Further. Forward. Faster. 6. Microsoft Private Cloud Solution for SAP: Configuration of Cluster Shared Volumes, VHDX, Disks & LUNs Authors Cameron Gardiner, Microsoft Senior Program Manager – SAP Technical Reviewers Juergen Thomas, Principal Lead Program Manager Hermann Daeubler, Microsoft Senior Program Manager – SAP Summary This document provides guidance for customers and partners to deploy SAP on a Microsoft Private Cloud Solution for SAP. This chapter discusses the Configuration of Cluster Shared Volumes, VHDX, Disks & LUNs CSV, VHDX & LUNs Table of Contents Table of Contents ........................................................................................................ 2 1 Summary ................................................................................................................ 4 2 Example Configuration – SQL Server VM ........................................................... 5 2.1 SQL SERVER VM CONFIGURATION DETAILS ......................................................................................... 5 2.1.1 Physical Server.............................................................................................................................. 5 2.1.2 Virtual Machine ............................................................................................................................ 5 2.1.2.1 Roles & Features Required for SQL Server ................................................................... 5 2.1.2.2 Windows Server 2012 Server Core................................................................................... 6 2.1.3 SQL Server Named Instance & Datafile Configuration ................................................. 6 2.1.4 Cluster Shared Volume & Virtual Hard Disk ..................................................................... 8 2.1.5 SAN Configuration...................................................................................................................... 8 3 Example Configuration – SAP Application Server VM ...................................10 3.1.1 Physical Server........................................................................................................................... 10 3.1.2 Virtual Machine ......................................................................................................................... 10 3.1.2.1 Roles & Features Required for SAP Application Server ........................................ 10 3.1.2.2 Windows Server 2012 Server Core................................................................................ 10 3.1.3 SAP Application Server Disk Configuration.................................................................... 10 3.1.4 Cluster Shared Volume & Virtual Hard Disk .................................................................. 13 3.1.5 SAN Configuration................................................................................................................... 13 4 Example Configuration – SAP ASCS SPOF VM ................................................15 4.1.1 Physical Server........................................................................................................................... 15 4.1.2 Virtual Machine ......................................................................................................................... 15 4.1.2.1 Roles & Features Required for SAP Application Server ........................................ 15 4.1.2.2 Virtual Machine Affinity..................................................................................................... 15 4.1.2.3 Windows Server 2012 Server Core................................................................................ 15 4.1.3 SAP ASCS Cluster Disk Configuration .............................................................................. 16 4.1.4 Cluster Shared Volume & Virtual Hard Disk .................................................................. 18 4.1.5 SAN Configuration................................................................................................................... 18 4.1.6 Shared Disk for SAP ASCS .................................................................................................... 18 4.1.6.1 iSCSI .......................................................................................................................................... 18 4.1.6.2 Virtual Fibre Channel (vHBA)........................................................................................... 18 5 Cluster Shared Volumes .....................................................................................21 5.1 HOW MANY VIRTUAL MACHINES PER CSV?...................................................................................... 21 5.2 RENAMING CSV PATHS ......................................................................................................................... 21 5.3 CLUSTER SHARED VOLUME CACHE ...................................................................................................... 22 5.3.1 CsvEnableBlockCache ............................................................................................................. 22 5.3.2 SharedVolumeBlockCacheSizeInMB ................................................................................. 22 SAP on Microsoft Private Cloud Page 2 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 5.3.3 6 How to Set CSV Cache ........................................................................................................... 22 Virtual Hard Disk Configuration - VHDX .........................................................23 6.1 VHDX TYPES ........................................................................................................................................... 23 6.1.1 Fixed, Dynamic & Differencing ........................................................................................... 