We help IT Professionals succeed at work.

Windows Server 2008 R2 GPT or MBR

LeonesIT
LeonesIT asked
on
Hi all. I am configuring a HP Proliant DL380G6 FILE-Server with the following config:
OS: Windows Server 2008R2
OS RAID: RAID1
DATA RAID: 5 146 GB 2-Port SAS Drive 15k in a RAID5 with 1 hotspare
RAID 5 stripe size: 64k
Purpose: File server
Roles/functions: Fileservices, shadow copy (previous versions), Access based enumeration

I want to know the following:
1. Should I use MBR or GPT when initializing the disk in Windows? Is there a performance difference between the two?
2. Is the 64k stripe size good for the purpose as a file server?

Thanks
Comment
Watch Question

Commented:
If i remember correctly GPT is used only for 2TB or more.
Senior IT Advisor
Top Expert 2008
Commented:
Here's a good link that describes the differences beter than I could

http://www.microsoft.com/whdc/device/storage/GPT_FAQ.mspx

2. What is wrong with MBR partitioning?
MBR disks support only four partition table entries. If more partitions are wanted, a secondary structure known as an extended partition is necessary. Extended partitions can then be subdivided into one or more logical disks. By convention, Windows creates MBR disk partitions and logical drives on cylinder boundaries based on the reported geometry, although this information no longer has any relationship to the physical characteristics of the hardware (disk driver or RAID controller)., For Windows Vista and Windows Server 2008, more logical boundaries will be selected when the hardware provides better hints at the true cache or physical alignment. Because this partition information is stored on the drive itself, the operating system is not dependent on the alignment. MBR partitioning rules are complex and poorly specified. For example, does cylinder alignment mean that each partition must be at least one cylinder in length? An MBR partition is identified by a two-byte field, and coordination is necessary to avoid collision. IBM originally provided that coordination; today there is no single authoritative list of partition identifiers.

Another common practice is to use partitioned or "hidden" sectors to hold specific information. That practice is undocumented and results in severe system problems that are difficult to debug. Over the years, vendor-specific implementations and tools have been released to the public, making support difficult.

3. Why do we need GPT?
GPT disks can grow to a very large size. The number of partitions on a GPT disk is not constrained by temporary schemes such as container partitions as defined by the MBR Extended Boot Record (EBR).

The GPT disk partition format is well defined and fully self-identifying. Data critical to platform operation is located in partitions and not in unpartitioned or "hidden" sectors. GPT disks use primary and backup partition tables for redundancy and CRC32 fields for improved partition data structure integrity. The GPT partition format uses version number and size fields for future expansion.

Each GPT partition has a unique identification GUID and a partition content type, so no coordination is necessary to prevent partition identifier collision. Each GPT partition has a 36-character Unicode name, which means that any software can present a human-readable name for the partition without any additional understanding of the partition.

Distinguished Expert 2019
Commented:
Use GPT, then in the future you may expand the array past 2TB without having to backup/restore.