Logical Arrangement of Virtual Disks, Host, Data Replication Folders

[fh-djb]

Greetings,
I was hoping to get some opinions, advise, and/or examples of how some of you logically setup your virtual disks, hosts, and data replication within Command View to determine if you have any better ideas than my own. Initially what I did was break up Virtual Disk folders by the type of file system the vDisk would be (e.g. NTFS, HPUX, VMFS, RDM). When we started playing with CA I found it might be more logical to create Virtual Disk folders based on the name of the system being replicated (e.g. EXCHANGE_SERVER_NAME) and put all the vDisks presented to the host within this subfolder.

Then we started playing around with virtualizing Exchange 2003 with VMware VI3 and now the replicated LUNs are a mix of VMFS volumes and RDM's (Raw Device Mappings).

Luckily we're in testing phase of this project so nothing is in production, but the my Virtual Disk, Host, and DR folders are becoming a huge mess! Any suggestions on how to keep a good house within CV? How do you all arrange your objects?

[Uwe Zessin]

I organize the virtual disks by single-host or cluster.

RDMs can be a bit hard to identify, so I use different LUN address ranges for VMFS and RDM disks. RDM disks can be collected in a per-VM subfolder. For example:

\\Virtual Disks\ESXCluster1\VMFS-1
\\Virtual Disks\ESXCluster1\VMFS-2
\\Virtual Disks\ESXCluster1\VM1-RDM\VM1-RDM-1
\\Virtual Disks\ESXCluster1\VM1-RDM\VM1-RDM-2
\\Virtual Disks\ESXCluster1\VM2-RDM\VM2-RDM-1

[Tim Nelson]

I also organize by host. The virtual disk name then includes the intended virtual guest.
VHost1>>VGuest1_disk1_etc_etc
>>VGuest2_disk1_etc_etc
>>Vguest2_disk2_etc_etc

You could also create additional folders from there.

If you have a large mix of platforms then an upper level of platform then host would be prudent.

HPUX>>host1>>etc..
WDOZE>>host3>>etc..
LINUX>>host2>>etc..

I find the more fully qualified the names then the items become self explanatory reducing error and bad/old documentation.

[Ivan Ferreira]

We have a grouping like this. Just think in how will be the easiest way to you to locate a disk if someone ask you:

Can you tell me on which disk is stored my database? How would you start searching?

Microsof
Dataware
hostname
BD
vdisk
vdisk
Logs
vdisk
vdisk
Groupware
hostname
BD
vdisk
vdisk
Logs
vdisk
vdisk
Linux
Monitoring
hostname
Aplication Server
hostname
vdisk
hostname
vdisk
Unix
Production
CluProd
applications
vdisk
vdisk
vdisk
database
datafiles
vdisk
vdisk
vdisk
logs
vdisk
vdisk
vdisk

[Gary McCready]

Our arrangement tries to take into account both keeping track of space via platform or location, especially when my admins like to change around what gets loaded on the blades in our bladecenter

For both hosts and VDisks, our folder structure usually is

OS Platform > physical blade, cabinet or cluster name > hostname

With the Vdisks belonging to the host or cluster in the hostname folder. Some of those folders for shared disks in a cluster (or ESX environment) have names like NAS0102 or ESXFARM1, etc.

For Vdisk names, we usually will start with the hostname, followed by lun and GB of the disks, such as

ESXFARM01_011_0512

And for some Vdisks, we will place a filesystem name/mount point as a last field at the end of a vdisk name.

We've found this useful in catching errors or even just parsing the SSSU output in excel, although there are much better ways to do that.