"DB has methods to handle a transaction lifecycle: commit(), rollback() and close().
One DB object represents single transaction. The example above uses single global transaction per store, which is sufficient for some usages:"
... but this doesn't say anything about concurrency or how to create additional transactions?
Basically I am looking for design guidelines for a store that processes concurrent updates that require isolated reads (increments existing values in the store); and also processes periodic reads.
Thanks!