I'm not expert, but I can guess that to make the whole thing scalable,
gluster needs to go through gluster client, then network, then gluster
server, then native disk write, while native write does native disk only.
Again, it doesn't make sense to do such a comparison, or if single node
fit you, get rid of gluster.
-C.B.
On 11/27/2013 4:57 PM, lei yang wrote:
On Wed, Nov 27, 2013 at 5:14 PM, Marcus Bointon
<[email protected] <mailto:[email protected]>> wrote:
On 27 Nov 2013, at 09:30, lei yang <[email protected]
<mailto:[email protected]>> wrote:
I have a machine which have 5 hard disk
I want to use glusterfs to fast my disk
You do know that gluster is not about single-node performance?
You'll get far better performance by using RAID 0/1/5/10 on your
local machine. Gluster gives you multiple machine redundancy (like
RAID, but spread across multiple machines), and size scalability
beyond what you can fit in one box.
root@ovpovp-S2600CP:/mnt/lyang0# dd if=/dev/zero bs=1000M
count=10 of=./1G
10+0 records in
10+0 records out
10485760000 bytes (10 GB) copied, 55.5562 s, 189 MB/s
From a single node that looks reasonable for gluster. You'd
probably get 3-4 times that with native local access.
but the truth is it's much worse than use my native harddisk, that's
is why I post my question, any setting is wrong with me ?
Lei
Marcus
_______________________________________________
Gluster-users mailing list
[email protected] <mailto:[email protected]>
http://supercolony.gluster.org/mailman/listinfo/gluster-users
_______________________________________________
Gluster-users mailing list
[email protected]
http://supercolony.gluster.org/mailman/listinfo/gluster-users
---
This email is free from viruses and malware because avast! Antivirus protection
is active.
http://www.avast.com
_______________________________________________
Gluster-users mailing list
[email protected]
http://supercolony.gluster.org/mailman/listinfo/gluster-users
