Microsoft Windows Server 2012 thin provisioning space ... · Microsoft Windows Server 2012 thin provisioning space ... VMware vSphere 5.0 storage APIs for ... such as those now included

Microsoft Windows Server 2012 thin provisioning space reclamation using

the IBM XIV Storage System

Author: Eric B. Johnson

Editors: Ted Gregg

Date: February 2013

Version: 1.0

Windows Server 2012 space reclamation using the IBM XIV Storage System

Page 2 ©Copyright IBM Corp. 2012

Table of Contents Executive summary ....................................................................................................................................... 3

Target audience ............................................................................................................................................ 3

Symantec Storage Foundation user considerations ..................................................................................... 4

VMware virtualization environment guidance ............................................................................................. 4

Space reclamation introduction .................................................................................................................... 4

Microsoft and IBM XIV Gen3 SCSI UNMAP support concepts ...................................................................... 5

Microsoft SCSI UNMAP support summary for the IBM XIV Storage System Gen3................................... 7

IBM XIV Storage System Gen3 ...................................................................................................................... 8

Introducing IBM XIV Storage System Gen3 Model 214 .......................................................................... 10

Microsoft thin provisioning and storage space reclamation features ........................................................ 11

Thin provisioning LUN identification....................................................................................................... 11

Threshold notification ............................................................................................................................. 12

Space reclamation using SCSI UNMAP .................................................................................................... 14

Storage device logical block mapping states .......................................................................................... 15

Microsoft Windows 2012 Data Deduplication ............................................................................................ 15

Microsoft Windows 2012 Data Deduplication support summary .......................................................... 16

Windows Server 2012 Data Deduplication installation steps ................................................................. 17

Triggering immediate Windows 2012 Data Deduplication for a NTFS volume ...................................... 24

Conclusion ................................................................................................................................................... 26

Resources .................................................................................................................................................... 27

Trademarks and special notices .................................................................................................................. 28



Executive summary With the release of IBM XIV Storage System Gen3 11.2.x microcode, customers can begin to capitalize

on the highly awaited space reclamation benefits associated with Small Computer System Interface

(SCSI) UNMAP support. While the IBM XIV supports SCSI UNMAP commands for both thick and thin

provisioned volumes, only the latter use case is presented in this white paper since SCSI UNMAP

benefits are typically associated with over-provisioned storage. For Microsoft data centers struggling to

mitigate explosive data propagation, SCSI UNMAP is a valuable option to help control storage sprawl.

Furthermore, SCSI UNMAP functionality is native to Windows Server 2012 which also offers new thin

provisioning developments and supplemental operating system (OS) space reclamation features.

This white paper reveals the simplicity in which Information Technology (IT) teams can now take

advantage of substantial data capacity savings using combined IBM XIV and Microsoft improvements

which include the introduction of the Windows Server 2012 Data Deduplication feature. Likewise,

Microsoft cloud administrators should gain practical insight into highly available (HA) Hyper-V capacity

savings tactics.

Target audience This documentation is intended for solutions architects, system, and storage administrators that

routinely manage IBM XIV Storage System Gen3 thin provisioned environments. Microsoft Windows

Server 2012 early adapters looking to maximize storage efficiency using the IBM XIV Gen3 should benefit

most from new multi-level space reclamation enhancements. Moreover, these latest thin provision

enhancements should appeal to large to midsize businesses evaluating or seeking additional storage

capacity savings strategies for both physical and virtual machines.

General experience or understanding of the required solution components including Microsoft Windows

Server administration, Microsoft Failover Clustering, Hyper-V virtualization technology, thin

provisioning, data deduplication and XIV Storage System Gen3 administration is recommended.

However, there are technical reviews as well as supplemental references in the remaining sections

below.

Prior to sharing information about new Microsoft space reclamation enhancements now available to the

IBM XIV Storage System Gen3, the following couple of sections direct Symantec Storage Foundation and

VMware users to solution-specific resources.



Symantec Storage Foundation user considerations Symantec Storage Foundation provides middleware functionality for its software-based dynamic disks

which is beyond typical Windows NTFS volume definitions. This differentiation carries over to its

alternate space reclamation approach. Rather than using SCSI UNMAP commands, Symantec Storage

Foundation employs SCSI WRITE SAME commands to achieve the same end result. However, SCSI WRITE

SAME space reclamation methods are outside of the scope of this paper. Thus, Symantec Storage

Foundation users can visit the following website for further details regarding space reclamation using

various IBM storage family products including the XIV Storage System:

http://public.dhe.ibm.com/common/ssi/ecm/en/tsw03164usen/TSW03164USEN.PDF

VMware virtualization environment guidance Unlike Symantec Storage Foundation, VMware vSphere supports SCSI UNMAP commands for its

datastore space reclamation. VMware began supporting SCSI UNMAP commands when it introduced the

VMware vSphere 5.0 storage APIs for Array Integration (VAAI) primitives. However, VMware discovered

issues which affected their Storage vMotion and VM Snapshot consolidation that led them to alter their

SCSI UNMAP support for its newest release – vSphere 5.1. Specifically, vSphere 5.1 does not provide

proactive or automatic space reclamation for SCSI UNMAP commands. Manual user intervention or

scripts must be implemented to realize the SCSI UNMAP benefits preferably outside of peak business

hours. For VMware virtualization environment guidance, review the following IBM Redpaper:

http://www.redbooks.ibm.com

Space reclamation introduction The demand for IT administrative productivity continues to increase simultaneously as hardware and

resource budgets decrease. This forces many organizations to re-evaluate their data center designs in

the midst of irrepressible data growth trends. With business critical data proliferation frequently

challenging storage-based teams, many IT administrators direct their attention towards maximizing

storage return on investments (ROI) and reducing costs. Hence, customers place much greater demand

on storage vendors for product portfolios with greater efficiency, advanced features and attractive,

competitive pricing. Historically to address storage efficiency, a natural progression ensued which

introduced thin provisioning.

