Answer the question
In order to leave comments, you need to log in
Can you recommend distributed network storage?
Advise please the distributed network storage. The data must be stored in key-value format. It needs to be able to lock the key, read its state, possibly change it and release the lock, any concurrent attempt to lock the key to change must wait for the lock to be released. When the connection is broken, the blocking should be removed. Ideally, this should be a transaction similar to:
begin; select for update; update; commit;
It should be possible to add servers and redistribute data to them both to speed up operations and to increase reliability.
At the moment, getting the hash of the key, calculating the hash of the mecached server and the postgresql server is used.
Accordingly, SQL and memcached / redis are not offered, other solutions are of interest.
Answer the question
In order to leave comments, you need to log in
Riak? True, in my opinion there are no locks and transactions, but you can always add functionality.
Mongodb sharding, there is atomic transactionality (+ it is more productive than regular transactions), only "dead locks" will need to be removed manually (or by timeout).
As an option, you can pay attention to Oracle NoSql Database . If there are no prejudices - there is a community version, but there is durability, consistensy, version - you will have to somehow get out through them, there are no locks at the record level, you can only require the application of value changes on all nodes, but this is apparently not what you need. Well, plus to this, the maximum consistency and durability (ALL) are guaranteed errors if at least one node has fallen or is unavailable.
Didn't find what you were looking for?
Ask your questionAsk a Question
731 491 924 answers to any question