Windows Server 2012 Storage Feature Overview

Microsoft is taking the wraps off of the latest version of Windows Server -- Windows Server 2012 -- at their BUILD conference in Anaheim, CA this morning. I joined a few other Windows IT Pro editors at an invite-only press event last week to get an advance look at what Windows Server 2012 (formerly called Windows Server 8) has to offer, and we've been holding our tongues (thanks to an NDA and press embargo) until then. The embargo broke just a few minutes ago, so we're now able to spill the beans about what we saw.

My Windows IT Pro colleagues and I divided and conquered when it came to covering the full spectrum of what Windows Server 8 has to offer: Michael Otey covers the massive number of improvements to Hyper-V, while Sean Deuby takes a look at Dynamic Access Control and improvements to Active Directory. Paul Thurrott is also covering Windows Server 8 from many angles at the Windows Supersite.

I drew the storage card, so I'll provide you a quick overview of all the new storage features in Windows Server 2012. And believe me, there are a TON of all-new features (and improvements to existing ones) to talk about in the storage arena. In this overview I'll focus on three: storage pools and spaces, improvements to CHKDSK, and data deduplication.

 

Storage Pools and Spaces
With many IT environments filled with a huge array of storage devices of varying types and sizes, finding a way to easily manage and allocate that storage has historically been an onerous task. Microsoft hopes to change that with storage pools and storage spaces, two new Windows Server 8 storage abstraction concepts.

You can think of Storage Pools as units of storage aggregation that provide administration and isolation; Storage Spaces give virtual disks performance, resiliency, and simplify storage provisioning.

As an example, you can use storage spaces to aggregate separate individual storage devices into a single unit of storage, and then provision and divvy up that storage space as you see fit. It's a powerful way to provide storage for virtual machines, and simplifies management of disparate storage types immensely. Storage pools and spaces also can scale from the SMB up the the largest enterprises, which promises a new level of storage flexibility.

CHKDSK Redux
The venerable CHKDSK has been around in some form or another since the days of MS-DOS, and has been the bane of many system administrator's existence on many occasions. Running on larger storage volumes, a CHKDSK process can sometimes take hours to complete, which can throw the best-laid plans of even the most organized of IT professionals into the toilet. Microsoft seems to have finally heard the psychic anguish of millions of IT pros crying out when the dreaded CHKDSK message appeared, and has some long-overdue improvements in store for CHKDSK in Windows Server 8.

The biggest news is that Microsoft has decided to split CHKDSK scanning into two separate phases: an "online" scan and corruption-logging phase where CHKDSK searches volumes for defects behind the scenes, and a vastly shortened "offline" fixing phase that only corrects defects in drive data.

Microsoft Senior Program Manager for Storage and File Systems Matt Garson gave a compelling demonstration of how much time the new CHKDSK can save over the old version. Here's an example: Scanning 300 million files with the old CHKDSK could take 350 minutes, while the same number of files in the new version takes less than 8 seconds. (See slide, below.) It's an amazing speed improvement, and shows that Microsoft isn't just focusing on producing only shiny new features in Windows Server 8, but is also addressing some Windows Server features that have caused system administrators angst and frustration for years.

 

Windows Server 8 CHKDSK

Data Deduplication
With demand for physical storage increasing exponentially, one way to reduce storage demands is to rely on technologies like data deduplication, which Microsoft leverages in Windows Server 8 to reduce file storage sizes.

Here's how it works: Let's say you have dozens of VHD (virtual hard disk) files. Many of the files on those VHDs are identical copies of each other, such as Microsoft Paint. Data deduplication removes all the copies of Microsoft Paint from all of those VHDs but one, puts all of that redundant data into a separate store in System Volume Information (SVI), then simply leaves a marker that points to the file that serves as the template. Imagine this used across thousands (if not millions) of files and throughout your storage network, and you can expect to see vast reductions in storage space.

I went through a hands-on Windows 8 lab to test data deduplication, and was impressed to discover that the technology works across networks to separate Windows Server 8 or Windows 8 machines as well -- when you copy files the time it takes to do so is vastly reduced. It's an impressive feature that should do wonders for storage efficiency and network utilization. Maybe this is the first of the many "better together" features of Windows Server 8 and Windows 8?

I've only scratched the surface of all the new storage features in Windows Server 8: I haven't mentioned the improvements to cluster shared volumes (CSV) and its expansion beyond Hyper-V, Bitlocker support for shared cluster disks, cluster-aware updating, SMB2.2 storage support, and continuously available Hyper-V storage on remote SMB2.2 shares. We'll be covering Windows Server 8 storage features in much more detail in the weeks and months to come.

For even more information on Windows Server 8 and Windows 8, check out our dedicated news landing pages here at Windows IT Pro and over at the Windows Supersite.

Discuss this Blog Entry 3

on Sep 14, 2011
Hi Richard, I spoke with a contact at Microsoft who has access to the team responsible for storage features for Windows Server 8, and they passed along the following answer: "This is below the filesystem but there are optimizations for sequential writes. We also preserve the disk locations for data on writes which helps reads out, and reads tend to dominate performance and tend to affect response times. Of course for the mirror options, we are just about as optimal as anyone can be since there is no parity." Let me know if that answers your questions. Thanks for the question! - Jeff
on Sep 14, 2011
The interesting thing that I saw during the keynote was the combination of disks in a JBOD configuration into a single pool at the OS level (without the use of traditional RAID). How is this going to impact performance vs. any of the standard RAID levels? Will Windows 8 be doing anything like ZFS ("RAID" at the file system level) or is this just another implementation of software RAID with all the usual performance overhead?
on Sep 17, 2011
Thanks so much for that information Jeff - it will be interesting to see the performance vs. traditional RAID levels. =D

Please or Register to post comments.

What's Business Technology Perspectives Blog?

Insight, expert commentary, and strategic perspective that will help business decision-makers (BDMs) make the right purchasing decisions.

Blog Archive

Sponsored Introduction Continue on to (or wait seconds) ×