23 6.1.2 Windows 2008 VHD vs. Windows Server 2012 VHDX................................................ 23 6.2 MBR, GPT, NTFS AUS SIZE AND OTHER OPTIONS .......................................................................... 23 7 Appendix 2 – Links Page ....................................................................................25 SAP on Microsoft Private Cloud Page 3 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 1 Summary Windows Server 2012 Hyper-V 3.0 is a powerful integrated Virtualization platform. Microsoft Private Cloud for SAP systems is a tailored solution optimized and simplified to enable customers and partners to quickly deploy SAP on Hyper-V. Configuration of SQL Server datafiles and logfiles, Virtual Hard Disks and placing these onto Cluster Shared Volumes is often confusing for customers. This document illustrates and simplifies this configuration through diagrams and tables. Customers are encouraged to copy the Visio diagrams and Excel spreadsheets contained in this document and customize to their specific configuration. There are mandatory requirements that must be followed: 1. Windows Server 2012 Hyper-V 3.0 physical cluster servers must be 2 processor systems with a minimum of 384GB RAM. 384GB is strictly mandatory and customers must not attempt to deploy this reference architecture on systems with less than 384GB RAM 2. SQL Server systems require different numbers of datafiles depending on the size of the system. Small sized systems: 4 data files. Medium sized systems: 8 data files. Larger sized systems: 16 data files. Xtra large sized systems: 32 for server with more than 160 or more Logical Processors. 3. SQL Server databases only require 1 large log file per database Small systems start with Logfile size 5 - 10 GB Medium systems start with Logfile size 50 - 100 GB Large systems start with Logfile size 300 - 500 GB 4. Tempdb must be configured with at least 4 datafiles and one log file. These should be “colocated” on the same Virtual Hard Disks and Cluster Shared Volumes as the SAP DB datafiles 5. Large busy systems such as ECC, BW, CRM and SCM can have up to 4 SAP DB datafiles per Virtual Hard Disk 6. Very small SAP systems such as non-productive EP, XI, GCR or Solman may consolidate many databases onto a single VM. Each SAP component requires a separate Named Instance of SQL Server. Small systems with very low workload may consolidate all SAP DB datafiles onto a set of VHDXs 7. SQL Server AlwaysOn is the preferred method of achieving high availability for the database layer. Traditional SQL Server clustered are supported for reverse compatibility reasons only 8. A single large C: Drive no less than 120GB must be created for SAP application servers or SAP ASCS cluster servers. Do not partition the Virtual Hard Disk into C:, D:, E: drives 9. SAP and SQL Servers are not to be used as file servers SAP on Microsoft Private Cloud Page 4 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 2 Example Configuration – SQL Server VM SAP on Microsoft Private Cloud reference deployment can be decomposed into several simple layers thereby allowing customers and partners to correctly configure Microsoft Private Cloud infrastructure for best performance and availability. The layers are: 1. Physical Servers running Windows Server 2012 Datacenter, Hyper-V and Windows Cluster with Cluster Shared Disks 2. Virtual Machines running Windows Server 2012, SQL Server 2012 SP1 or higher and SAP Host Monitoring Agent 3. SQL Server Named Instance running one or more SAP databases with a specified number of datafiles, tempdb files and log file 4. Cluster Shared Volumes holding one or more sets of Virtual Hard Disk (VHDX) files 5. Physical SAN Layer contains physical disks aggregrated and presented to the Physical Servers running Windows Server 2012 as a number of LUNs 2.1 SQL Server VM Configuration Details SQL Server Virtual Machines require some specific configuration to achieve good performance. The following sections should be reviewed in conjuction with the diagram on the following page. This section depicts a dedicated SQL Server such as that used on a three tier system. When installing two tier systems (SQL Server and SAP on the same VM) simply combine the configuration details in section 2 and 3 of this document. 2.1.1 Physical Server See the Chapter 6 and 9 for information on the physical server configuration 2.1.2 Virtual Machine Windows Server 2012 Standard Edition or greater should be used for running SQL Server. See the Sizing Chapter for information on the Virtual Machine sizing. 2.1.2.1 Roles & Features Required for SQL Server If a SQL Server VM is part of an AlwaysOn HA/DR configuration then Windows Cluster Feature must be installed. Otherwise only the minimum Roles and Features should be installed. SQL Server Installer required .Net 4.0 or 3.5.1. SAP Software Provisioning Manager (SWPM) may install File Services Feature onto a dedicated SQL Server during installation. This is not required for the operation of the SAP system and should be removed once SWPM procedure is completed. SAP on Microsoft Private Cloud Page 5 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 2.1.2.2 Windows Server 2012 Server Core To reduce patching requirements it is supported to run SQL Server on a Windows Server 2012 core server1 See the Security Chapter for more information on hardening the Guest Virtual Machines 2.1.3 SQL Server Named Instance & Datafile Configuration To allow consolidation of multiple databases onto the SQL Server VM, SQL Server should be installed as a Named Instance. Larger SQL Server systems might require a dedicated VM. Small sized systems: 4 data files. Systems usually run on servers which have between 4 and 8 Logical Processors2. Medium sized systems: At least 8 data files. Systems usually run on servers which have between 8 and 16 Logical Processors. Larger sized systems: A minimum of 16 data files. Systems which today run on hardware between 16 and 120 Logical Processors. Xtra large sized