How to optimize Windows Server performance

Optimizing Windows Server Disk Performance

In the use of a single server, disk often occupies a very important position. Server disk configuration and maintenance have a significant impact on the performance of the server as a whole. In fact, the way to optimize the server disk takes only a little time. In this article, I'll share some ways to improve the efficiency of your server's disk subsystem.

Note: Almost all servers are configured differently. That is, the disk configuration and optimization techniques that work best for your organization may be detrimental to servers in another organization. The only way to fully optimize a single server disk is to analyze what applications are running on this server, and how much effort these applications are taking on the server's disk subsystem.

So here is no way to provide the most comprehensive and perfect way to get your server running with optimized performance. This article will bring some common methods that can be used to improve the performance of most types of servers.

File layout

One trick to optimizing server performance is to put files on the appropriate disk. For example, the Windows server operating system is typically installed on the C disk. In this case, I recommend that C disk be used only for operating system files, although in some cases you can apply some other to C disk without having a bad effect on performance.

Because the operating system component consumes a lot of disk input/output time and generates pagefile system files, I recommend placing it on a dedicated disk drive (not an exclusive partition). This prevents disk input/output from associating virtual memory paging without interrupting other parts of the system.

Any data that is stored on the server should normally be stored on a dedicated volume. Because the vast majority of server disk input/output is usually associated with read and write data, it typically requires data to be stored on a RAID (redundant array of independent disks) array, depending on the performance level and fault-tolerant control you need. RAID (redundant array of independent disks) 10 arrays are often a better choice because they are as fast as RAID 1 arrays, but they provide fault-tolerant control through mirroring (mirroring).

Memory

Whether you believe it or not, the total amount of memory in your system can make a huge difference to the performance of your disk subsystem. I've talked about storing paging files on an exclusive disk to minimize the impact of the virtual memory pages that are brought to the system. The concept of virtual memory is relative to physical memory, when the system's physical memory space to make ends meet, the operating system will open up a disk space on the hard disk as memory.

However, this technique is quite passable. First, the disk's speed is even slower than the slowest physical memory. The disk's access speed is at the order of milliseconds, while physical memory is accessed at a One-zero-second order of magnitude.

Another problem is that the operating system cannot take advantage of the paging of memory stored on virtual memory. If the operating system needs to perform an instruction in one memory paging, the first paging must be moved to physical memory. This step is called commutative (swapping) or paging (a mapping of the linear space to the physical address space).

If the physical memory of a single server is loaded, and when it needs to access a page of memory stored in virtual memory, at this point the server will have to swap a temporarily unused paging from physical memory to virtual memory to make space on the physical memory, The ability to swap the paging of memory needed for access to physical memory from virtual memory.

This paging process consumes not only disk resources, but also a large amount of CPU time and memory. Some of the server's physical memory tends to record which paging is on random memory (RAM) and which pages are on virtual memory. That's why the paging method is so inefficient. Some paging is normal, but excessive paging can severely restrict the performance of a system. Excessive paging can result in an unchanging disk activity.

The best way to reduce paging is to install as much memory on your server as possible. As the number of random memory (RAM) in the server increases, the size of the pagefile file increases accordingly. However, the system dependencies on the pagefile file will be reduced. This will free up a considerable amount of disk processing time and CPU processing time.

Disk Defragmenter

One of the most important ways to do this is to ensure that you have a good performance disk and you should go to regular Disk Defragmenter routine. The disk is designed to write data in a linear fashion. However, when a file is deleted, the disk file is not stored continuously, and where the file is deleted, the space is not available.

The disk calculates the available space before writing the data to free space. If the free space does not provide enough space for the data being written, files are dispersed to different parts of the disk: a portion of the file is inserted into a gap in the available space, and the remainder may be stored in a slot in the next available space, or at the end of the disk.

This kind of file fragmentation has a bad effect on system performance, because too much file fragmentation can cause systems to look back and forth while reading and writing files, making them less efficient. Defragment the disk and allow the file to be stored in a linear fashion, which will speed up disk access and read and write.

Index

Depending on your server's past experience, the use of indexes should be a mixed blessing. The index includes reading all the files on the disk and creating an index that tells the operating system what files it contains. If a user uses this routine search to find special files in a large amount of data, the index makes this method more efficient.

However, many system administrators do not like to use indexes because the indexing process consumes a lot of CPU processing time and memory. Some disk space also tends to store indexes.

My advice: It is appropriate to use the Server-by-server basis to determine whether an index is used. In my organization I do index file servers because I often have to go through my file search to see if I have written articles about certain issues. However, my mail servers, database servers, primary domain Servers (DNS), domain controllers, and so on are not indexed. It makes no sense to index these types of servers. In fact, if you want to improve system performance, you should check your infrastructure servers to make sure they are not indexed.