Thin provisioning allows administrators to allocate logical capacity that is greater than a storage

system’s total physical capacity. It does so by using on-demand block allocation of data based on host

writes versus allocating all of the blocks during the initial volume creation. As a result of this on-demand

approach to allocating actual physical storage capacity, customers can realize significant economic

benefits by over-provisioning or thin provisioning their storage. By and large, this is due to not having to

commit considerable storage capacity up front (as with thick provisioning) to users or business groups

that often consume only a fraction of the allocated physical capacity. Consequently, cascading cost

http://public.dhe.ibm.com/common/ssi/ecm/en/tsw03164usen/TSW03164USEN.PDF

http://www.redbooks.ibm.com/



reductions are observed by diminished storage capacity requirements that contribute to smaller data

center footprints with lower administrative overhead and power and cooling expenses. Even with these

welcomed benefits, for many, thin provisioning lacked distinctive maturity due to space reclamation

deficiencies. More or less, many desired storage array dead space reclamation without having to use

limited or primitive utilities.

In order to fully appreciate dead space reclamation, one must examine the host front-end and the

storage back-end. Once a host writes to a thin provisioned volume, physical capacity is allocated to the

host file system. Unfortunately, if the host deletes the file, only the host file system frees up that space.

The physical capacity of the storage system remains unchanged. In other words, the storage system

does not free up the capacity from the deleted host file which is commonly referred to as “dead space”.

Obviously, this is not the most effective method for handling back-end block-level storage. Ideally, when

a host deletes files, that space is not only reclaimed by the host file system but also the back-end

storage system.

To address this thin provisioning limitation, the T10 Technical Committee established the T10 SCSI Block

Command 3 (SBC3) specification which defines the “UNMAP” command for a diverse spectrum of

storage devices including hard disk drives (HDDs) and numerous other storage media. Using SCSI

UNMAP, IT administrators can now reclaim host file system space and back-end storage dead space

typically within 30 seconds of a host file deletion. However, not only does SCSI UNMAP require T10 SBC3

compliant SCSI hardware, it also requires necessary software application programming interfaces (APIs)

such as those now included in Windows Server 2012 or Windows 8. That being said, previous Windows

OS releases do not support the necessary APIs.

Microsoft and IBM XIV Gen3 SCSI UNMAP support concepts In order to avoid common misconceptions about Microsoft Windows storage space reclamation using

the T10 SBC3 specification, key additional information is shared. As previously albeit partially disclosed,

Microsoft space reclamation requires APIs which are now included in Windows 8 and Windows 2012

that allow native or 3rd party applications to send TRIM or SCSI UNMAP hints to compliant storage

media. Notably for the IBM XIV Storage System Gen3, storage media refers to SCSI HDDs.

From an OS perspective, Microsoft Windows uses file TRIM or FILE_LEVEL_TRIM code to trigger storage

space reclamation using either TRIM or SCSI UNMAP requests depending on the target device types.

Consequently for Microsoft Windows solutions, file TRIM and space reclamation are often used

interchangeably due to their obvious relationship. However, this can be confusing since TRIM used in

this context is completely different than T13 Advanced Technology Attachment (ATA) device standards

which define ATA-specific TRIM commands. To help clarify, only SCSI devices can interpret T10 SBC3

UNMAP commands and only ATA devices, better known as Integrated Drive Electronics (IDE) devices,

can interpret T13 TRIM commands. That said, Windows determines the storage media type when the

device, or XIV LUN for the practical purposes of this paper, is attached or mapped to the host.

http://www.t10.org/

http://www.t10.org/



During the device enumeration process, Windows detects the device identity and if it is SCSI UNMAP or

ATA TRIM capable. When a large OS file deletion occurs, Windows triggers a FILE_LEVEL_TRIM and

converts the deletion or FILE_LEVEL_TRIM notification into an equivalent UNMAP request to pass to the

storage device. Acting as interpreter, the storage port driver stack translates the UNMAP request into a

SCSI UNMAP command or ATA TRIM command according to the detected storage device type. For all

intents and purposes of this document, IBM XIV Storage System Gen3 administrators should only be

concerned with SCSI UNMAP but it helps to be aware of ATA TRIM concepts for other applicable

solutions.

For Microsoft Windows physical environments, SCSI UNMAP behavior is fairly straightforward both at

the operating system and XIV storage level. Within Microsoft Windows, when a file is written to an NTFS

volume, the XIVGUI or XCLI interface immediately reflects an increase in the XIV volume used capacity

for the consumed space. When a file is permanently deleted on that NTFS volume, the XIVGUI or XCLI

used capacity for the XIV volume usually decreases within 30 seconds. It’s a valuable new real-time

space reclamation advancement that’s simple to comprehend and almost instantaneous. Even though

SCSI UNMAP behavior for Microsoft virtual machines is slightly more complex, it is also greatly

welcomed since Hyper-V clouds continue to multiply throughout global data centers.