systems: 32 for server with more than 160 or more Logical Processors. Small systems start with Logfile size 5 - 10 GB Medium systems start with Logfile size 50 - 100 GB Large systems start with Logfile size 300 - 500 GB Only one Logfile should be created – SQL Server will only use one Logfile at one time Check and eventually set autogrowth rate to 10% To achieve a good balance between performance and administration it is recommended to place four SAP DB datafiles per Virtual Hard Disk. Tempdb files can be placed on the same Virtual Hard Disk as the SAP DB datafiles/logfile. Traditional Clustering of SQL Server is supported on Microsoft Private Cloud for SAP via the use of the vHBA or iSCSI. To achieve High Availability at the database level SQL Server 2012 AlwaysOn is highly recommended. Support for traditional SQL Server clustering is for reverse compatibility reasons only. New systems should be installed with SQL Server AlwaysOn. See the High Availability Chapter for more information If SQL Server is running on Windows Core a “management station” with SQL Server Management Studio must be available as well as remote server management 1 http://blogs.msdn.com/b/saponsqlserver/archive/2010/09/28/windows-2008-r2-groupsprocessors-sockets-cores-threads-numa-nodes-what-is-all-this.aspx 2 SAP on Microsoft Private Cloud Page 6 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs VM Host Name Home Node (1 – 5) DB File Path on VM Virtual Hard Disk Name Cluster Shared Volume VHDX Path on Host SAN LUN CSV Cache sapeccdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapdb1 sapsolman1 sapdb2 sapdb2 sapdb2 sapdb2 sapdb2 sapdb2 sapdb2 sapdb2 sapdb2 sapdb2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 3 3 3 3 3 3 3 3 3 3 OS disk C:\Program Files\SQLServer F:\ECCDATA1\ECCDATA1.mdf F:\ECCDATA2\ECCDATA2.ndf F:\ECCDATA3\ECCDATA3.ndf F:\ECCDATA4\ECCDATA4.ndf G:\ECCDATA5\ECCDATA5.ndf G:\ECCDATA6\ECCDATA6.ndf G:\ECCDATA7\ECCDATA7.ndf G:\ECCDATA8\ECCDATA8.ndf H:\ECCDATA9\ECCDATA9.ndf H:\ECCDATA10\ECCDATA10.ndf H:\ECCDATA11\ECCDATA11.ndf H:\ECCDATA12\ECCDATA12.ndf I:\ECCDATA13\ECCDATA13.ndf I:\ECCDATA14\ECCDATA14.ndf I:\ECCDATA15\ECCDATA15.ndf I:\ECCDATA16\ECCDATA16.ndf L:\ECCLOG1\ECCLOG1.ldf F:\TEMPDB1\TEMPDB1.mdf G:\TEMPDB2\TEMPDB2.ndf H:\TEMPDB3\TEMPDB3.ndf I:\TEMPDB4\TEMPDB4.ndf L:\ TEMPLOG1\TEMPLOG1.ldf OS disk C:\Program Files\SQLServer F:\SLMDATA1\SLMDATA1.mdf G:\SLMDATA2\SLMDATA2.ndf H:\SLMDATA3\SLMDATA3.ndf I:\SLMDATA4\SLMDATA4.ndf L:\SLMLOG1\SLMLOG1.ldf F:\TEMPDB1\TEMPDB1.mdf G:\TEMPDB2\TEMPDB2.ndf H:\TEMPDB3\TEMPDB3.ndf I:\TEMPDB4\TEMPDB4.ndf L:\ TEMPLOG1\TEMPLOG1.ldf sapdb1_ECC_C_DISK.VHDX sapdb1_ECC_F_DISK.VHDX sapdb1_ECC_F_DISK.VHDX sapdb1_ECC_F_DISK.VHDX sapdb1_ECC_F_DISK.VHDX sapdb1_ECC_G_DISK.VHDX sapdb1_ECC_G_DISK.VHDX sapdb1_ECC_G_DISK.VHDX sapdb1_ECC_G_DISK.VHDX sapdb1_ECC_H_DISK.VHDX sapdb1_ECC_H_DISK.VHDX sapdb1_ECC_H_DISK.VHDX sapdb1_ECC_H_DISK.VHDX sapdb1_ECC_I_DISK.VHDX sapdb1_ECC_I_DISK.VHDX sapdb1_ECC_I_DISK.VHDX sapdb1_ECC_I_DISK.VHDX sapdb1_ECC_L_DISK.VHDX sapdb1_ECC_F_DISK.VHDX sapdb1_ECC_G_DISK.VHDX sapdb1_ECC_H_DISK.VHDX sapdb1_ECC_I_DISK.VHDX sapdb1_ECC_L_DISK.VHDX sapdb2_ECC_C_DISK.VHDX sapdb2_SLM_F_DISK.VHDX sapdb2_SLM_G_DISK.VHDX sapdb2_SLM_H_DISK.VHDX sapdb2_SLM_I_DISK.VHDX sapdb2_SLM_L_DISK.VHDX sapdb2_SLM_F_DISK.VHDX sapdb2_SLM_G_DISK.VHDX sapdb2_SLM_H_DISK.VHDX sapdb2_SLM_I_DISK.VHDX sapdb2_SLM_L_DISK.VHDX C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume1\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume1\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume1\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume1\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume2\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume2\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume2\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume2\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume3\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume3\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume3\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume3\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume4\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume4\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume4\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume4\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume5\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume1\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume2\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume3\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume4\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume5\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume1\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume2\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume3\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume4\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume5\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume1\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume2\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume3\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume4\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume5\SLM\<vm name>\Virtual Hard Disks\ 6 1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4 5 1 2 3 4 5 6 1 2 3 4 5 1 2 3 4 5 Yes No No No No No No No No No No No No No No No No No No No No No No Yes No No No No No No No No No No Table shows two SAP systems running on separate VMs on a 5 node Microsoft Private Cloud Infrastructure. ECC 6.5 system SID = ECC and a Solution Manager system SID = SLM. ECC 6.6 is a large database and requires 16 datafiles. The Solution Manager system is small and only requires four datafiles. To ensure IO is balanced on the Microsoft Private Cloud Infrastructure for SAP every system must have a minimum of four VHDX, one placed on each of the four Cluster Shared Volumes SAP on Microsoft Private