Want to protect your cyber security and still get fast solutions? Ask a secure question today.Go Premium

x
?
Solved

Best practice for lun presentation/carving for SQL instances

Posted on 2011-02-28
6
Medium Priority
?
1,099 Views
Last Modified: 2012-06-21
Hi,

We are about to upgrade our current SQL environment which will utilise our current SAN environment. We currently have two SQL instances, one physical and one virtual (ESX).

I have configured three LUN’s on our SAN which are:

-  VMFS_LUN (ESX Server storage)
-  Data_LUN (SQL data)
-  Logs_LUN (SQL logs)

Below are the two SQL server instance and their disk configurations.

SQL Physical Server
C:  Local disk - OS + App install
D:  Shared storage (Log_LUN)
E:  Shared storage (Data_LUN)

SQL Virtual Server
C:  Shared Storage (VMFS_LUN) - OS + App install
D:  Shared storage (Log_LUN)
E:  Shared storage (Data_LUN)


My question is related to best practice for lun presentation/carving for SQL instances.
Is it better to create a different LUN for each SQL instance I have? For example: Log_Lun1 (SQL Physical Server) &  Log_Lun2 (SQL Virtual Server)

Also on our SQL virtual environment D and E drives, is it better to create these drives as VMFS which point to the Log and Data LUN’s? or should I used the isci initiator in windows?  


Any advice or suggestions will be welcomed!!!!
0
Comment
Question by:cmga
6 Comments
 
LVL 42

Expert Comment

by:Paul Solovyovsky
ID: 35002869
What are your IOPS requirements, SAN model, and Backup capability?
0
 
LVL 125

Accepted Solution

by:
Andrew Hancock (VMware vExpert / EE MVE^2) earned 1500 total points
ID: 35002890
Personally I would create, different LUNs for different SQL Instances, for Data DB and Data Logs.

VMFS versus iSCSI initaitor?

How will you backup the SQL servers, if it's all virtual, third party backup tools will back it all up.

If you present an iSCSI LUN via Microsoft Initiator, you'll need another mechansim to back this up, because it's not a virtual presentation (e.g. vmdk), or you could consider an RDM to the virtual machine as well. Remember VMDK on VMFS, give is lots of flexibility, versus the trade off of some performance.

also if you do iSCSI in the virtual machine, the virtual machine CPU is doing the iSCSI work, rather than presenting the LUN to ESX, and formatting the LUN as vmfs, and using a standard VMDK on the LUN.

I would recommend, presenting LUN direct to ESX server.
0
 

Author Comment

by:cmga
ID: 35003190
IOPS Requirments? Still looking at this.

SAN Model? HP LeftHand Networks P4500

Backup Capability? Databases are backups by SQL and we do a file level backup of all the *.bak files on a nightly basis.
There are no requirements for a VCB image backup of the virtual SQL server as yet, but we will configure an image back of the virtual sql server which will exclude d & e drives.
0
Technology Partners: We Want Your Opinion!

We value your feedback.

Take our survey and automatically be enter to win anyone of the following:
Yeti Cooler, Amazon eGift Card, and Movie eGift Card!

 
LVL 4

Expert Comment

by:virtualxistance
ID: 35003395
Is the Physical / Virtual SQL instance for redundancy or just 2 seperate Instances?

Unless you are trying to use some sort of an Active Passive SQL setup, with a SQL Cluster / File Cluster  i don't think you'd be able to point to the same LUN from both servers and do a simultaneous RW.

If for seperate instances, I would suggest having 2 different sets of LUNS for the Data and Logs on each server.

Let ESX take care of the LUN than using iscsi initiator in Windows.
0
 
LVL 4

Expert Comment

by:Vijay kumar Mohanraj
ID: 35017253
The simple answer is you want a config that's balanced and easy to understand without referring to a spreadsheet each time you allocate LUNs. So here's what I've recommended for some of my customers:

RG/LU Carving:

All LU's are the same size. Larger LU's are created by merging multiple LU's together.

Within any raid group, every LU is assigned to the same controller.
RG0/2/4/6/8 -- Even raid groups (RGs) go have all their LU's assigned to CTL 0
RG1/3/5/7/9 -- Odd RGs have all their LU's assigned to CTL 1

A LU can be any four digit number from 0000 to 4095.

First two digits of the LU number matches the RG number, last two digits represent the "slice" within the RG.

So the LU's within RG6 would look like:

0600, 0601, 0602, 0603, ...

You can also make the first ten LU's in a RG small, the rest big, and separate them by gaps in the slice, i.e.: 0600-0612 = 30GB, 0640-0667 = 120GB, ...

This numbering system allows easy space recovery and reallocation when a server is decomissioned.
0
 

Author Closing Comment

by:cmga
ID: 35178122
No all items where answered.
0

Featured Post

What does it mean to be "Always On"?

Is your cloud always on? With an Always On cloud you won't have to worry about downtime for maintenance or software application code updates, ensuring that your bottom line isn't affected.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

August and September have been big months for VMware—from VMworld last month to our new Course of the Month in VMware Professional - Data Center Virtualization. We reached out to Andrew Hancock, resident VMware vExpert, to have a more in-depth discu…
In this article will go through how to backup a vPostgres DB from a broken vCenter Appliance and restore to a new vCenter Appliance.
This Micro Tutorial will teach you how to reformat your flash drive. Sometimes your flash drive may have issues carrying files so this will completely restore it to manufacturing settings. Make sure to backup all files before reformatting. This w…
This video shows you how easy it is to boot from ISO images for virtual machines with the ISO images stored on a local datastore on the ESXi host.
Suggested Courses

580 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question