For Microsoft Hyper-V virtual cloud environments, SCSI UNMAP behavior is more complex since space

reclamation can occur at multiple levels and often involves failover clusters. For Hyper-V host-based file

deletions that involve VM files or virtual machine storage migrations to other local or remote

destinations, space reclamation is identical to the aforementioned physical machine description. If the

Hyper-V host file (i.e. VHD, VHDX, ISO, etc.) is permanently removed from the host NTFS volume, the

host file system and XIV volume used capacity decreases and free space increases.

For Hyper-V VM guest-based file deletions, space reclamation occurs at 3 different levels:

Guest OS NTFS volume frees up space

XIV volume used capacity decreases

Host VHDX file size decreases

The first two processes are dynamic and near-instantaneous while the latter requires manual user

intervention.

Manual user intervention is mandatory since Hyper-V host VHDX files are mostly read-write files locked

by the VM guest OS which require the file to be unlocked in order to shrink in size. Therefore host VHDX

files do not currently allow real-time space reclamation. To be more precise, the guest-based file

deletions will not trigger a decrease in the host VHDX file size until additional administrative actions

occur. So when a Hyper-V VM user deletes files, only the guest NTFS volume and XIV volume reclaim

space at that time.

To trigger Hyper-V host space reclamation for the VHDX file, the VM must be offline. Hyper-V cloud

administrators can perform the following to decrease the host VHDX file size:



1. Within the VM guest OS, launch the Windows PowerShell with administrator privileges.

2. Run this command:

‘Optimize-Volume -DriveLetter F -ReTrim –Verbose’ (substitute appropriate drive letter)

Note: If space reclamation results do not occur in a timely or predicted fashion, always initiate

host or guest OS volume optimizations such as above. Quite often this course of action

accelerates the desired space reclamation.

3. Shutdown or restart the VM guest OS.

4. Upon shutdown of the VM guest OS, administrators observe an immediate decrease in the host

file size of the VM VHDX file(s).

Note: Since these steps require production VMs to be offline, cloud administrators should

synchronize host VHDX space reclamation with other guest OS maintenance windows that

require downtime. Combining these tasks should help minimize production VM outages.

Even though Hyper-V space reclamation is more involved, the rewards far outweigh the slight virtual

complexities. This also applies to Microsoft Hyper-V failover clusters. HA SCSI UNMAP behavior is not

hindered or negatively impacted by failover cluster resources. Subsequently, space reclamation is fully

supported for Hyper-V HA VMs that reside on cluster disks or cluster shared volumes (CSVs). VM quick

and live migrations are also supported and do not affect space reclamation functionality for Hyper-V

cluster hosts or their corresponding guests. Additionally, SCSI UNMAP is supported for HA Hyper-V VM

IDE and SCSI controller hard disks consisting of fixed and dynamically expanding VHDX files, pass-

through disks, fibre channel adapter (FCA) N-Port ID Virtualization (NPIV) disks and differencing disks.

Important: No matter the additional hypervisor or cluster stack layers, it is the detected storage media

that determines how Windows passes FILE_LEVEL_TRIM to the target device (using SCSI UNMAP for XIV

HDDs). For example, Hyper-V VM default system volumes that reside on an XIV volume and use virtual

IDE controllers, still benefit from SCSI UNMAP.

Microsoft SCSI UNMAP support summary for the IBM XIV Storage System Gen3 Requires IBM XIV Storage System Gen3 code level 11.2.x or greater

IBM XIV Gen3 FC and iSCSI volumes

Note: The smallest atomic unit that can be reclaimed via SCSI UNMAP is 1MB which corresponds to the fundamental building block size of an XIV volume known as a partition. Note: Although XIV volume used capacity reflects space reclamation upon host deletions, actual pool capacity is reclaimed in 17 GB units, also known as slices. XIV pool space reclamation occurs once an entire 17 GB slice is freed. This is due to the XIV grid architecture which always defines logical volume physical capacity in multiples of 17 GB (decimal) since each volume spans all physical drives in the system.



IBM XIV Gen3 mirror volumes

Note: SCSI UNMAP is supported for volumes that use synchronous mirroring as long as all XIV

Storage Systems are using 11.2 code. SCSI UNMAP is not supported for volumes that use

asynchronous mirroring.

IBM XIV Gen3 volumes that contain snapshots

Note: Depending on XIV thin provisioning threshold values, volume file deletions may trigger

snapshot deletions due to space exhaustion. Snapshot deletion priorities can be defined during

the initial snapshot creation and extra precautions should be taken not to exhaust XIV hard or

physical capacity.

IBM XIV Gen3 snapshot volumes

Microsoft Windows Server 2012

Windows 8

Microsoft failover cluster

o Cluster shared volume (CSV)

o Cluster disk

Hyper-V

o VHDX

o PassThru disks

o Virtual SCSI Controller or Virtual Fibre Channel Adapter disks

IBM XIV Storage System Gen3 Even without the latest space reclamation benefits, the IBM XIV Storage System Gen3 consistently

exceeds customer expectations by continuing to evolve and offer enhanced efficiency, more advanced

features and attractive, competitive pricing to an ever-increasing number of businesses. Building upon

this tradition, IBM introduced a new XIV Gen3 model as well as microcode 11.2, which satisfies T10 SCSI

UNMAP compliance and produces highly anticipated space reclamation advantages. All of which

strengthens an existing and immensely popular enterprise-class feature set for a storage system that

departs from the unnecessary complexities linked to conventional redundant array of independent disks

(RAID) solutions.