Cloud Page 7 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 2.1.4 Cluster Shared Volume & Virtual Hard Disk Windows Server 2012 Cluster Shared Volumes can support many VHDXs and many VHDXs from different VMs. 10-25 Virtual Machines can share the same set of CSV without performance impact when configured correctly. To balance SQL Server IO across sufficient LUNs the Microsoft Private Cloud reference architecture for SAP systems required four CSVs for SQL Datafiles and one CSV for SQL Log files. It is not supported to place all SQL Datafiles and the Log file on one single CSV (leveraging only one LUN). Up to four SAP database SQL Datafiles can be placed on one Virtual Hard Disk (VHDX). Distributing SQL Server datafiles onto separate VHDX on separate Cluster Shared Volumes will improve IO performance and ensure that the IO load on all the Cluster Shared Volumes is balanced. A small SAP system with only 4 datafiles should have one datafile per VHDX. One VHDX should be placed on each Cluster Shared Volume thereby balancing the IO load on the Cluster Shared Volumes and LUNs. A small SAP system with only 8 datafiles should have two datafile per VHDX. One VHDX should be placed on each Cluster Shared Volume thereby balancing the IO load on the Cluster Shared Volumes and LUNs. SAP Landscapes normally have a number of very low SQL transaction volume systems such as Enterprise Portal, GRC, XI, Solution Manager etc. The load on SQL Server is very small. Such systems can be consolidated as multiple Named Instances on one Virtual Machine. In such cases it is possible to consolidate 4-6 SAP components such as EP, GRC etc onto one VM provided the SQL Server workload is observed to be very low. In this scenario each SAP component has its own SQL Server Named Instance with 4-6 Named Instances consolidated on one VM. The datafiles for all of the Named Instances can be consolidated into the four Virtual Hard Disks (F:, G:, H: and I:) for SQL datafiles and one Virtual Hard Disk for logfiles. 2.1.5 SAN Configuration Each Cluster Shared Volume requires a separate LUN at the SAN level. The LUNs for the CSV must be presented to all nodes in the Hyper-V 3.0 cluster. RAID 5 (or similar) is suitable for the Cluster Shared Volumes holding the Virtual Hard Disks that in turn hold SQL Server Datafiles. RAID 1 (or similar) should be used for the Cluster Shared Volumes holding the Virtual Hard Disks that in turn hold SQL Server Log. SAP on Microsoft Private Cloud Page 8 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs MS Private Cloud Physical Layer Windows 2012 Hyper-V 3.0 Cluster VM Layer Large VM = 16vCPU, 192GB vRAM In this example this VM is only running a single SQL Server instance with a single database for a 1,500 concurrent user ECC 6.6 system. DB size is 1TB fully PAGE compressed SAP DB Layout: 16 datafiles, 1x 200GB Log Tempdb Layout: 4x 1GB datafiles, 1x 1GB Log VHDX: 4 SQL datafiles per VHDX CSV: 5 CSV shared by all SAP systems LUN: 5x Data + Log LUN 2TB ECC 6.5 DB Datafile Layer Datafile 9 Datafile 5 Datafile 6 Datafile 1 Datafile 2 Datafile 3 Datafile 4 Datafile 7 Datafile 8 tempdb 1 tempdb 2 sapdb1_ECC_F_DISK.VHDX sapdb1_ECC_G_DISK.VHDX Datafile 13 Datafile 10 Datafile 11 Datafile 15 Datafile 12 Datafile 14 Datafile 16 tempdb 4 tempdb 3 sapdb1_ECC_H_DISK.VHDX sapdb1_ECC_I_DISK.VHDX SAP DB Log File 200GB Temp log sapdb1_ECC_L_DISK.VHDX Virtual Hard Disk -> CSV Layer C:\ClusterStorage\ Volume1 C:\ClusterStorage\ Volume2 C:\ClusterStorage\ Volume3 C:\ClusterStorage\ Volume4 C:\ClusterStorage\ Volume5 { LUN 1 LUN 2 SAP on Microsoft Private Cloud LUN 3 LUN 4 Page 9 of 25 LUN 5 SAN Layer Created: 15-Mar-16 CSV, VHDX & LUNs 3 Example Configuration – SAP Application Server VM SAP on Microsoft Private Cloud reference deployment can be decomposed into several simple layers thereby allowing customers and partners to correctly configure Microsoft Private Cloud infrastructures. SAP Application Server Virtual Machines require some specific configuration to achieve good performance. The following sections should be reviewed in conjuction with the diagram on the following page. This section depicts a dedicated SAP Application Server (non-clustered) such as that used on a three tier system. When installing two tier systems (SQL Server and SAP on the same VM) simply combine the configuration details in section 2 and 3 of this document. 3.1.1 Physical Server See the Sizing Chapter for information on the physical server configuration 3.1.2 Virtual Machine Windows Server 2012 Standard Edition or greater should be used for running a SAP Application Server. See the Sizing Chapter for information on the Virtual Machine sizing. 3.1.2.1 Roles & Features Required for SAP Application Server The SAP Application Server requires the File Services Role in order to share \SAPLOC. Roles or Features are typically required 3.1.2.2 No other Windows Server 2012 Server Core It is not supported at this time to run the SAP application server on Windows Server Core or Hybrid. The only supported “Shell” is Full GUI. See the Security Chapter for more information on hardening the Guest Virtual Machines 3.1.3 SAP Application Server Disk Configuration The SAP ABAP and Java Application Servers do not cause significant IO on their own file systems. Typically the IO activity on the \usr\sap file system is limited to writing trace files and other activity. The vast majority of the IO on a SAP system is caused by the Database. In order to simplify the configuration one large C: Drive should be used to hold the following components: 1. Windows Server 2012 Guest Operating System 2. Pagefile.sys 3. C:\usr\SAP\<SID>\<instance> 4. SAP Host Agent C:\Program Files\SAP\HostCtrl\exe SAP on Microsoft Private Cloud Page 10 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs Typically 120GB Virtual Hard Disk partitioned into one single large C: Drive is sufficient for this purpose. It is highly unlikely that the disk would ever fill up