Since RAID technology allows multiple disks to read and write simultaneously, this significantly increases the performance requirements for a server's CPU, memory, I/O bandwidth, etc., compared to when a single disk is used. Moreover, as the read/write performance of individual disks improves, the requirements on the server also increase accordingly. In the past, when using Ultra Wide SCSI with 40Mb/s disk arrays, the demand for CPU speed was not stringent because SCSI itself was not very fast. However, as SCSI evolved to Ultra2 at 80Mb/s, the demands on the CPU became more critical, especially on Intel Pentium processors, as well as on specialized processors in high-performance servers from companies like IBM, HP, Compaq, and Sun based on RISC architecture. Nowadays, the latest processors from Intel and AMD can support the fastest current RAID technology, Ultra3 at 320Mb/s.
In addition, server bus technology is another key factor limiting RAID application. Therefore, the I/O bus architecture supporting RAID technology has shifted from traditional I/O structures to 120 (Intelligent I/O, abbreviated as I2O) structures, aiming to reduce the load on the server's CPU.
In terms of server network operating systems, currently NetWare supports RAID1 (mirroring and duplexing). Windows NT, Windows 2000, Linux, and other systems support RAID0, RAID1, and RAID5 levels. To achieve data security redundancy, there must be certain requirements regarding the use of disks in the disk array. Ensuring data security depends on the design and quality of the disk array. There are several features that need to be considered: Is there an environmental monitor for temperature, voltage, power supply, cooling fans, and disk status? Are the disks in the disk array connected via SCA-2 integrated backplanes or just through SCSI cables? Are there isolation chips on the SCA-2 integrated backplane to prevent system voltage fluctuations caused by high/low voltages during hot swapping, which could destabilize the system and cause data loss? These aspects need attention. Since many disks in a disk array share the same SCSI bus, one disk being hot-swapped should not affect others. 80-pin SCSI disks support hot-swapping, while 68-pin disks do not. Whether hot-swapping is supported is reflected in the circuit design by whether protective circuits are included. Similarly, there are differences between genuine and fake hot-swappable disk trays.
When configuring a disk array, it’s important to note the order of disks within the array (not all of them). This means that some RAID controller cards require disks to be reinserted in the correct sequence after removal to ensure data can still be accessed normally. For example, if six disks are used in an array and were initially placed in sequential order (first, second... sixth), then if the array has an order requirement, after removing a disk for maintenance, it must be reinserted in its original position. Otherwise, due to mismatched disk order, the controller may fail to recognize the disks, leading to data loss.
Anhui Server Data Recovery Center: http://www.ahserver.com/plus/view-74-1.html