[Search for users]
[Overall Top Noters]
[List of all Conferences]
[Download this site]
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 |
1164.0. "Thrupt Problem OSF/1->aVMS via FDDI" by 17311::KENOYER (Greg Kenoyer { SecureAMC/CSC }) Wed Dec 01 1993 10:36
(cross-posted in AOSG::ALPHA_OSF_IFT and UPSAR::FDDI)
I've a customer who seems to be having a problem with FDDI thruput. He
reports ethernet speeds (2 Mb/sec) instead of FDDI speeds (60+ Mb/sec).
He is doing a memory-to-memory transfer via TCP. The machines are on the
same subnet (hosts 131.1.225.1 and 131.1.225.2 with netmask 255.255.255.0)
One oddity is that ftp over this link reports 6 Mb/s; this seems to imply
an application problem?
Questions:
1) Searching the FDDI, OSF, and VMS conferences I find most folks getting
well over the (expected) 60 Mb/s on both OSF and VMS. There is no mention
of special tuning or other non-obvious steps taken. Is special tuning
required?
2) There are multiple devices on the Turbochannel, could there be some sort
of bandwith/timing problem?
3) Can anyone point me to a simple (and releasable) piece of benchmarking
code? (OSF -> VMS)
5) Is the 6 Mb/s ftp thruput reasonable?
4) What else should we examine? Note that while I'm quite happy to use the
U-word, I've limited experience on OSF/1 and none on FDDI; the actual
command needed to obtain the info would be appreciated.
*********************************************************************
*********************************************************************
These two systems are (usually) connected via a DECconcentrator 500,
however the same thruput is seen when directly connected.
*********************************************************************
*********************************************************************
Transmitting system:
DEC 3000/500 with 256 MB memory
OSF/1 v1.2 (with RT kernal)
DEFZA controller
Genroco (sp?) drives (on IPI?), with patched drivers
VME i/f, with patched drivers
--> note that they're afraid to move to v1.3 due to the drivers for the
above two special devices. The patches/drivers were obtained both from
DEC and the folks who supplied the orig. hw.
uerf on the startup record (rec 300)
__________________________________________
CPU SUBTYPE: KN15AA
Alpha Boot: available memory from
_0x8f0000 to 0xe000000
DEC OSF/1 [RT] V1.2 (Rev. 10); Sat Nov
_20 20:12_02 MST 1993
physical memory = 222.00 megabytes
available memory = 206.00 megabytes
using 852 buffers containing 6.65
_megabytes of memory
tc0 at nexus
scco at tc0 slot 7
asco at tc0 slot 6
rz3 at asc0 bus 0 target 3 lun 0 (DEC
_ RZ26 (C) DEC T386)
rz4 at asc0 bus 0 target 4 lun 0 (DEC
_ RRD42 (C) DEC 4.5d)
asc1 at tc0 slot 6
tz8 at asc1 bus 1 target 0 lun 0 (DEC
_ TKZ09 90097000 045H)
rb0 at tc0 slot 8
1280X1024
ln0: DEC LANCE Module Name: PMAD-BA
lm0 at tc0 slot 7
ln0: DEC LANCE Ethernet Interface,
_hardware address: 08:00:2b:37:8a:17
fza0 at tc0 slot 0
fza0: DEC DEFZA FDDI Interface,
_hardware address: 08:00:2b:35:9c:41
_ROM rev 1.0 Firmware rev 1.2
ipi0 cache size = 0 Mbytes
ipi0 at tc0 slot 1
ipi0 at tco slot 2 (3VIA/MVIB)
* VME probe routine - SVA addr1
fffffc0140080000
dmaex0 at vba0 csr 0x4000 vmea16d32
_supervisory data mode vec 0x40
_priority 3
DEC3000 - M500 system
Firmware re4vision: 2.1
PALcode: OSF version 1.28
lvm0: configured.
lvm1: configured.
ra1 at ipi0 slave 1 (IPI91)
ra0 at ipi0 slave 0 (IPI91)
_
output of netstat -m
__________________________________________
25/128 mbufs in use:
18 mbufs allocated to data
7 mbufs allocated to socket names and addresses
20/50 mapped pages in use
400 Kbytes allocated to network (40% in use)
1 requests for memory denied
0 requests for memory delayed
0 calls to protocol drain routines
__________________________________________
output of netstat -i
__________________________________________
Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll
ln0 1500 LAT none 1008 0 1010 0 6
ln0 1500 <Link> 1008 0 1010 0 6
ln0 1500 131.1.224 IDS 1008 0 1010 0 6
fza0 4352 LAT none 98 0 2003 0 0
fza0 4352 <Link> 98 0 2003 0 0
fza0 4352 131.2 ids_fddi 98 0 2003 0 0
s10 296 <Link> 0 0 0 0 0
s11 296 <Link> 0 0 0 0 0
lo0 1536 <Link> 89266 0 89266 0 0
lo0 1536 loop localhost 89266 0 89266 0 0
__________________________________________
output of fddi_config -I fza0 -s
__________________________________________
Firmware Revision: 1.2
ROM Revision: 1.0
SMT Version ID: 1
Requested TRT: 8.000 ms
Maximum TRT: 173.015 ms
Valid Transmission Time: 2.621
LEM Threshold: 8
Restricted Token Timeout: 1000.000 ms
PMD Type ANSI Multimode
fza0 ANSI FDDI settable parameters
Token Request Time:: 8.0000 ms
Valid Transmission Time: 2.6214
LEM Threshold: 8
Restricted Token Timeout: 1000.000 ms
Ring Purger State: Purger off
>> the fza0 FDDI counters show no errors, unrecognized frames, dups, rejects,
etc.
*********************************************************************
*********************************************************************
Receiving system:
DEC 3000/400 with 64 MB
OpenVMS v1.5
UCX 3.0
DEFTA controller
mtu is 4348
No more info on this one: tell me what you need!
T.R | Title | User | Personal Name | Date | Lines |
---|
1164.1 | | NETRIX::thomas | The Code Warrior | Wed Dec 01 1993 12:10 | 1 |
| DEFZAs suck. Upgrade it to a DEFTA.
|