on a SAP application server as no files are stored on SAP Application Servers. In the majority of cases where a file system has filled up the root cause has been the use of a SAP or SQL Server as a File Server. SAP and SQL Servers must never be used as File Servers. Chapter 4 discusses security in detail. It is a mandatory requirement of this deployment architecture that File Sharing is blocked by the Windows Firewall, preventing any of the SAP on Microsoft Private Cloud infrastructure from being used as a File Server. SAP ABAP programs that download content to a text file should be configured to write to ABAP Logical File Names (defined in transaction FILE) on a separate dedicated File Server. SAP and SQL Servers are not to be used as File Servers under any circumstance. SAP on Microsoft Private Cloud Page 11 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs VM Host Name Home Node (1 – 5) SAP Path on VM Virtual Hard Disk Name Cluster Shared Volume VHDX Path on Host SAN LUN CSV Cache sapapp1 1 C:\usr\SAP\ECC\D00 C:\usr\SAP\ECC\D01 C:\Program Files\SAP\HostCtrl\exe C:\Windows C:\Pagefile.sys C:\usr\SAP\SLM\D00 ??Host Agent Path?? C:\Windows C:\Pagefile.sys sapapp1_ECC_C_DISK.VHDX sapapp1_ECC_C_DISK.VHDX sapapp1_ECC_C_DISK.VHDX C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ 6 6 6 Yes Yes Yes sapapp1_ECC_C_DISK.VHDX sapapp1_ECC_C_DISK.VHDX sapapp2_SLM_C_DISK.VHDX sapapp2_SLM_C_DISK.VHDX sapapp2_SLM_C_DISK.VHDX sapapp2_SLM_C_DISK.VHDX C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\SLM\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\SLM\<vm name>\Virtual Hard Disks\ 6 6 6 6 6 6 Yes Yes Yes Yes Yes Yes sapapp2 Table shows two SAP systems running on separate VMs on a 5 node Microsoft Private Cloud Infrastructure. ECC 6.6 system SID = ECC and a Solution Manager system SID = SLM. ECC 6.6 has two application server instances running on one VM3. The Solution Manager system has only one instance. The Virtual Hard Disks for the SAP application servers can all reside on a dedicated Cluster Shared Volume (using its own dedicated SAN LUN). 3 For HA reasons it is always recommended to have multiple VMs for each production application component SAP on Microsoft Private Cloud Page 12 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 3.1.4 Cluster Shared Volume & Virtual Hard Disk Windows Server 2012 Cluster Shared Volumes can support many VHDXs and many VHDXs from different VMs. 10-25 Virtual Machines can share the same set of CSV without performance impact when configured correctly. If more than one SAP Component (eg. ECC, CRM, PI and Solman) is installed on a single SAP application server then the VHDX file holding the C: drive may need to be increased in size. The VHDX file should have the naming convention: C:\ClusterStorage\Volume6\<SID>\<vm hostname>\Virtual Hard Disks\<vm hostname>_<SID>_C_DISK.VHDX The value for SID on SAP application servers with multiple components should be the SID of the first system installed. 3.1.5 SAN Configuration A single Cluster Shared Volume on a dedicated single separate LUN is required. The LUN for the CSV must be presented to all nodes in the Hyper-V 3.0 cluster. RAID 5 (or similar) is suitable for the Cluster Shared Volumes holding the Virtual Hard Disks that in turn hold SAP Application Servers. The single CSV running on one LUN can support 10-25 Virtual Machines (likely many more). SAP on Microsoft Private Cloud Page 13 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs MS Private Cloud Physical Layer Windows 2012 Hyper-V 3.0 Cluster VM Layer Small VM = 4vCPU, 32GB vRAM In this example this VM is only running 2 SAP ABAP instances – 00 and 01 for an Large sized ECC 6.6 system sapapp1_ECC_C_DISK.VHDX PHYS_MEMSIZE = 16000 All profile parameters kept in DEFAULT.PFL VHDX: 1 x VHDX for a single large C: Drive CSV: 1 x CSV shared by all SAP systems LUN: 1 with CSV Cache NTFS File System Layer LOCAL C:\ -> 120GB GPT disk C:\usr\SAP\ECC\D00 -> ECC instance 00 C:\usr\SAP\ECC\D01 -> ECC instance 01 C:\usr\SAP\<SID>\<Instance Number> -> additional instances C:\pagefile.sys -> pagefile set to the same size as vRAM = 32GB C:\Windows -> 32GB minimum disk space Virtual Hard Disk -> CSV Layer C:\ClusterStorage\ Volume6 { SAN Layer LUN 6 SAP on Microsoft Private Cloud Page 14 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 4 Example Configuration – SAP ASCS SPOF VM SAP on Microsoft Private Cloud reference deployment can be decomposed into several simple layers thereby allowing customers and partners to correctly configure Microsoft Private Cloud infrastructures. The Single Points of Failure (SPOF) on a SAP system run on the ASCS. SAP ASCS Server Virtual Machines require some specific configuration to deliver a very Highly Available solution. The following sections should be reviewed in conjuction with the diagram on the following page. This section depicts four ASCS running on a two node cluster in Active/Active configuration. 4.1.1 Physical Server See the Sizing Chapter for information on the physical server configuration. The ASCS cluster nodes must all be “Small” configuration VMs. It is not permitted to run any other software or SAP application servers in conjunction with the ASCS. The reason for this restriction is to expedite troubleshooting of any cluster or failover issues. Note: all nodes of a ASCS cluster must be “Small” size. See the Sizing Chapter for details. It is not permitted to install any other software such as SQL Server or additional SAP application servers onto an ASCS cluster on this reference architecture. The reason for this is to simplify configuration and troubleshooting 4.1.2 Virtual Machine Windows Server 2012 Standard Edition or greater should be used for running an SAP ASCS Cluster. Windows Server 2012 Standard Edition and Datacenter Edition support the same number of cluster nodes. See the Sizing Chapter for information on the Virtual Machine sizing. 