Upon examining the IBM XIV Storage System Gen3 closer, there are no traditional RAID requirements for

assigning specific spindles (disks) to individual storage pools. Both XIV regular (thick) and thin

provisioned pools are essentially managed the same and there are no major physical distinctions

between the pool types. Pool types are mainly distinguished by minor metadata differences.

Accordingly, pools and their constituent volumes are created instantaneously since only XIV metadata

tables are modified. Due to this inherent architectural design, the IBM XIV provides unorthodox thin-like



provisioning for all of its virtual storage. Actual thin provisioned pools are primarily differentiated by

their ability to over-provision logical capacity that is greater than the pool’s physical capacity.

Consequently and as noted previously, the latest SCSI UNMAP space reclamation benefits apply to both

regular (thick) and thin provisioned pools which form one of many important elements in the IBM XIV

Gen3’s unique storage design.

Contained within the IBM XIV Gen3 pools are volume partitions which provide the foundation of its

unique virtual storage design. XIV partitions are the fundamental building blocks used to create XIV pool

volumes and consist of primary and secondary mirrored data copies mapped to separate disks. Using

sophisticated data distribution algorithms which preserve data redundancy and performance

equilibrium, XIV partitions are automatically and dynamically distributed across all disks. Since the

distribution algorithms pseudo-randomly distribute the data across all of the spindles, there is no LUN

locality of reference to a particular disk or subset of disks, ensuring no hot-spots regardless of workload

changes over time. As a result, the IBM XIV Storage System self-tunes in response to application I/O

patterns as well as configuration and capacity changes commonly associated with physical and virtual

environments all while providing some of the storage industry’s best data protection properties.

XIV data protection employs active/active N+1 redundancy of all data modules, disks, interconnect

switches, and uninterruptible power supply (UPS) units. The XIV also contains multipath FC and iSCSI

host connections for each interface module. Three built-in UPS units protect all disks, cache and

electronics with redundant power supplies and fans, which further promote hardware and software

business critical availability and reliability. The XIV uses predetermined data distribution algorithms

which help ensure fast recovery from major and minor faults using pre-failure detection and proactive

corrective healing. In the event of module or disk failures, global spares striped across all disks quickly

redistribute data back to a fully redundant state. During such events, the performance impact is notably

minimized.

This minimal performance impact is due to the IBM XIV Storage System Gen3 extremely competitive

performance characteristics which readily meet demanding physical or virtual workloads. The XIV highly

scalable, distributed architecture provides a combined total of up to 360 GB of cache and individual

modules powered by quad-core Intel Xeon processors. Up to six dedicated host interface modules

ensure optimal, balanced data distribution among all 180 2TB or 3TB disks to help eliminate hot spots.

This data distribution feature is quite significant due to the popularity of using larger LUNs for multiple

VMs typically deployed in cloud environments. In addition to data integrity benefits, since every LUN is

striped across all 180 disks, the chance of saturating storage IO is greatly reduced when compared to

conventional architectural approaches using RAID sets and hot spares.

As conveyed thus far, the IBM XIV Gen3 unique and sophisticated storage design is very user-friendly

and provides extremely low administrative overhead. To no surprise, the actual management utilities

are categorically intuitive and extremely easy to use. So much so, that IBM continues to develop almost

identical graphical user interfaces for other members of its storage portfolio in response to customer



popularity. This unique evolutionary architecture coupled with truly uncomplicated management

software is vastly improved with the XIV Storage System Gen3 family and helps provide the following

key benefits:

Up to 4 times the throughput (10 GB/sec) of the previous XIV generation, improving performance for business intelligence, archiving and other I/O-intensive applications

Up to 3 times improved response time of the previous XIV generation, enabling faster transaction processing and greater scalability for online transaction processing, database and email applications

Industry-leading rebuild times in the event of disk or module failures (less than 60 minutes for 2 TB drives)

Innovative snapshot functionality that includes snap-of-snap, restore-of-snap and a nearly unlimited number of snapshots

Non-disruptive maintenance and upgrades

Per host/cluster QoS capability to prioritize workloads based on business criticality

Power to serve even more applications from a single system with a substantial hardware upgrade that includes an InfiniBand interconnect, larger cache (up to 360 GB of combined memory), faster SAS disk controllers and increased processing power—plus, each Gen3 interface module delivers 8 Gb FC and 1 Gb iSCSI connectivity

Boost highly random application workloads up to 4.5X with optional SSD Cache

With the XIV family’s “all-inclusive” pricing model, there are no hidden costs for multipath software or replication features — specifically, every XIV includes the following with purchase: snapshot capability, thin provisioning, asynchronous and synchronous data replication, advanced management, performance reporting, monitoring and alerting, and full support of Microsoft technologies including GeoClustering, Volume Shadow Copy Services (VSS) and MPIO

Introducing IBM XIV Storage System Gen3 Model 214 In order to maintain product leadership and increase customer satisfaction, host performance, and

support the IBM smarter planet program with energy star compliance, IBM welcomes its latest XIV

family member – the IBM XIV Storage System Gen3 Model 214. Building upon an existing strong

foundation, the newest XIV Gen3 model provides the following improvements:

