Geovision System recording to RAID 6 Setup

[simo923 0]

Hi,

 

I have 4 servers, each with 2 x GV 2008 in. So there are 16 cameras and 16 audio feeds going to each server.

 

In each server there are also 16 1Tb HDDs which make up the RAID 6, so the actual recording space i have is about 12.7Tb.

 

The RAID setup has split the 12.7Tb into 6 logical partitions, so essentially i can store 2.7 cameras on each logical partition.

 

So in the Geovision software i have setup my storage locations as the 6 logical partitions, with the 2 high movement cameras in some locations and 3 low movement partition cameras in the other, obviously because i cant save 2.7 cameras.

 

Does this seem like a sensible/workable setup or is there any better way to do it.

 

If one of my harddrives (or two as the RAID can handle 2 failures) goes down, how does the Geovision react? Will it stop recording the cameras? This is one area i am unsure of as i have not had the chance to simulate the scenario yet.

 

 

TIA

[Soundy 1]

GeoVision shouldn't know the difference. Disk failure handling is taken care of entirely by the RAID controller and its subsystem. The array itself just appears to most software as just another drive letter on the system.

 

I'm not quite sure why you'd split the array up into multiple partitions, unless it's to separately control the retention time for groups of cameras (for example, you don't want your high-framerate cameras "stealing" space from the others).

 

I'm not that familiar with GeoVision, but I've done something like this for several Vigil systems on multiple sites for one client. In their main site, I have three 1TB drives in the machine, and 8x1TB drives in a RAID5 array using an Enhance R8IP rack unit. The DVR simply connects to the array via iSCSI, and appears as a drive letter with about 6.5TB available. Vigil handles disks as a "pool" - each additional drive letter is just added to the pool, and the system cycles through them as space is needed.

 

I did have a disk failure in that first array, too... as long as your array supports hot-swapping, you simply pop the disk out, and pop a new one in, all without shutting down the system. The RAID controller should then automatically rebuild the array. And again, your software should never even know the difference.

[g00fy 0]

why to use raid at all?

tb disks are cheap (less then 100$),

I personaly have 20x1TB disks with 24 cameras (16 analog +8 IP),

each camera has its own disk (some analog are 2 per disk as they don't require much diskpace), if one disk fails (and why would that be?), i loose only 1 camera fotage, that I can easily recover, and I replace broken drive with another. If the footage from that camera is realy important, i can send it to a data recovering company, and they will do it for less than 200$ unless the disk is physically demaged.

[Soundy 1]

Hmmm, I already replied to this one... looks like the board has swallowed a post again.

 

So why RAID? Well... how do you fit 20 3.5" drives into a machine, let alone connect all of them? Or are you using externals (eSATA, USB or Firewire)? Externals are going to take up a lot of desk/shelf space, nevermind the powerbars needed to plug in all their power supplies... plus power supplies for USB hubs, if applicable. Sure it could work, but it's a pretty bulky, inelegant solution.

 

Rather then risking ANY data, RAID 5 or 6 allow you to simply swap out a failed drive (assuming they're in a hot-swap enclosure), and the system will automatically rebuild the array - no need to take anything offline, no need for downtime on any cameras.

 

But as with the "VPN-vs-port-forwarding" thread, a lot will depend on the individual situation and the needs of the client.