4.1.2.1 Roles & Features Required for SAP Application Server Windows Clustering Feature is required to install the ASCS. The SAP ASCS Cluster Server requires the File Services Role in order to share \SAPLOC & \SAPMNT. No other Roles or Features are typically required 4.1.2.2 Virtual Machine Affinity SAP ASCS cluster nodes (the VMs that form the ASCS 2 node cluster) must not run on the same physical host. This is documented in SAP Note 1374671. See the High Availability Chapter for more information 4.1.2.3 Windows Server 2012 Server Core It is not supported at this time to run the SAP application server on Windows Server Core or Hybrid. The only supported “Shell” is Full GUI. SAP may release the ASCS server on Windows Core in the future. SAP on Microsoft Private Cloud Page 15 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs See the Security Chapter for more information on hardening the Guest Virtual Machines 4.1.3 SAP ASCS Cluster Disk Configuration The SAP ASCS Cluster does not cause significant IO. Typically the IO activity on the \usr\sap file system is limited to writing trace files and other activity. The vast majority of the IO on a SAP system is caused by the Database. In order to simplify the configuration one large C: Drive should be used to hold the following components: 1. Windows Server 2012 Guest Operating System 2. Pagefile.sys 3. Enqueue Replication -> Instance Number 11, 12, 13, 14 4. SAP Host Agent Typically 120GB Virtual Hard Disk partitioned into one single large C: Drive is sufficient for this purpose. It is highly unlikely that the disk would ever fill up on a SAP application server as no files are stored on SAP Application Servers. The cluster components must reside on a cluster shared disk – Drive letters P:, Q:, R: and S: are reserved for holding the following components: 1. P:\usr\sap\<SID>\ASCS -> ASCS System Number 00 2. Q:\usr\sap\<SID>\ASCS -> ASCS System Number 01 3. R:\usr\sap\<SID>\ASCS -> ASCS System Number 02 4. S:\usr\sap\<SID>\ASCS -> ASCS System Number 03 Typically 30GB 4 Virtual Hard Disk partitioned into one single large P: Drive is sufficient for this purpose. Some SAP components may require more. Always check the SAP Installation Guide 4 SAP Installation Guides mention about 5GB as the minimum requirement. SAP on Microsoft Private Cloud Page 16 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs VM Host Name sapascs1 sapascs2 Home Node (1 – 5) 1 2 SAP Path on VM C:\Windows C:\Pagefile.sys C:\Program Files\SAP\HostCtrl\exe C:\usr\SAP\ECC\ERS P:\usr\SAP\ECC\ASCS C:\usr\SAP\BWP\ERS Q:\usr\SAP\BWP\ASCS C:\usr\SAP\XIP\ERS C:\usr\SAP\SLM\ERS C:\Windows C:\Pagefile.sys C:\Program Files\SAP\HostCtrl\exe C:\usr\SAP\XIP\ERS R:\usr\SAP\XIP\ASCS C:\usr\SAP\SLM\ERS S:\usr\SAP\SLM\ASCS C:\usr\SAP\ECC\ERS C:\usr\SAP\BWP\ERS Virtual Hard Disk Name host or SHARED sapascs1_ECC_C_DISK.VHDX sapascs1_ECC_C_DISK.VHDX sapascs1_ECC_C_DISK.VHDX Cluster Shared Volume VHDX Path on Host C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ SAN LUN 6 6 6 CSV Cache Yes Yes Yes sapascs1_ECC_C_DISK.VHDX SHARED vHBA or iSCSI disk sapascs1_ECC_C_DISK.VHDX SHARED vHBA or iSCSI disk sapascs1_ECC_C_DISK.VHDX sapascs1_ECC_C_DISK.VHDX C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ directly presented to VMs (Node A and Node B) C:\ClusterStorage\Volume6\BWP\<vm name>\Virtual Hard Disks\ directly presented to VMs (Node A and Node B) C:\ClusterStorage\Volume6\XIP\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\SLM\<vm name>\Virtual Hard Disks\ 6 7 6 8 6 6 Yes No Yes No Yes Yes sapascs2_ECC_C_DISK.VHDX sapascs2_ECC_C_DISK.VHDX sapascs2_ECC_C_DISK.VHDX C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ 6 6 6 Yes Yes Yes sapascs2_ECC_C_DISK.VHDX SHARED vHBA or iSCSI disk sapascs2_ECC_C_DISK.VHDX SHARED vHBA or iSCSI disk sapascs2_ECC_C_DISK.VHDX sapascs2_ECC_C_DISK.VHDX C:\ClusterStorage\Volume6\XIP\<vm name>\Virtual Hard Disks\ directly presented to VMs (Node A and Node B) C:\ClusterStorage\Volume6\SLM\<vm name>\Virtual Hard Disks\ directly presented to VMs (Node A and Node B) C:\ClusterStorage\Volume6\ECC\<vm name>\Virtual Hard Disks\ C:\ClusterStorage\Volume6\BWP\<vm name>\Virtual Hard Disks\ 6 9 6 10 6 6 Yes No Yes No Yes Yes Table shows a Multi-SID SAP ACS Cluster running on two VMs (in a guest cluster) on a 5 node Microsoft Private Cloud Infrastructure. ECC 6.5 system SID = ECC, a BW system SID = BWP, a XI system SID = XIP and a Solution Manager system SID = SLM. There is a standard Windows Cluster configured between sapascs1 and sapascs2 with Node and File Share Witness Majority. Each SAP ASCS instance is clustered. The ASCS for ECC and BWP run on sapascs1 typically. The ASCS for XIP and SLM run on sapascs2 typically. Enqueue Replication is configured to preserve the lock table if there is a failover. Each SAP system must have its own Shared Disk – P:, Q: R: and S: are Cluster Shared disks for the cluster between the two VMs. The Virtual Hard Disks for the SAP application servers boot disk (<host>_<SID>_C_DISK.VHDX) can all reside on the same dedicated Cluster Shared Volume (using its own dedicated SAN LUN). SAP on Microsoft Private Cloud Page 17 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 4.1.4 Cluster Shared Volume & Virtual Hard Disk Windows Server 2012 Cluster Shared Volumes can support many VHDXs and many VHDXs from different VMs. 10-25 Virtual Machines can share the same set of CSV without performance impact when configured correctly. 4.1.5 SAN Configuration A single Cluster Shared Volume on a dedicated single separate LUN is required. The LUN for the CSV must be presented to all nodes in the Hyper-V 3.0 cluster. RAID 5 (or similar) is suitable for the Cluster Shared Volumes holding the Virtual Hard Disks that in turn hold SAP Application Servers. The single CSV running on one LUN can support 10-25 Virtual Machines (likely many more). 