- Lựa chọn các đời main từ thấp đến cao

- Phân loại các main cùng đời vào một nhóm

- Quá 2 đời main hoặc 2 nền tảng thì làm 2 image

- Cài driver cho từng nhóm main

- Phân bố hardware profile theo card màn hình

A page file (also known as a "paging file") is an optional, hidden system file on a hard disk. The page file can be used to "back" (or support) system crash dumps and extend how much system-committed memory (also known as “virtual memory”) a system can back. It also enables the system to remove infrequently accessed modified pages from physical memory to let the system use physical memory more efficiently for more frequently accessed pages.

64-bit versions of Windows and Windows Server support more physical memory (RAM) than 32-bit versions support. However, the reason to configure the page file size has not changed. It has always been about supporting a system crash dump, if it is necessary, or extending the system commit limit, if it is necessary. For example, when a lot of physical memory is installed, a page file might not be required to back the system commit charge during peak usage. The available physical memory alone might be large enough to do this. However, a page file or a dedicated dump file might still be required to back a system crash dump.

Use the following considerations for page file sizing for all versions of Windows and Windows Server:

• Crash dump setting: If you want a crash dump file to be created during a system crash, a page file or a dedicated dump file must exist and be large enough to back the system crash dump setting. Otherwise, a system memory dump file is not created.
• Peak system commit charge: The system commit charge cannot exceed the system commit limit. This limit is the sum of physical memory (RAM) and all page files combined. If no page files exist, the system commit limit is slightly less than the physical memory installed. Peak system-committed memory usage can vary greatly between systems. Therefore, physical memory and page file sizing also varies.
• Quantity of infrequently accessed pages: The purpose of a page file is to back infrequently accessed modified pages so that they can be removed from physical memory. This provides more available space for more frequently accessed pages. The "\Memory\Modified Page List Bytes" performance counter measures, in part, the number of infrequently accessed modified pages that are destined for the hard disk. However, be aware that not all the memory on the modified page list is written out to disk. Typically, several hundred megabytes of memory remains resident on the modified list. Therefore, consider extending or adding a page file if all the following conditions are true:
  • More available physical memory (\Memory\Available MBytes) is required.
  • The modified page list contains a significant amount of memory.
  • The existing page files are fairly full (\Paging Files(*)\% Usage).
  
  Notes
  • Some products or services may require a page file for reasons other than those discussed here. For more information, check your product documentation. For example, Windows Server domain controllers and DFS replication, certificate, and LDS servers (also Client editions) are not supported without a configured page file. The algorithm of the database cache for ESENT (ESE, in Microsoft Exchange Server) depends on the "\Memory\Transition Pages RePurposed/sec" performance monitor counter. A page file is required to make sure that the database cache can release memory if memory is requested by other services or applications. In summary, page file sizing depends on the system crash dump setting requirements and the system commit charge peak usage or expected usage. Both considerations are unique to each system, even for systems that are identical to other systems. This means that page file sizing is unique to each system and cannot be generalized.
  • For Windows Server 2012 Hyper-V and Windows Server 2012 R2 Hyper-V, the page file of the management OS (commonly called the host OS) should be left at the default of setting of "System Managed." This is per the Hyper-V product group.

System committed memory

The system commit limit is the sum of physical memory and all page files combined. It represents the maximum system-committed memory (known as the “system commit charge”) that the system can back. The system commit charge is the total committed or “promised” memory of all committed virtual memory in the system. If the system commit charge reaches the system commit limit, the system and processes might not obtain committed memory. This condition can cause hangs, crashes, and other malfunctions. Therefore, make sure that you set the system commit limit large enough to back the system commit charge during peak usage. 

The system commit charge and system commit limit can be measured on the Performance tab in Task Manager or by using "\Memory\Committed Bytes" and "\Memory\Commit Limit" performance counters. The "\Memory\% Committed Bytes In Use" counter is a ratio of the "\Memory\Committed Bytes" to "\Memory\Commit Limit" values. 

Note System-managed page files automatically grow up to three times physical memory or 4 GB (whichever is larger) when the system commit charge reaches 90 percent of the system commit limit. This assumes that enough free disk space is available to accommodate the growth.

System crash dumps

A system crash (also known as a “bug check” or a "Stop error") occurs when the system cannot run correctly. The dump file that is produced from this event is called a system crash dump. A page file or dedicated dump file is used to write a crash dump file (memory.dmp) to disk. Therefore, a page file or a dedicated dump file must be large enough to back the kind of crash dump selected. Otherwise, the system cannot create the crash dump file.

