zzsheng

6 ms·

1.
Benchmark object storage in objects/s, not GB/s(fractalbits.com)
8 points by zzsheng 3mo ago|0 comments
2.
by zzsheng 4mo ago
we manage memory instead of relying on page cache
3.
by zzsheng 4mo ago
sorry, typo. data will *not* be lost.
4.
by zzsheng 4mo ago
check s3 express one zone
5.
by zzsheng 4mo ago
journal file was pre-alloacated and we use direct-io for journal write so no need to call fsync.
6.
by zzsheng 4mo ago
nop. we will not ack any write which is not in data or journal. please check the put details in the blog.
7.
by zzsheng 4mo ago
actually both crash consistentcy and durability. after we ack, we make sure data will be lost due to crash, restart or power loss.
8.
by zzsheng 4mo ago
we also use direct-io for reads.
9.
by zzsheng 4mo ago
thanks for feedback. actually it was pointed out in blog that we do not use append only log to avoid fsync due to size change. what we use is preallocate fixed size log file and we do write journal data and space reclaim by 4KB unit, also w
10.
Removing fsync from our local storage engine(fractalbits.com)
66 points by zzsheng 4mo ago|78 comments
11.
by zzsheng 4mo ago
Author here. This is not a general argument against fsync; the design depends on SSD-only deployment, preallocated files, O_DIRECT, single-key atomicity, and device write guarantees.
12.
To fsync or not, that is a question(fractalbits.com)
2 points by zzsheng 5mo ago|1 comments
13.
Bootstrapping a Storage Cluster with S3 Based Workflow and Barriers(fractalbits.com)
5 points by zzsheng 6mo ago|0 comments