New cloud and virtualization enhancements

o 10 GbE iSCSI host connectivity

o Microsoft Windows Server 2012 support for latest features such as Hyper-V, VSS, and

failover clustering

Improved storage system self-optimization and performance

o Reduce disk rebuilds by 60% - during heavy workloads, fully utilized 2 TB drives can be

rebuilt in 26 minutes

o Faster sequential read performance up to 13.7 GB/s, enabling high service level

agreements for VM cloning, multimedia, and backup workloads

o Up to 15 six-core Intel Xeon processors providing 90 physical cores or 180 logical cores

using Hyper-Threading technology



Decreased TCO through greater energy efficiency and capacity optimization

o Promotes green data center strategies by lowering carbon footprints and aggregate

power consumption by up to 16% with new power supply units

o SCSI UNMAP space reclamation for both physical and virtual hosts

Microsoft thin provisioning and storage space reclamation features Supported by both IBM XIV Gen3 Storage System models, Microsoft storage reclamation is built into its

Windows thin provisioning features, not to be confused with OS features or roles. A Windows Thin

Provisioning feature literally does not exist. Instead, the Windows thin provisioning feature refers to an

OS framework that yields supplemental thin provisioning functionality. Since Microsoft Windows 8 and

Windows 2012 both adopted the T10 SCSI Block Command 3 (SBC3) standard specification which now

provides the necessary OS thin provisioning framework, storage vendors can use the same standard to

augment customer storage monitoring and utilization without requiring additional or proprietary

applications.

In order to ensure storage vendor T10 SBC3 compliance, Microsoft also introduced new Windows Server

2012 Thin Provisioning Logo certification tests. Storage product participation involves passing numerous

thin provision functionality and performance tests which include storage space reclamation. Once the

passing test results are submitted to and reviewed by Microsoft, storage products become Windows

Thin Provisioning certified. The IBM XIV Storage System Gen3 with 11.2.x code is currently undergoing

Windows Thin Provisioning certification testing. However, the IBM XIV Gen3 has implemented the T10

SBC3 standard and thus unlocks new Windows thin provisioning benefits even though the official

Microsoft certification is still pending.

Microsoft Windows 8 and 2012 coupled with the T10 SBC3-compliant IBM XIV Storage System Gen3

(code level 11.2.x) now provide:

Thin provisioning LUN identification

Threshold notification as well as permanent resource exhaustion handles

Storage space reclamation using SCSI UNMAP storage hints

Storage device logical block mapping states

Thin provisioning LUN identification In the past, only storage administrators could tell whether a volume was thick or thin provisioned. With the latest Windows 2012 thin provisioning framework enhancements, all Windows users can determine the LUN identification. To verify the new thin provisioning LUN identification, launch the Windows Defragment and Optimize Drives utility located in the Administrative Tools. Under the Media type column of the Optimize Drives window, notice IBM XIV Gen3 thin volumes show up as ‘Thin provisioned drives’ as depicted in the screenshot below.



Threshold notification Pool and volume threshold alerts which were formerly detectable using IBM XIV storage management tools are now also available in Windows Server 2012 too. When a threshold is exceeded on the IBM XIV Storage System Gen3, the alerts are triggered as Windows System Event logs using Windows temporary or permanent resource exhaustion handles which allow the SAN storage to directly update the OS. Fortunately, there is no need to configure anything for the IBM XIV Gen3 to immediately profit from the new Windows Server 2012 resource exhaustion handles. The XIV only supports and enables permanent resource exhaustion by default. For Windows System Event Log threshold notification, storage administrators can define pool alert threshold values using the XIVGUI as depicted below.

XIVGUI > Systems > System Settings > Pool Alerts Thresholds



Once the XIV pool alert thresholds are exceeded, there are two types of events that are triggered. At the storage level, the XIVGUI reports the pool alerts at the bottom left of the system view as well as the monitor alerts view.

Note: At the bottom left of the XIVGUI, mouseover the red alert warning icon to view additional details.

XIVGUI > View > Monitor > Alerts

Note: For testing purposes, the XIVGUI alerts above were triggered with pool threshold values lower than the default values.

At the OS level, the Windows System Event log will report the following system events:

Event ID General Information

144 Threshold notification without additional information

145 Threshold notification without specific information

146 Threshold notification with used LUN capacity and available LUN capacity information

147 Threshold notification with used LUN capacity and available pool capacity information

148 Threshold notification with used pool capacity and available LUN capacity information

149 Threshold notification with used pool capacity and available pool capacity information



Example System Event Log ID 144 for XIV thin provisioned volume threshold trigger:

Log Name: System Source: disk Date: 1/10/2013 10:22:29 PM Event ID: 144 Task Category: None Level: Warning Keywords: Classic User: N/A Computer: isvx8.xiv.tucson.ibm.com Description: Disk 13 has crossed a capacity utilization threshold.

If a permanent resource exhaustion threshold is reached, the following also occurs:

A Windows System Event log ID 150 is generated

The thin-provisioned XIV volume is taken offline and becomes unavailable until the storage administrator increases the volume capacity and/or pool capacity

Space reclamation using SCSI UNMAP By default, Windows 8 and 2012 enable real-time space reclamation using SCSI UNMAP. As previously

described, anytime a large file is deleted, multi-level space reclamation occurs. However, this may

impact performance depending on how often users or applications delete large files. Proper planning

should help to alleviate any real-time space reclamation performance impacts and can be accomplished

establishing performance baselines.

