LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Server (http://www.linuxquestions.org/questions/linux-server-73/)
-   -   Can Intel SCSI RAID controller See Existing Disk Array in another Server? (http://www.linuxquestions.org/questions/linux-server-73/can-intel-scsi-raid-controller-see-existing-disk-array-in-another-server-577362/)

Runge_Kutta 08-15-2007 04:13 PM

Can Intel SCSI RAID controller See Existing Disk Array in another Server?
 
The Intel controller is a SRCU42E (U320) in a Dell PowerEdge 2970, and I want the Dell server to see an existing RAID array from an APPRO SR2012G box with an Adaptec AIC-7902 SCSI RAID controller.

So, basically, I want to know if it is possible to:

plug a scsi cable between the two server boxes and have the Dell see the Appro's raid array (already formatted with data on it?)

is this possible? if yes, is it risky? and how would I go about doing this if there are other steps involved besides cabling up?

ajg 08-17-2007 07:09 PM

It would help if I knew what you're trying to achieve.

Short answer is a big fat NO.

Long answer is ...

Possibly, but only if it's RAID1. Any other RAID level and the Intel controller won't understand how the data is structured as each manufacturer uses subtly different methods to organise the array.

If it is RAID1, there is a slim possibility that it would talk to the drive, but once again, it depends on how the Adaptec is structuring the array. The general recommendation is that both controllers should be identical, right down to the firmware revision on the RAID cards.

If you're trying to access one volume from two servers at once by simply hooking both servers up to a single SCSI bus, you're heading for a disaster. Two servers accessing a single filesystem will bork very quickly as there is no control over which server is accessing what data - you can imagine the mess it would make of your filesystem. Just try one of those internet drawing things where everybody can draw on the same page at once. The results would be similar, only it would be your data getting scribbled all over! If you want to do this, you need to look at some sort of clustering software, with distributed lock management to arbitrate access to the shared volume(s), but it is a bit of a black art, and there are limitations on what you can do with it.

If you're just wanting to copy data over to another server, you can do it simply with the Adaptec RAID server powered up, but all the caveats above apply, although I would guess you aren't wanting to do that otherwise you'd just copy the data over the network.

Final point ... hooking up different manufacturers' controllers to a RAID array is very risky, as you can never be sure what the outcome will be. It may completely destroy all the data on the array. It may work flawlessly. It may do any combination of weird things in-between. Be sure that the vendors haven't tested it, so you're on your own.

I would strongly advise against any of this without identical RAID controllers, and clustering software to arbitrate access between the servers.

IMO it would be simpler and far safer to put an additional gigabit NIC in each server with a crossover cable between, and rsync whatever data you need between the servers, although that may not be possible depending on what you're trying to achieve.

Runge_Kutta 08-18-2007 03:33 PM

thanks dude, my question above is pretty much answered.

You bring up an interesting point about the extra gigabit nic:
as a matter of fact, both servers have a spare gigabit nic already installed. I could just create an ip for the server serving out the data (the one with the Adaptec), hook it up to the back of a spare port on a gigabit switch, and have the other server as the only one that can access data from that ip. This should be a safer way for second server to have fast access to data on first server (though it won't be U320 speeds).

again, thanks!

ajg 08-18-2007 04:04 PM

Happy to help!

Unless they're some distance apart, you'd be better off just using a crossover cable because you want to avoid having the traffic anywhere near your LAN. It also adds an additional point of failure to the system.

Use a private IP subnet just for the traffic between the systems.

I do a this between my web server and database server - they talk over their own private network which keeps all the database traffic away from my LAN, and ensures the database gets maximum network I/O at all times.

Always remember K-I-S-S ... Keep It Simple and Stupid - you can't go far wrong!


All times are GMT -5. The time now is 03:38 AM.