MCDFS is inspired by Heystack. Its stand-alone storage is the simplified weed-fs, so it has better performance. raft is used as MCDFS's consensus Algorithm
# start two processes
$ MCDFS -vl YOUR_VOLUME_LOCATION /tmp/node.1
$ MCDFS -join localhost:4001 -vl YOUR_ANOTHER_LOCATION /tmp/node.2
$ curl -F "[email protected];type=image/jpg" http://127.0.0.1:4001/write
1
{'Cookie':13412123,'Offset':'0','Size':1234123}
$ wget http://127.0.0.1:4002/read/1/0/1234123/13412123
(get file just uploaded)
run with no replcation @ Intel i5 CPU M480 2.67GHz, 6G Ram, 5400 rpm disk
payload: 10K file
cluster config: one instance
write via http
400000 ops in 53.458s
qps:7482.50
cluster config: three instance
write via http
80000 ops in 36.83s
qps:2172.14