If Windows space reclamation planning identifies a high probability of performance impact, consider the

following option:

Real-time space reclamation can be disabled for large file deletions in the Windows registry.

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\FileSystem

Set the DisableDeleteNotification value to 1

Note: This Windows registry setting affects all LUNs. For further information, visit the following

website:

Plan and Deploy Thin Provisioning

http://technet.microsoft.com/en-us/library/jj674351.aspx

http://technet.microsoft.com/en-us/library/jj674351.aspx



Storage device logical block mapping states Windows Server 2012 now has the ability to detect the storage device logical block mapping states of thin provisioned LUNs. For storage management operations, Windows contains a new API implementation known as an IOCTL DSM allocation which retrieves the logical block address (LBA) status of thin provisioned LUNs. All logical blocks are grouped into slabs or clusters which are classified into mapped, de-allocated or anchored states which Windows considers unmapped states. This is transparent to users and ensures the Windows thin provisioning framework, which includes space reclamation, performs as intended.

For further information about the Windows 2012 Thin Provisioning features, reference the following website:

http://msdn.microsoft.com/en-us/library/windows/hardware/hh770514.aspx

Microsoft Windows 2012 Data Deduplication In addition to the space reclamation benefits associated with SCSI UNMAP, Microsoft Windows Server

2012 introduced a new Data Deduplication feature that operates at the storage block level. Windows

Data Deduplication is very compelling and highly effective for reducing file server, user profile, Hyper-V

library, and other widespread data capacity requirements. The Windows Data Deduplication feature also

provides compounded space reclamation benefits. After Data Deduplication is configured and enabled

for a volume, not only does the file system reclaim space but Windows Data Deduplication also triggers

SCSI UNMAP to reclaim the SAN storage free space. However, unlike the default space reclamation

behavior during Windows Server 2012 file deletions, data deduplication does not take place in real-time.

It can be easily automated through scheduling as well as manually triggered but falls into the post-

processing category.

Even though Windows Data Deduplication is designed for low memory and CPU use and the default

deduplication policy settings are usually sufficient to achieve excellent savings without impacting normal

server workloads, it is post-processed to help eliminate potential performance impacts. Basically, a

server needs one CPU core and 350 MB of free memory to run a deduplication job per volume which

scales with additional CPU core processors and available memory. CPU and memory resources are used

for a sub-file chunking algorithm which enables Data Deduplication to find and remove duplicate data by

breaking files into variable sizes, typically 32-128 KB, all without compromising file integrity or fidelity.

The file segments form chunk boundaries which are defined by a Rabin fingerprint-based sliding window

hash and eventually compressed and placed into stores. These chunk stores are hidden in the volume

root System Volume Information (SVI) folder. In the end, the files are replaced by reparse points which

map the data streams and chunks in order to reconstitute the files if later required.

In general, NTFS volumes that contain infrequently modified files are the best candidates for Windows

Data Deduplication. Additionally, customers may elect to automate or run deduplication jobs after peak

business production hours to ensure maximum performance and optimization. For in-depth information

about Windows 2012 Data Deduplication, reference the following website:




Data Deduplication Overview

http://technet.microsoft.com/en-us/library/hh831602.aspx

Microsoft Windows 2012 Data Deduplication support summary Windows 2012 Data deduplication supports:

Windows Server operating systems beginning with Windows Server 2012

NTFS data volumes including regular cluster disk resources

SQL Server and Exchange Server backup volumes

Windows 2012 Data Deduplication does not support:

System or boot volumes

Remote mapped or remote mounted drives

Cluster shared volumes (CSV)

Live VM files

Active SQL Server databases

Files that are open and constantly changing or have high I/O requirements

Setting hard quotas on volume root folders that have deduplication enabled

Note: Microsoft suggests not using Windows Data Deduplication for Hyper-V hosts, VDI VHDs, WSUS,

SQL Server or Exchange Server, and large files that are 1 TB or greater in size.

Windows omits the following from data deduplication:

Encrypted files

Windows system state data files

Files smaller than 32 KB

Non data deduplication reparse points

Files with metadata extended attributes

Note: User-defined folders and file extensions can also be excluded when configuring volume data

deduplication.

During the planning stages, consider using the Data Deduplication Savings Evaluation Tool to help

determine which server solutions benefit most. This Windows 2012 native utility is extracted to

%systemroot%\system32 once the Data Deduplication feature is installed as described in the next

section. For further information about the Data Deduplication Savings Evaluation Tool, reference the

following website:

http://technet.microsoft.com/en-us/library/hh831602.aspx



Evaluate savings with the Deduplication Evaluation Tool (DDPEVAL.exe)

http://blogs.technet.com/b/klince/archive/2012/08/09/evaluate-savings-with-the-deduplication-

evaluation-tool-ddpeval-exe.aspx

To launch the utility from the command line, run the following:

C:\Windows\System32>ddpeval /?

Data Deduplication Savings Evaluation Tool

Copyright (c) 2012 Microsoft Corporation. All Rights Reserved.

