FDDI switching in network recovery
|
|
|||
|
|
Sign up to receive this and other networking newsletters in your inbox.
One of the key attractions of FDDI as a backbone technology is that devices "dual attached" in a ring can "wrap" around a break in a few milliseconds or less. In comparison, traditional bridging and routing protocols can take up to a minute or more to recover, potentially causing catastrophic session loss. Even proprietary solutions can take up to three seconds, preserving sessions but resulting in possible data loss, transport retransmissions and momentary interruption in network activity. For environments in which a network outage of even a few seconds is disastrous, FDDI is an ideal topology. This is perhaps one reason that, by some estimates, FDDI is still the most prevalent backbone technology.
So the question arises: What happens if users choose to replace the shared, dual-ring FDDI LAN with FDDI switching to improve performance? A switched network usually means a single connection between a switch port and an attached device. Does installing an FDDI switch necessarily mean removing redundant links and forfeiting FDDI's fault tolerance? Fortunately for FDDI customers, the answer is a resounding no.
In addition to the dual attachment described above, FDDI provides a second, less well-known method for link recovery, know as dual homing. A dual-homed device maintains an active connection to one switch (or concentrator) and a hot standby to another. If the primary connection fails, the standby assumes the active role so quickly that not only are sessions preserved, but the network recovers with near-zero frame loss.
This means that customers with FDDI switches can accrue both the performance benefits of dedicated 100M bit/sec connections and the high availability of FDDI's built-in network recovery through dual homing.
However, there are two catches to this solution, both involving cost. First, a dual-homed device requires two FDDI ports, known as an A and a B port; devices with such ports are generally more expensive than, for example, an S port. (The switch must also support M port configurations on both the active and the standby port, but most switches do.) Second, dual homing ties up two switch ports - one active, one standby - at potentially considerable expense. This becomes even more significant if the active and standby connections are on different switches for maximum resilience. Thus, while network uptime is maximized, the associated costs can be considerable.
Next week, we'll look at FDDI switching vs. Fast Ethernet switching, and consider if and when to replace FDDI with Fast Ethernet.
RELATED LINKS
Hub review and buyer's guide
We take a detailed look at several hubs, including their FDDI support, and suggest what to look for when considering hubs. Network World, 6/9/97.
FDDI diagrams, specifications and recovery techniques from Cisco
LAN Emulation: Q & As for the End User from the ATM Forum
FDDI Protocol overview from Bay Networks
Additional High Speed LANs newsletters
Sign up for a free subscription to this and other newsletters
