In docs for boltdb it is stated that NoSync should be used very carefully 
because it can leave the database in an inconsistent state.

If a full fledged distributed database was the case, a no-sync manner would 
be acceptable. But for an embedded database, it is not a plus (to be 
performant in no-sync mode, assuming that mode would not leave database in 
an inconsistent state, in case of a dirty shutdown of the app or the 
machine).

On Saturday, January 20, 2018 at 3:56:10 PM UTC+3:30, Jordan Krage wrote:
>
> Are you using 'SyncWrites=false'? (Or 'NoSync' with boltdb?)
>
> It looks like the badger benchmarks from github.com/dgraph-io/badger-bench 
> do not sync. 
>
>
> On Thursday, January 18, 2018 at 1:33:02 PM UTC-6, dc0d wrote:
>>
>> Badger write performance is a bit worse than boltdb. And badger suggests 
>> to batch writes. But so does boltdb.
>>
>> At the same time at badger's GitHub page it says it has higher write 
>> performance compared to boltdb.
>>
>> Is there a sample of how to do high performance/throughput writes with 
>> badger?
>>
>

-- 
You received this message because you are subscribed to the Google Groups 
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to