4.1.6 Shared Disk for SAP ASCS The following two sections provide information about creating a shared disk for the SAP (A)SCS cluster between two VMs. The VM is to a large extent “unaware” of the fact that it is running on a Hypervisor and sees the VHDX on the Cluster Shared Volume as just a direct attached SCSI disk. Shared is is required for SAP and SQL Server cluster5 scenario. These shared cluster disks must be presented directly to the VMs and not to the physical Hyper-V hosts themselves. Shared disks should not be used for Quorum detection. The only supported Quorum mode for Microsoft Private Cloud for SAP systems is “Node and File Share Witness” with Dynamic Quorum. There are two technologies available for presenting cluster shared disks to a VM – iSCSI and vHBA (Virtual Fibre Channel) 4.1.6.1 iSCSI Many storage systems support iSCSI and can present a LUN to a Windows Server 2012 VM via the iSCSI initiator. It is important that the shared disk presented to both VMs that form the ASCS cluster 4.1.6.2 Virtual Fibre Channel (vHBA) Windows Server 2012 Hyper-V has a new feature to allow Fibre Channel devices to be presented to a “Virtual” HBA within a VM and also to preserve connectivity during a LiveMigration. The steps to setup a Virtual Fibre Channel Shared Cluster Disk are as follows: 1. Confirm that the SAN storage and physical HBA model in the Hyper-V hosts supports NPIV 2. Open Hyper-V manager on each of the physical hosts and click on Virtual SAN Manager… 3. Press Create button, name the Virtual Fibre Channel SAN and Tick at least 2 active ports (WWNN) SQL Server AlwayOn is generally recommended. SQL Server clustering is for reverse compatibility only. SQL Server AlwaysOn does not require any shared disks and is preferred for this reference implementation 5 SAP on Microsoft Private Cloud Page 18 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 4. Repeat this step on all Hyper-V hosts and ensure the naming and configuration is identical to avoid problems during LiveMigrations 5. Shutdown the VMs that are used for the ASCS cluster (if the VMs are active shutdown one at a time) 6. Open Failover Cluster Manager6 and select Add Hardware -> Fibre Channel Adapter -> Add 7. Select the Virtual SAN created in step 3 8. Take a screenshot and/or copy the WWNN and WWPN and pass this information to the SAN administrator 9. Request SAN administrator to present LUNs to the VM hosts (both Node A and Node B) 10. Start VM and open Device Manager –> Storage Controllers and check Hyper-V Fibre Channel HBA is added and working correctly 11. Open Computer Management -> Disk Management: check the disk is present and format the disk NTFS if required 12. Perform functionality testing: a. Run Failover Cluster Manager on the Hyper-V host and perform a LiveMigration. Expected outcome is no interruption of service and no error messages b. Run Failover Cluster Manager on one of the VMs and perform a Failover. Expected outcome is no interruption of service and no error messages Highly available Virtual Machines must be managed with Failover Cluster Manager and not Hyper-V Manager 6 SAP on Microsoft Private Cloud Page 19 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs MS Private Cloud Physical Layer Windows 2012 Hyper-V 3.0 Cluster Cluster Node 1 = (A)SCS + ERS ACTIVE Cluster Node 2 = ERS Small VM = 4vCPU, 32GB vRAM In this example there is a 2 node Active/Passive (A)SCS cluster for an Large sized ECC 6.6 system. Node 1 is running ASCS + ERS. Node 2 is running ERS. There is a single shared disk P: Drive. Each 2 node cluster can run many (A)SCS for ECC, BW, EP, PI etc (Multi-SID cluster) VM Layer ERS vHostname for ASCS vIP for ASCS ASCS installed on cluster disk ERS installed on local C: disk VHDX: 2 x VHDX for a local C: Drive and cluster shared P: Drive for ASCS CSV: 1 x CSV shared by all SAP systems LUN: Large – hold all ASCS and all boot disks CSV for boot disks and ASCS has CSV caching enabled vHBA or iSCSI Shared Disk sapascs1_ECC_C_DISK.VHDX sapascs2_ECC_C_DISK.VHDX NTFS File System Layer LOCAL C:\ -> 120GB disk C:\usr\SAP\ECC\ERS Instance 11 C:\pagefile.sys -> C:\Windows -> 32GB GPT – ERS = 32GB space SHARED DISK P:\ -> 30GB GPT disk P:\usr\SAP\ECC\ASCS – ECC Instance 00 \\vhostname\SAPMNT LOCAL C:\ -> 120GB disk C:\usr\SAP\ECC\ERS Instance 11 C:\pagefile.sys -> C:\Windows -> 32GB GPT – ERS = 32GB space sapascs1_ECC_C_DISK.VHDX Virtual Hard Disk -> CSV Layer vHBA or iSCSI Disk Presented Directly to VMs sapascs2_ECC_P_DISK.VHDX C:\ClusterStorage\ Volume6 { LUN 6 SAP on Microsoft Private Cloud LUN x Page 20 of 25 SAN Layer Created: 15-Mar-16 CSV, VHDX & LUNs 5 Cluster Shared Volumes Windows 2008 R2 introduced a Cluster Shared Volume restricted to support Hyper-V 2.0 LiveMigration only. Windows Server 2012 Cluster Shared Volumes are a full Cluster Shared File System that can be used for any application including Hyper-V 3.0. Cluster Shared Volumes are disks that can be read and written to by all nodes in a Windows cluster. The disk appears as a special directory under the C: drive (or whichever disk is booted from). Each cluster shared volume is named “C:\ClusterStorage\Volume” followed by the disk number. The procedure for converting a disk from a standard cluster shared disk to a CSV is documented online7. SQL Server cannot directly run on Cluster Shared Volumes, but SQL Server can run inside a VM which is running on Cluster Shared Volumes. 