How we did it
 
|
|||
 | |||
|
|
|
|||
| |||
|
Also: |
We then tested for specific performance criteria in the DWDM products. These included:
Transparency - This is the capability to transparently transport various protocols and bit rates - specifically OC-3 ATM, Gigabit Ethernet and Fast Ethernet. We checked each vendor's shelves to assure that it was transparent to the subtending edge equipment and the various protocols. If it was, the edge devices could see each other and were oblivious to the DWDM gear in the middle of the link.
Channel independence - This is the ability to protect the various lambdas from interfering with each other. We tested this by simultaneously running Fast Ethernet, Gigabit Ethernet and ATM streams through the DWDM equipment and looking for channel interaction. We created glitches in each stream, one at a time, and noted if these glitches also appeared on any other the other streams.
Fault tolerance and module failures - We simulated module failures by shutting off redundant power supplies one at a time and by pulling chassis cards and noting the results. Going card-by-card, shelf-by-shelf, where possible we disabled the cards in shelf and noted the results. Once a card was tested, it was reinserted into the shelf and the disrupted service given time to stabilize. While the specific card was being removed, and reinserted, we looked to see if doing so created any side effects in transports not fed through the card. We noted if the shelves survived the simulated module failures and, if not, how long it took for them to recover.
Stability and reliability - To measure the stability of the shelves, we set up simultaneous and continuous pings on all three transports (ATM, Fast Ethernet, and Gigabit Ethernet). We used three different workstations to ping devices on the far end of the connection, assuring that all pings had to go through the shelves. We let the pings run continuously for 72 hours, while keeping our fingers crossed that our UPS held in the event of a power failure. At the end of the test we counted the number of lost pings. Lost pings, of course, can be due to a variety of issues - not just failures in the shelves. However, no significant loss of pings was a very good sign.
Management interface - We determined how each vendor's DWDM gear was managed. When equipment is really in trouble or the network is down, it is important to directly connect to the shelf to determine the problem and correct it. Therefore, we looked to see if each shelf had serial port connectivity. We checked to see if a Web interface was supported, and if the shelf also had telnet access. We also looked for SNMP capability, although we did not test this feature.
RELATED LINKS
Fritz, who is the principal network engineer for West Virginia University's Network Services department, is responsible for advanced network technology development for the WVU campus networks and has directed the university's Advanced Network Applications Lab since 1988. Fritz holds a master's degree in Electrical Engineering from West Virginia University and is the author of Remote LAN Access: a guide for networkers and the rest of us and Sensible ISDN Data Applications.
The West Virginia University Network Services test team consisting of Matthew Glotfelty, Jeff Davis and Bird Vilseck.
Enterprise optical boxes see the light
Nortel's OPTera Metro 5200 tops the list of DWDM gear tested.
Interactive scorecard and NetResults
Some established vendors bow out