Usage:

ddpeval <path> [/V] [/E] [/O:<filepath/name>]

<path> Drive, directory or share to be evaluated.

/V Verbose.

/E Process files with extended attributes.

/O Output to file.

<filepath/name> Output file name.

/? Help.

C:\Windows\System32>ddpeval f:\ /V /O:c:\ddpeval_f_test1.txt



Progress: 95%

C:\Windows\System32>ddpeval f:\ /V /O:c:\ddpeval_f_test1.txt



Evaluated folder: f:\

Evaluated folder size: 647.96 GB

Files in evaluated folder: 16438

Processed files: 8638

Processed files size: 647.91 GB

Optimized files size: 335.88 GB

Space savings: 312.03 GB

Space savings percent: 48

Optimized files size (no compression): 366.41 GB

Space savings (no compression): 281.50 GB

Space savings percent (no compression): 43

Note: Test durations for the Data Deduplication Savings Evaluation Tool vary based on volume

size, file quantities, file types, file or folder exclusions, file sizes, and server resource availability.

Windows Server 2012 Data Deduplication installation steps Perform the following steps to install and test the new Windows Server 2012 Data Deduplication

feature:

http://blogs.technet.com/b/klince/archive/2012/08/09/evaluate-savings-with-the-deduplication-evaluation-tool-ddpeval-exe.aspx




1. In the left pane of the Server Manager, highlight Local Server, expand the top right menu bar.

Under Manage, select Add Roles and Features.

2. In the Before you begin section of the Add Roles and Features Wizard, click Next.



3. In the Select installation type of the Add Roles and Features Wizard, select Role-based or

feature-based installation and click Next.

4. In the Select destination server section of the Add Roles and Features Wizard, choose Select a

server from the server pool. Make sure to select the desired server for adding the new feature

and click Next.



5. In the Select server roles section of the Add Roles and Feature Wizard, expand File and Storage

Services. Expand File and iSCSI Services and select Data Deduplication.

6. Upon selecting Data Deduplication above, the following feature dependency popup appears.

Click Add Features.



7. Review the Server Role selection and click Next.

8. In the Features section of the Add Roles and Features Wizard, simply click Next.



9. In the Confirmation Installation selections section of the Add Roles and Features Wizard,

review the selected role and its dependencies and click Install.

10. In the Results section of the Add Roles and Features Wizard, confirm the successful installation

and click Close.



11. In the left pane of the Server Manager, highlight File and Storage Services and click on

Volumes. Right click the desired volume and select Configure Data Deduplication.

12. In the Deduplication Settings dialog box, select “Enable data deduplication” and add other

desired parameters.

Note: If testing deduplication, place a value of 0 days in the “Deduplicate files older than…” field

as highlighted above.



13. Once deduplication is enabled on the volume, take note of the “Deduplication Rate” and

“Deduplication Savings” values. The Deduplication values start at 0% until the deduplication job

is executed as scheduled or immediately using the appropriate PowerShell commands.

Triggering immediate Windows 2012 Data Deduplication for a NTFS volume Once Windows 2012 Data Deduplication is enabled, perform the following steps to manually trigger the

capacity savings outside of default or user-defined schedules:

Note: Prior to forcing deduplication to run on a volume, record the volume used capacity in the XCLI or

XIVGUI. Capturing the before and after values provides the exact space reclamation savings. Of course,

the deduplication durations will vary as previously noted.

From the XCLI:

Before: XIV 7820045>>vol_list

Name Size (GB) Master Name Consistency Group Pool Creator Used Capacity (GB)

hyper_v_library_vol_01 2219 xiv_UNMAP_test_thin_pool_01 admin 695

After: XIV 7820045>>vol_list

Name Size (GB) Master Name Consistency Group Pool Creator Used Capacity (GB)

hyper_v_library_vol_01 2219 xiv_UNMAP_test_thin_pool_01 admin 368



Note: By running the XCLI vol_list command multiple times, the storage administrator can periodically

check the Windows Deduplication and SCSI UNMAP progress. The XCLI volume "Used Capacity (GB)"

should decrease in value as seen above.

From the Windows PowerShell (PS) using administrator privileges:

PS C:\Users\administrator.XIV> Start-DedupJob L: -type Optimization

Type ScheduleType StartTime Progress State Volume

---- ------------ --------- -------- ----- ------

Optimization Manual 0 % Queued L:

PS C:\Users\administrator.XIV> Get-DedupJob


---- ------------ --------- -------- ----- ------

Optimization Manual 3:23 PM 61 % Running L:

PS C:\Users\administrator.XIV> Get-DedupJob


---- ------------ --------- -------- ----- ------

Optimization Manual 3:23 PM 96 % Running L:

Note: The data deduplication process can also be checked using the following command. Just like the

other XCLI and Windows PS examples, it can be run multiple times to view the progress. The

"SavedSpace" and "SavingsRate" values should increase over time.

From the Windows PowerShell (PS) using administrator privileges:

PS C:\Users\administrator.XIV> Get-DedupVolume L:

Enabled SavedSpace SavingsRate Volume

------- ---------- ----------- ------

True 245.48 GB 38 % L:

PS C:\Users\administrator.XIV> Get-DedupVolume L:

Enabled SavedSpace SavingsRate Volume

------- ---------- ----------- ------

True 288.75 GB 45 % L:



Conclusion The IBM XIV Storage System Gen3, including the latest model and 11.2 microcode, now offers enhanced

efficiency and more advanced features which allow customers to tap into the latest Microsoft Windows

Server 2012 thin provisioning benefits. These new Windows framework benefits include thin

provisioning LUN identification, threshold notification, SCSI UNMAP space reclamation, and storage

device logical block mapping states. Thin provisioning dead space deficiencies are no longer a concern

and IT administrators can manage data capacity demands and control storage sprawl much better.