5.1 How Many Virtual Machines per CSV? Unlike VMware VMFS there is no technical reason to limit the number of VMs per CSV volume. Due to enhancements in the way metadata updates and locking are handled in Windows Server 2012 Cluster Shared Volumes adding many VMs to a single CSV will not cause performance lantencies. On very large Hyper-V 3.0 clusters (more than 20 nodes) it may be possible that the physical servers collectively generate so much IO that the single LUN that comprises a single CSV is eventually overloaded inside the SAN. Older generation SAN may have issues with this more so than modern SAN. It is recommended to separate Sandbox, Development, QAS and Production systems onto separate CSVs. Therefore most SAP on Microsoft Private Cloud deployments should have at least 3 sets of CSVs if deployed on the same Hyper-V cluster (Note: Large configurations would typically have a Hyper-V cluster for non-Prod and a separate Hyper-V cluster for Production) The following configuration should be followed: 1. C:\ClusterStorage\Volume1 to 9 is reservered for Development 2. C:\ClusterStorage\Volume10 to 19 is reservered for QAS 3. C:\ClusterStorage\Volume20 to 30 is reservered for Production 5.2 Renaming CSV Paths It is technically possible to rename CSVs from the default “C:\ClusterStorage\Volume1” to ““C:\ClusterStorage\VolumeXX”. Right click on any node and say “Rename..”. Windows Server 2012 will automatically rename the directory on all cluster nodes. 7 http://technet.microsoft.com/en-us/library/jj612868.aspx SAP on Microsoft Private Cloud Page 21 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 5.3 Cluster Shared Volume Cache Cluster shared volume caching can dramatically improve VM boot times. This feature should only be used on the CSV holding the boot disk and SAP ASCS disk (LUN 6) and not any other CSV. Do not enable CSV cache on any CSV holding SQL Server datafiles. 5.3.1 CsvEnableBlockCache 0 = Disabled (default) 1 = Enabled for that LUN Requires recycling the resource to take effect This is a private property of the cluster Physical Disk resource. It allows you to enable CSV Cache on an individual disk. This gives you the flexibility to configure cache for read intensive VMs running on some disks, while allowing you to disable and prevent random I/O on other disks from purging the cache. For example parent VHDX’s with high reads you would enable caching on Disk1, and high writes for differencing disks the CSV cache could be disabled on Disk2. The default setting is 0 for disabled, setting to a value of 1 enables CSV Block Cache on that disk. 5.3.2 SharedVolumeBlockCacheSizeInMB 0 = Disabled Non-zero = the amount of RAM in MB to be used for cache on all nodes in the cluster 1024 recommended value No downtime required to modify 5.3.3 How to Set CSV Cache To enable CSV cache follow the procedure below 1. Open an elevated Windows PowerShell prompt 2. Define the size of the size of the cache to be reserved (Get-Cluster). SharedVolumeBlockCacheSizeInMB = 1024 3. Enable CSV Cache on an individual disk Get-ClusterSharedVolume “<disk name>” | SetClusterParameter CsvEnableBlockCache 1 SAP on Microsoft Private Cloud Page 22 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 6 Virtual Hard Disk Configuration - VHDX Windows Server 2012 has a new Virtual Hard Disk format called VHDX. The new Virtual Hard Disk format has a much larger storage capacity8 than Windows 2008 R2 VHDX files. A Virtual Hard Disk file is a “Disk in a File”. The disk inside the Virtual Hard Disk file can be formatted, partitioned and handled just as any other physical disk. 6.1 VHDX Types There are several different Virtual Hard Disk types for Windows Server 2012. 6.1.1 Fixed, Dynamic & Differencing Only Fixed disks are supported for SAP on Microsoft Private Cloud reference architecture. Do not use pass through disks. 6.1.2 Windows 2008 VHD vs. Windows Server 2012 VHDX Windows Server 2012 VHDX disks should be used in all cases. Customers planning Hybrid Windows Azure and on premise Private Cloud deployments (which is not covered in this documentation) should note that Windows Azure is not compatible with VHDX. Windows Azure is currently not supported by SAP for Production systems 6.2 MBR, GPT, NTFS AUS Size and other options Disks should be formatted for GPT – Guid Partition Table. GPT is more resilant than MBR9. MBR is limited in size to 2TB. All Virtual Hard disks (the VHDX file system) that hold SQL Server datafiles or log files must be formatted with 64k NTFS Allocation Unit size. The NTFS allocation unit size on the Cluster Shared Volumes themselves should be formatted with the default NTFS allocation unit size (4k). To verify the NTFS allocation unit size run the command CHKDSK /i/c Windows has scanned the file system and found no problems. No further action is required. 234067967 107766292 14968 0 171883 65536 126114824 KB KB KB KB KB KB KB total disk space. in 44845 files. in 10960 indexes. in bad sectors. in use by the system. occupied by the log file. available on disk. 4096 bytes in each allocation unit. 8 http://technet.microsoft.com/en-us/library/hh831446.aspx 9 http://msdn.microsoft.com/en-us/windows/hardware/gg463525.aspx SAP on Microsoft Private Cloud Page 23 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 58516991 total allocation units on disk. 31528706 allocation units available on disk. C:\> SAP on Microsoft Private Cloud Page 24 of 25 Created: 15-Mar-16 CSV, VHDX & LUNs 7 Appendix 2 – Links Page Required OSS Notes: 1732161 - SAP Systems on Windows Server 2012 1811298 - SAP NetWeaver Upgrade/Update on Windows Server 2012 1753578 - Windows Server 2012 in Virtual Environments Blog Articles: CSV Cache http://blogs.msdn.com/b/clustering/archive/2012/03/22/10286676.aspx Channel9: Cluster Shared Volumes Reborn in Windows Server 2012: Deep Dive http://channel9.msdn.com/Events/TechEd/Europe/2012/WSV430 Failover Clustering and Hyper-V: Multi-Site Disaster Recovery http://channel9.msdn.com/Events/TechEd/NorthAmerica/2011/VIR303 Whitepapers: Technet MSDN: Use Cluster Shared Volumes in a Windows Server 2012 Failover Cluster http://technet.microsoft.com/en-us/library/jj612868.aspx CSV Cache http://technet.microsoft.com/en-us/library/jj612868.aspx Hyper-V Scalability in Windows Server 2012 http://technet.microsoft.com/en-us/library/jj680093.aspx Hyper-V: How to get the most from your virtualized disk performance (IDE vs. SCSI vDisks) http://blogs.msdn.com/b/tvoellm/archive/2007/12/12/which-is-better-ide-or-scsi-windows-servervirtualization-08-code-name-viridian-controller-performance.aspx SAP on Microsoft Private Cloud Page 25 of 25 Created: 15-Mar-16