T.R | Title | User | Personal Name | Date | Lines |
---|
893.1 | | NPSS::MDLYONS | Michael D. Lyons DTN 226-6943 | Tue Dec 17 1996 09:40 | 11 |
893.2 | SAS and DAS again | NSDP01::RV | | Wed Feb 05 1997 05:55 | 30 |
| Hi,
Can someone give an advise on this configuration.
Two GIGAswitches interconnected with DAS
Each GIGASW/is using SAS cards
DECSwitch900 EF DAS are dual homed on SAS card of different GIGASW.
I can't verify that activity of the modules for the moment but I try to
realise what is happening.
1. A and B port are CMT Thrue because connecetd on SAS
2. Only one MAC is present for the DECSWitch so the GIGASWitch(es) see
a loop and use the Spanning Tree to resolve the issue.
My question is what, the point to use S interfaces rather than M ones.
I the example above everything works due to spanning tree and if
recovery need to be started (link failure) it will take more or less
more 1 minute to recover from that event.
Should it not be better to reconfigure the SAS ports of the GIGASW in
M port and use FDDI dual homing functionnality that will be faster to
recover from link failure ?
Thanks,
Robert
|
893.3 | | NPSS::MDLYONS | Michael D. Lyons DTN 226-6943 | Wed Feb 05 1997 09:47 | 16 |
| If the EFs are connected to "S" ports, then dual homing is not
involved. Dual homing in FDDI only refers to connections to "M" ports.
Connecting a DAS device to two SAS devices results in a wrapped
ring from a FDDI perspective. From the perspective of spanning tree,
all three devices are on a ring - spanning tree doesn't care about
whether the ring is wrapped or not.
Whether or not dual homing is preferable depends on the situation.
In general, if you are trying to create a redundant configuration, dual
homing yields faster failover times.
There are lots of notes in this conference discussing redundant
configurations - see the keyword failover-configurations
MDL
|
893.4 | | NPSS::MDLYONS | Michael D. Lyons DTN 226-6943 | Wed Feb 05 1997 09:50 | 4 |
| P.S. Spanning tree recovery using typical defaults will take less than
a minute.
failover time = Max Age + (2 * Forward Delay)
|
893.5 | Yes you are right | NSDP01::RV | | Wed Feb 05 1997 12:29 | 6 |
| Taking the default values
Max Aging Time =15 sec
Forwarding Delay Time 15 sec
we have 45 secs
Robert
|