![]() |
|
||||||||
|
|
Overview | Designs | Projects | Rates | Log-in | About | ISP-HowTo | Home | |
|
|
Design FAQ Why use Linux? Unix (and variants, including AIX, HP-UX, Solaris, and Linux) is the operating system running around 80 percent of all Internet servers in the world. Unix was developed in the 1960s. It is a mature and stable product with outstanding reliability. For further details, read the Why Linux? of our How-to section. Why use SCSI hard disks? SCSI is an alternative to IDE, providing higher levels of performance and reliability. SCSI drives operate at 7200 to 15,000 (and even 20,000 RPM) spindle speeds and seek times almost half those of IDE drives. As of this writing, Ultra-320 SCSI is readily available, offering raw potential three times that of the theoretical IDE limit. SCSI costs more than IDE, but the benefits are often worth it. What is RAID and why would I want it? RAID stands for Redundant Array of Inexpensive Disks. RAID offers constant data backup and/or an increase in server throughput by spreading data across multiple hard disks that are operating in unison. There are many types and levels of RAID. The types, from slowest to fastest, include IDE RAID, IDE-SCSI RAID, and SCSI RAID. RAID can be implemented in hardware or software (hardware is best). RAID levels are numbers (0, 1, 5, etc.) that indicate the data spreading method being used. The earliest form of RAID was developed solely to provide data redundancy, so that if one drive fails the data is not lost and the server continues operating. Later, RAID was expanded to provide data 'striping' for increased speed, and then the two methods were combined. In the 2003 model year, all packages from the Community upward include hardware RAID. In all cases we use RAID for redundancy; on larger servers we also use RAID for speed. What are 'cold spares' and why would I want them? A redudant power supply module mounted in a server is a hot spare; an extra module sitting on the shelf is a cold spare. Hot spares are ideal because there is zero down-time when one unit fails. Not all parts lend themselves to this, however. For example, a server cannot have two video cards. So having a combination of hot and cold spares provides the best guarantee of maintaining carrier-class 99.9% up-time. Every ISP Ltd. system comes with a well planned set of spare parts. Spare hard disk drives are pre-configured from the factory to minimize swap time. A complete server is provided as a cold spare in every system (even the Neighbourhood) which reduces the repair time for any server failure to mere minutes. Why use a proxy server? Proxy servers are most commonly used for networks behind a firewall, to provide increased levels of security for the ISP and its users. A caching proxy server goes one step further by storing and delivering copies of popular Internet web content. The cache can be used to buffer a busy web server inside the ISP's network or, the opposite, to buffer web servers outside the ISP's network. Buffering outside web sites can dramatically increase the speed of subscriber access, especially when the demand for content exceeds the backbone connection speed. An intelligent caching proxy server enhances this function with a great degree of flexibility to match the needs of individual ISPs. The proxy server used in ISP Ltd's systems provide the highest level of intelligence / efficiency available. A transparent intelligent caching proxy server provides all of these services transparently to the subscribers. This has the benefit of being instantly available to all subscribers; no proxy settings are required to subscriber computers. |
![]() |