Note During startup, system-managed page files and system-managed dedicated dump files are sized respective to the system crash dump settings. This assumes that enough free disk space exists.

* 1 MB of header data and device drivers can total 256 MB of secondary crash dump data.

Automatic memory dump

Windows 8 and Windows Server 2012 introduced the “Automatic memory dump” feature. This feature is enabled by default. This is a new setting, not a new kind of crash dump. This setting automatically selects the best system crash dump based on the frequency of system crashes. 

The Automatic memory dump setting at first selects a Small memory dump, which requires a page file or a dedicated dump file of at least 256 KB. If the system crashes, the Automatic memory dump feature selects a Kernel memory dump at startup. Then, it increases the minimum size of the system-managed page file or the system-managed dedicated dump file to back this kind of crash dump. 

Kernel memory crash dumps require enough page file space or dedicated dump file space to accommodate the kernel mode side of virtual memory usage. If the system crashes again within four weeks of the previous crash, a Complete memory dump is selected at restart. This requires a page file or dedicated dump file of at least the size of physical memory (RAM) plus 1 MB for header information plus 256 MB for potential driver data to support all the potential data that is dumped from memory. Again, the system-managed page file or the system-managed dedicated dump file will be increased to back this kind of crash dump. If the system is configured to have a page file or a dedicated dump file of a specific size, make sure that the size is sufficient to back the crash dump setting that is listed in the table earlier in this section together with and the peak system commit charge.

For more information about system crash dumps, click the following article number to go to the article in the Microsoft Knowledge Base:

969028 How to generate a kernel or a complete memory dump file in Windows Server 2008 and Windows Server 2008 R2

Dedicated dump files

Computers that are running Microsoft Windows or Microsoft Windows Server usually must have a page file to back a system crash dump. System administrators now have the option to create a dedicated dump file instead by using the following software packages to start with:

• Windows 7 Service Pack 1 with hotfix 2716542 applied
• Windows Server 2008 R2 Service Pack 1 with hotfix 2716542 applied

A dedicated dump file is a page file that is not used for paging. Instead, it is “dedicated” to back a system crash dump file (memory.dmp) when a system crash occurs. Dedicated dump files can be put on any disk volume that can support a page file. We recommend that you use a dedicated dump file when you want a system crash dump but you do not want a page file. 

For more information about dedicated dump files, click the following article numbers to go to the articles in the Microsoft Knowledge Base:

969028 How to generate a kernel or a complete memory dump file in Windows Server 2008 and Windows Server 2008 R2

950858 Dedicated dump files are unexpectedly truncated to 4 GB on a computer that is running Windows Server 2008 or Windows Vista and that has more than 4 GB of physical memory

System-managed page files

By default, page files are system-managed. This means that the page files increase and decrease based on many factors, such as the amount of physical memory installed, the process of accommodating the system commit charge, and the process of accommodating a system crash dump. 

For example, when the system commit charge is more than 90 percent of the system commit limit, the page file is increased to back it. This continues to occur until the page file reaches three times the size of physical memory or 4 GB, whichever is larger. This all assumes that the logical disk that is hosting the page file is large enough to accommodate the growth. 

The following table lists the minimum and maximum page file sizes of system-managed page files.

* See system crash dumps.

Performance counters

Several performance counters are related to page files. This section describes the counters and what they measure.

\Memory\Page/sec and other hard page fault counters

The following performance counters measure hard page faults (which include, but are not limited to, page file reads):

• \Memory\Page/sec
• \Memory\Page Reads/sec
• \Memory\Page Inputs/sec

The following performance counters measure page file writes:

• \Memory\Page Writes/sec
• \Memory\Page Output/sec

Hard page faults are faults that must be resolved by retrieving the data from disk. Such data can include portions of DLLs, .exe files, memory-mapped files, and page files. These faults might or might not be related to a page file or to a low-memory condition. Hard page faults are a standard function of the operating system. They occur when the following items are read:

• Parts of image files (.dll and .exe files) as they are used
• Memory-mapped files
• A page file

High values for these counters (excessive paging) indicate disk access of generally 4 KB per page fault on x86 and x64 versions of Windows and Windows Server. This disk access might or might not be related to page file activity but may contribute to poor disk performance that can cause system-wide delays if the related disks are overwhelmed. 

