Disks often occupy a very important position in the use of a server. The configuration and maintenance of server disks has a major impact on overall server performance. In fact, the method of optimizing the server disk takes only a little time. In this article, I will 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 are best for your organization can be detrimental to another organization's servers. The only way to fully optimize a server disk is to analyze which applications are running on that server and how much work these applications consume in the server disk subsystem. So, there is no way to provide the most comprehensive and complete method to get your server running with optimal performance. This article will bring some general methods that can often be used to improve the performance of most types of servers. File Layout One of the tricks to optimize server performance is to place files on the appropriate disk. For example, the Windows Server operating system is usually installed on the C drive. In this case, I recommend using the C drive only for storing operating system files, although in some cases you can apply some other to the C drive without adversely affecting performance. Because the operating system component consumes a lot of disk input/output time and generates a pagefile system file, I recommend placing it on a dedicated disk drive (not a dedicated partition). This prevents disk input/output from correlating virtual memory paging without interrupting other parts of the system. Any data 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 data and write data, it typically requires data to be stored on a RAID (Redundant Array of Independent Disks) array, depending on the level of performance you need and Fault tolerant control. RAID (Redundant Array of Independent Disks) 10 arrays are often a good choice because they are as fast as RAID 1 arrays, but provide fault-tolerant control through mirroring. Memory Whether you believe it or not, the total amount of memory in your system can make a huge difference in disk subsystem performance. I have already discussed storing page files on a dedicated disk to minimize the impact of the virtual memory pages that are left unused by the system. The concept of virtual memory is relative to physical memory. When the system's physical memory space is not enough, the operating system will open up a disk space on the hard disk as memory. However, this technique is very unsatisfactory. First, the corresponding speed of the disk is even slower than the slowest physical memory. Disk access speeds are on the order of milliseconds, while physical memory access speeds are on the order of a billionth of a second. Another problem is that the operating system cannot take advantage of memory paging stored on virtual memory. If the operating system needs to execute an instruction in a memory page, then the page must first be moved to physical memory. This step is called swapping or paging (linear space to physical address space mapping). If the physical memory of a server is full, and when it needs to access a memory page stored in virtual memory, then at this time, the server will have to go to a temporary unused page from physical memory. Switched to virtual memory so that there is space on physical memory to swap the memory pages that need to be accessed from virtual memory to physical memory. This paging process not only consumes disk resources, but also consumes a lot of CPU time and memory. There are some servers whose physical memory tends to record which pages are on random access memory (RAM) and which pages are on virtual memory. This is why the pagination method is so inefficient. Some paging is normal, but excessive paging can severely limit the performance of a system. Excessive paging can result in constant disk activity. The best way to reduce paging is to install as much memory as possible on your server. As the number of random access memories (RAMs) in the server increases, correspondingly, the size of the pagefile file also increases. However, the system dependencies on the pagefile 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 methods here is to ensure that you have a good performance disk, you should go to the regular execution of the disk defragmentation program. Disks are designed to write data in a linear fashion. However, when a file is deleted, the disk file will not be stored continuously, and where the file is deleted, an unusable space will be formed. The disk will calculate these free spaces before writing the data to free space. If the available space does not provide enough space for the data being written, the file is scattered to different places on the disk: part of the file is inserted into the gap of the available space, and the rest may be stored The gap in the next available space, or the end of the disk. This file fragmentation has a bad impact on system performance, because too much file fragmentation can cause the system to look back and forth while reading and writing files, making the efficiency low. Clean up disk fragmentation and allow files to be stored in a linear fashion, which will speed up disk access and read and write speeds. Index According to the past experience of your server, the use of the index in it should be mixed. Indexing involves reading all the files on the disk and creating an index telling the operating system what files are included. If the user searches for a particular file in a large amount of data through this routine search, the index will make 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 suggestion: It is appropriate to use the server-by-server basis to determine whether to use the index. I am doing an index file server in my organization because I often have to go through my file search to see if I have written an article about some issues. However, my mail server, database server, primary domain server (DNS), domain controller, etc. are not indexed. There is no point in indexing these types of servers. In fact, if you want to improve system performance, you should check your underlying server to make sure they are not indexed.