When look for a storage, a simple thing for system admin was a NAS. With a NAS appliance, your life is easier. Plug and forget, if ever happend to the harddisk, just replace it as already have RAID system. But still, the storage size is limited to the disk bay provided and a disk size available, also with RAID, disk size is limited to the smallest disk available.
So when we need a future expanding of the storage size, NAS is not an easy way to do the storage expansion. Come the GlusterFS for storage cluster. With GlusterFS you can easily expand / scale out your storage size. just add a server and a disk storage, put it on the glusterFS cluster, your storage size is bigger.
Some of notes when using GlusterFS :
1. Run on Linux server
2. Use XFS file system (recommended by Redhat)
3. Use LVM system
4. Use Hardware RAID for disk level redundancy
5. Use native glusterFS client
The important thing to highlight is use the native glusterFS client as it will provide redundancy, because all the data stored on the storage nodes, the client will fetch it automaticly by connecting to all the nodes inside the storage cluster. The initial connection only to the mount point node of the storage cluster. This will provide load balance of load inside the storage.
Also the RAID on hardware level. If Disk problem, just swap it as provided by the RAID hardware. Easier to do than setup RAID on the OS level.
For the default configuration, GlusterFS distributed your file to each storage nodes. You can have other options in GlusterFS base on your needs as :
1. Distributed Volumes
2. Replicated Volumes
3. Striped Volumes
4. Distributed Striped Volumes
5. Distributed Replicated Volumes
6. Distributed Striped Replicated Volumes
7. Striped Replicated Volumes
8. Dispersed Volumes
To provide redundancy and availablility, i am using Distributed Replicated setup, where i set 2 replica and distributed the store to every nodes.
The minimum brick required are 4 (assumes each server have 1 brick only).
So we distribute the data to 4 storage node, with 2 replica on each data.
 | |
| GlusterFS Distributed Replicated Volume. Source: GlusterFS Documentation |
This storage setup i used for Owncloud storage backend.
With this, we able to expand the storage size anytime with adding more VM or server node to the cluster.
0 comments:
Post a Comment