Therefore, we recommend that you monitor the disk performance of the logical disks that host a page file in correlation with these counters. Be aware that a system that has a sustained 100 hard page faults per second experiences 400 KB per second disk transfers. Most 7200 RPM disk drives can handle about 5 MB per second at an IO size of 16 KB or 800 KB per second at an IO size of 4 KB. No performance counter directly measures which logical disk the hard page faults are resolved for.

\Paging File(*)\% Usage

The \Paging File(*)\% Usage performance counter measures the percentage of usage of each page file. 100 percent usage of a page file does not indicate a performance problem as long as the system commit limit is not reached by the system commit charge, and if a significant amount of memory is not waiting to be written to a page file.

Note The size of the Modified Page List (\Memory\Modified Page List Bytes) is the total of modified data that is waiting to be written to disk. 

If the Modified Page List (a list of physical memory pages that are the least frequently accessed) contains a lot of memory, and if the % Usage value of all page files is greater than 90, you can make more physical memory available for more frequently access pages by increasing or adding a page file.

Note Not all the memory on the modified page list is written out to disk. Typically, several hundred megabytes of memory remains resident on the modified list.

Multiple page files and disk considerations

If a system is configured to have more than one page file, the page file that responds first is the one that is used. This means that page files that are on faster disks are used more frequently. Also, putting a page file on a “fast” or “slow” disk is important only if the page file is frequently accessed and if the disk that is hosting the respective page file is overwhelmed. Be aware that actual page file usage depends greatly on the amount of modified memory that the system is managing. This means that files that already exist on disk (such as .txt, .doc, .dll, and .exe) are not written to a page file. Only modified data that does not already exist on disk (for example, unsaved text in Notepad ) is memory that could potentially be backed by a page file. After the unsaved data is saved to disk as a file, it is backed by the disk and not by a page file.

Provisioning Services based servers

The next set of registry changes are recommended for images deployed using standard mode vDisk images with Citrix Provisioning services. Standard mode images are unique in that they are restored to the original state at each reboot, deleting any newly written or modified data. In this scenario, certain processes are no longer efficient. These configurations may also apply when deploying persistent images and in many cases should be implemented in addition to the changes recommended in the preceding section.

Note: With Provisioning Services 6.0 and above it is no longer necessary or recommended to disable the Large Send Offload and TCP/IP Offload features.

 

* The Optimizer column indicates whether this registry change is included in the XenConvert Optimizer tool that is installed with the Provisioning Services target device software.

Services

Windows Server 2008 R2 includes a set of services, many of which are enabled as default. When using the system as Citrix XenApp server, some of these services provide little value increases but significantly increase the memory footprint of the OS. Therefore the following services could potentially be disabled.

 

Service	Recommendation	Justification
Desktop   Windows  Manager Session  Manager	Potential	This service is responsible for Windows 7 Aero theme. Organizations need to determine if Aero theme is something they wish to support at a later date
Network List   Service  Network  Location  Awareness	Disable Disable	The network location of a XenApp Server is unlikely to change over time. Thus automatic network location detection is not required.  Please note that some 3rd party software products, such as the AppSense User Virtualization Platform, may require these services.  Note: Disabling these services may cause an initial delay of 15 seconds when visiting a HTTPs web site.
Themes	Potential	Allows users to manage the themes, which includes backgrounds, sounds and visual effects. Although this service does take resources and will impact overall scalability, each organization needs to determine if this functionality should be supported. It does allow the user to better personalize the environment and helps improve acceptance.
Windows Defender	Disable	Most enterprise deployments will have their own anti-malware solutions. As the corporate version will most likely be used, the integrated service should be disabled.
Windows Update	Potential	For XenApp servers based on a Provisioning Services vDisk or in case an Enterprise Software Deployment Systems is used to deploy Windows hotfixes, the Windows Update Service is not required.

SMB tuning parameters:

User Settings

 

Unlike machine settings, user settings must be applied to each user and typically cannot be applied as part of the base image. In addition, user settings typically do not depend on whether the image is deployed in private or standard mode; however, the method in which the settings should be applied can be influenced by the image mode. None of the configurations in this section are included in the XenConvert Optimizer tool. These factors make user settings more difficult to implement.

As a recommended approach for the application of the user settings is to utilize group policy settings and/or preferences, as they provide administrators a central site for configuration and allow for greater customization based on the desktop.