Couch file significantly, making the constant maintenance faster and much more efficient. You lose a few of the "All-in-oneInch a part of getting everything in one document, obviously.