I'm working on a project with a client who recently had a large system replacement for a dedicated ERP application. I'm inheriting this hardware setup and want to understand what it means.
The server is a ProLiant DL580 G7 (4 x 8-core CPUs, 64GB RAM). The customer has a set of ~500GB of data that historically grows 1-2% per year. The system has plenty of onboard storage in the form of 8 x 300GB 6G SAS internal disks. The server is connected to an MSA2324sa SAS-attached SAN unit, fully populated with 24 x 146GB disks. This SAN has two SAS controllers and the server has two HBAs, each connected via SFF8088 cable to one port on each controller. This is the only server connected to the SAN.
While this is fine and appears to be a standard cabling setup, the confusion comes in the presentation to the OS. The server is running RHEL 5.7. Once I gained access to the system, I noticed the following
This seems odd to me. Unfortunately, I can't find any information to confirm the validity of this setup. Most of my experience has been with Smart Array controllers and "dumb" JBOD expansion units, but I tend to use HP's facilities to carve LUNs and manage volumes.
Looking at this setup, why was it necessary to use LVM striping? Is this uncommon? Does it have more to do with the multiple SAS connections to the server? The vdisk/volume arrangement of the MSA seems like it would provide the necessary flexibility. As-is, it seems like this SAN has no room for additional servers or provisions for spare disks, etc.
I just want to check with engineers who are more familiar with these deployments. Is there anything wrong with this setup?