发表于:2007-09-26 12:30:00
楼主
As Industrial Ethernet increases in popularity, redundant communication schemes often become an issue in the factory, plant or in any of many other industrial settings. Often redundancy involves extra data storage, redundant DCE equipment (switches and routers), redundant DTE equipment (network interface modules, operator work stations, HMIs), uninterruptible power supplies and backup network paths. Note that the Internet is already highly redundant, so rarely is its packet-switching activity of concern in redundant considerations.
For backing up network paths, the STP and RSTP standards are available, as well as many proprietary schemes. But when considering a redundant network, the fundamental issue of the thoroughness of redundancy needs attention. Some network engineers aim for what is sometimes called "five nines", which means an uptime of 99.999% for a specified time. For example, this design criterion allows for only about five minutes of total downtime during a one-year period. But what factors in to such a calculation?
A network is never completely redundant; there is always a lack of redundancy at some level. Therefore, an essential decision to be made is what level of redundancy is necessary. The more thorough the redundancy, the more costly its implementation.
Often the only concern is to safeguard against cable failure. Achieving this level of redundant coverage is straightforward. It typically involves a relatively minor cost to provide the backup cabling and the devices needed to activate the backup cabling.
Sometimes it is decided to also supply redundant copies of the DCE devices -- typically managed switches -- that activate the backup cabling. Of course, this adds to the cost of the installation.
Occasionally, backup power is provided to maintain uninterrupted power to the DCE equipment -- both primary and backup. But to fully guard against power interruption, the power conductors themselves must be duplicated.
What is seldom provided is backup of the equipment being serviced by the network. And even when such backup exists, there will always be some threat to the facility itself -- earthquake, fire, etc.
In short, it is impossible to provide 100% redundancy for any system or activity. Th