| Title: | FDDI - The Next Generation |
| Moderator: | NETCAD::STEFANI |
| Created: | Thu Apr 27 1989 |
| Last Modified: | Thu Jun 05 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2259 |
| Total number of notes: | 8590 |
I work at the CSC in Colorado and one of our customer is having a
performance problem. They beleave it maybe caused by the FDDI. When the
customer looks at the counters on two of the ALPHA systems he notices
+20 Unalign 1seg xcopies and +24 Unalign 2seg xcopies incrementing.
The values currently are 33144755 and 46533630.
I tried to find an explanation on these two counters but was
unsuccessful. These are the only two counters that are indicating that
there maybe a problem on the FDDI. The systems are alpha station 1000
with DEFEA controlers, storage works boxes.
A simplified configuration is each storage works box has a full duplex
connection to a giga switch. The operating systems are in an fddi
ring attached to a port on the gigi switch.
Any documentation that I could find at the CSC does not explain all
the errors that an fddi port may log. I could only find these two errors
listed in the frdriver listing but there was no explanation as to
what they were.
An explanation of these errors or a pointer to where I could find
an explanation would be appreciated. I would like to have an FDDI
specification manual in the CSC library.
Tony Abdella Networks Support
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 1936.1 | NETCAD::STEFANI | Machines to humanize | Fri Jan 26 1996 17:52 | 7 | |
>>+20 Unalign 1seg xcopies and +24 Unalign 2seg xcopies incrementing.
>>The values currently are 33144755 and 46533630.
These counters mean nothing to me. I'd QAR this against the
device driver you're running.
- Larry
| |||||
| 1936.2 | FRDRIVER version + | 32895::SLAGLE | Mon Jan 29 1996 09:43 | 29 | |
The driver is:
Image Identification Information
image name: "SYS$FRDRIVER"
image file identification: "X-3"
image file build identification: "X61Q-SSB-0000"
link date/time: 4-MAY-1995 22:59:21.84
linker identification: "All-12"
SDA> Device name FR_DEFEA
SDA> HW version 322E3320
These two counters are are incrementing similarly on two
Alphaserver 1000 4/2 machines running VMS 6.2. These two Alpha boxes
are part of an HS241 Storageworks FDDI server. Both Alphas are
connected point-to-point to GIGAswitch ports and are operating in FDX
mode.
The 4 other cluster members are also connected point-to-point to the
same GIGAswitch but are not operating in FDX mode. They all run VMS
6.1 and driver version:
"SYS$FRDRIVER"
"X-3"
"X5SC-SSB-0000"
14-APR-1994 11:47:30.53
"T11-11"
and are HW version 322E3436 according to SDA. This version of the
driver does not provide these counter fields in the SDA printout.
Also, this version of VMS does not enable FDX on the DEFEAs.
| |||||
| 1936.3 | STAR::STOCKDALE | Wed Jan 31 1996 11:30 | 17 | ||
>> +20 Unalign 1seg xcopies and +24 Unalign 2seg xcopies incrementing. >> The values currently are 33144755 and 46533630. These counters are merely internal counters that the driver maintains that are displayed by SDA. They simply say that the driver was forced to copy that many transmit packets to aligned transmit buffers to work around a problem with that revision of the DEFEA which requires longword-aligned transmit buffers. To determine if the performance problem the customer sees is caused by this extra buffer copy, you'll have to give more details why the customer thinks there is a problem. The DEFEA can only do about 80 mbits transmit anyway, and at that transmit rate if all packets are unaligned there is about a 20% cpu performance hit. The performance hit is much less under more typical usage. - Dick | |||||