Moreover, Microsoft cloud administrators realize multi-level physical and virtual space reclamation

benefits which help mitigate Hyper-V data capacity concerns. Finally, these storage optimization and

efficiency advances combined with Microsoft Windows Server 2012 Data Deduplication allow IT

administrators to maximize storage ROI all while helping to reduce costs.



Resources The following websites provide useful references to supplement the information contained in this

paper:

IBM XIV Storage System series

http://www-03.ibm.com/systems/storage/disk/xiv/index.html

T10 SBC3 Specification

http://www.t10.org/cgi-bin/ac.pl?t=f&f=sbc3r27.pdf

New API allows apps to send "TRIM and Unmap" hints to storage media (Windows) http://msdn.microsoft.com/en-us/library/windows/desktop/hh848053%28v=vs.85%29.aspx

Thin Provisioning in Windows Server 2012


Introduction to Data Deduplication in Windows Server 2012

http://blogs.technet.com/b/filecab/archive/2012/05/21/introduction-to-data-deduplication-in-

windows-server-2012.aspx

About Data Deduplication (Windows)

http://msdn.microsoft.com/en-us/library/windows/desktop/hh769303%28v=vs.85%29.aspx

Evaluate savings with the Deduplication Evaluation Tool (DDPEVAL.exe)

http://blogs.technet.com/b/klince/archive/2012/08/09/evaluate-savings-with-the-deduplication-

evaluation-tool-ddpeval-exe.aspx

http://www-03.ibm.com/systems/storage/disk/xiv/index.html

http://www.t10.org/cgi-bin/ac.pl?t=f&f=sbc3r27.pdf



http://blogs.technet.com/b/filecab/archive/2012/05/21/introduction-to-data-deduplication-in-windows-server-2012.aspx

http://blogs.technet.com/b/filecab/archive/2012/05/21/introduction-to-data-deduplication-in-windows-server-2012.aspx






Trademarks and special notices

© Copyright IBM Corporation 2012.

References in this document to IBM products or services do not imply that IBM intends to make them

available in every country.

IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of International Business

Machines Corporation in the United States, other countries, or both. If these and other IBM trademarked

terms are marked on their first occurrence in this information with a trademark symbol (® or ™), these

symbols indicate U.S. registered or common law trademarks owned by IBM at the time this information

was published. Such trademarks may also be registered or common law trademarks in other countries. A

current list of IBM trademarks is available on the Web at "Copyright and trademark information" at

www.ibm.com/legal/copytrade.shtml.

Java and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or

its affiliates.

Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the

United States, other countries, or both.

Intel, Intel Inside (logos), MMX, and Pentium are trademarks of Intel Corporation in the United States,

other countries, or both.

UNIX is a registered trademark of The Open Group in the United States and other countries.

Linux is a trademark of Linus Torvalds in the United States, other countries, or both.

Other company, product, or service names may be trademarks or service marks of others.

Information is provided "AS IS" without warranty of any kind.

All customer examples described are presented as illustrations of how those customers have used IBM

products and the results they may have achieved. Actual environmental costs and performance

characteristics may vary by customer.

Information concerning non-IBM products was obtained from a supplier of these products, published

announcement material, or other publicly available sources and does not constitute an endorsement of

such products by IBM. Sources for non-IBM list prices and performance numbers are taken from publicly

available information, including vendor announcements and vendor worldwide homepages. IBM has not

tested these products and cannot confirm the accuracy of performance, capability, or any other claims

related to non-IBM products. Questions on the capability of non-IBM products should be addressed to the

supplier of those products.

All statements regarding IBM future direction and intent are subject to change or withdrawal without

notice, and represent goals and objectives only. Contact your local IBM office or IBM authorized reseller

for the full text of the specific Statement of Direction.

Some information addresses anticipated future capabilities. Such information is not intended as a

definitive statement of a commitment to specific levels of performance, function or delivery schedules with

respect to any future products. Such commitments are only made in IBM product announcements. The

http://www.ibm.com/legal/copytrade.shtml



information is presented here to communicate IBM's current investment and development activities as a

good faith effort to help with our customers' future planning.

Performance is based on measurements and projections using standard IBM benchmarks in a controlled

environment. The actual throughput or performance that any user will experience will vary depending

upon considerations such as the amount of multiprogramming in the user's job stream, the I/O

configuration, the storage configuration, and the workload processed. Therefore, no assurance can be

given that an individual user will achieve throughput or performance improvements equivalent to the

ratios stated here.

Photographs shown are of engineering prototypes. Changes may be incorporated in production models.

Any references in this information to non-IBM websites are provided for convenience only and do not in

any manner serve as an endorsement of those websites. The materials at those websites are not part of

the materials for this IBM product and use of those websites is at your own risk.

Documents

Microsoft Windows Server 2012 thin provisioning space ... · Microsoft Windows Server 2012 thin provisioning space ... VMware vSphere 5.0 storage APIs for ... such as those now included