Re: FTP /performance issue question

The routing table has been changed to this:
# netstat -rn
Routing tables
Destination Gateway Flags Refs Use If PMTU Exp
Groups

Route Tree for Protocol Family 2 (Internet):
default 172.18.4.1 UG 316 1429609 en0 - -
-
127/8 127.0.0.1 U 29 925194 lo0 -
- -
172.18.0.0 172.18.4.11 UHSb 0 0 en0 - -
- =>
172.18/16 172.18.4.11 U 1359 4144497 en0 - -
-
172.18.4.11 127.0.0.1 UGHS 18 717836 lo0 - -
-
172.18.4.12 127.0.0.1 UGHS 2 2731 lo0 - -
-
172.18.255.255 172.18.4.11 UHSb 0 0 en0 - -
-

Route Tree for Protocol Family 24 (Internet v6):
::1 ::1 UH 0 16 lo0 - -
-
uxsybproa:/home/unixsa/atownsen
#

What we were seeing is that all traffic local to the subnet was going to
the default gateway first. We observed that using traceroute. Now we are
not seeing that there.

The above routes were created by bringing the adapter down and then back
up.

Here is an odd thing, current if I do an lsdev -Cc if I get this output:

# lsdev -Cc if
en0 Stopped 0E-08 Standard Ethernet Network Interface
en1 Defined 0E-09 Standard Ethernet Network Interface
en2 Defined 0G-08 Standard Ethernet Network Interface
en3 Defined 0L-08 Standard Ethernet Network Interface
en5 Defined 0P-09 Standard Ethernet Network Interface
en6 Defined 0a-08 Standard Ethernet Network Interface
en7 Defined 0a-09 Standard Ethernet Network Interface
en8 Defined 0l-08 Standard Ethernet Network Interface
en9 Defined 0l-09 Standard Ethernet Network Interface

yet the interface that is currently on the network is en0. I'm not sure why
we are seeing the above scenario.

This machine was an HACMP and that we recently stopped (It's a long story
why).


IBM AIX Discussion List <aix-l@xxxxxxxxxxxxx> wrote on 08/15/2006 05:48:58
PM:

Looks correct. AIX 5.x ( not sure what version ) basically puts all
local traffic to the loopback ( which is memory ) rather than actually
sending it out over the wire. Was there something unusual you see ?

lamar

-----Original Message-----
From: IBM AIX Discussion List [mailto:aix-l@xxxxxxxxxxxxx] On Behalf Of
Andrew.Townsend@xxxxxxxxx
Sent: Tuesday, August 15, 2006 8:44 AM
To: aix-l@xxxxxxxxxxxxx
Subject: Re: FTP /performance issue question

Thank you all for the input. To ask another question, what do you think
the
routing table should look like with two IP Addresses on the same subnet.
Currently (based on Lamar's advice) the two IP addresses are on the same
nic. The primary address is 172.18.4.11 and the alias is 172.18.4.12.

Currently this is what the rougint table looks like:

# netstat -rn
Routing tables
Destination Gateway Flags Refs Use If PMTU Exp
Groups

Route Tree for Protocol Family 2 (Internet):
default 172.18.4.1 UG 1752 473278839 en0
- - -
127/8 127.0.0.1 U 29 735210
lo0 - - -
172.18.4.11 127.0.0.1 UGH 18 3153826 lo0
- - -
172.18.4.12 127.0.0.1 UGH 1 24323
lo0
- - -

Route Tree for Protocol Family 24 (Internet v6):
::1 ::1 UH 0 16 lo0 - -
-

Does anyone care to make a comment on the above route table?

Thanks In Advance,


IBM AIX Discussion List <aix-l@xxxxxxxxxxxxx> wrote on 08/14/2006
04:24:21
PM:

One thing you should look at is, traceroute.

Traceroute other_node_name.

Compare your traceroute outputs from both nodes.

Make sure you are not hopping to more network nodes than what is
actually needed.





-----Original Message-----
From: IBM AIX Discussion List [mailto:aix-l@xxxxxxxxxxxxx] On Behalf
Of
Andrew.Townsend@xxxxxxxxx
Sent: Monday, August 14, 2006 12:16 PM
To: aix-l@xxxxxxxxxxxxx
Subject: FTP /performance issue question

Hello,

I'm running a test FTPing files between two AIX servers.

Server proa has two network adapters on the same subnet, both Gig-E.

I've tested with TSM server, FTPing files from the TSM server directly
to
each interface on proa. I've also FTPed files back. Below are some
averages
of transfer times.

X.X.1.11 interface:

Binary: 65536000 bytes sent in 0.7721 seconds (8.289e+04 Kbytes/s)
ascii: 65536000 bytes sent in 2.203 seconds (2.905e+04
Kbytes/s)


X.X.1.12 interface:

Binary: 65536000 bytes sent in 0.7652 seconds (8.364e+04
Kbytes/s)
ascii: 65536000 bytes sent in 2.224 seconds (2.877e+04
Kbytes/s)

FTPing back to the TSM server from proa, the results were pretty
consistent, taking much longer to FTP the files back from proa to TSM
(this
would seem likely considering that I cannot choose which interface to
send
files from):

Binary: 65536000 bytes sent in 1.099 seconds (5.821e+04
Kbytes/s)
ascii: 65536000 bytes sent in 1.297 seconds (4.935e+04
Kbytes/s)

Binary: 6536000 bytes sent in 1.121 seconds (5.707e+04
Kbytes/s)
ascii: 65536000 bytes sent in 1.307 seconds (4.898e+04
Kbytes/s)


I'm trying to troubleshoot why it takes so much longer to send a file
to
proa regardless of which adapter I connect to, rather then sending a
file
from proa. I do realize that when sending the file from proa, I cannot
choose which adapter to send from.

The actual command I am using to send the file once I connect using
ftp
is:
ftp> put "|dd if=/dev/zero bs=32k count=2000" /dev/null

If anyone would care to shed any light why I may be seeing this
symptom,
it
would be much appreciated.

TIA



Privileged and Confidential. This e-mail, and any attachments there
to, is intended only for use by the addressee(s) named herein and
may contain privileged or confidential information. If you have
received this e-mail in error, please notify me immediately by a
return e-mail and delete this e-mail. You are hereby notified that
any dissemination, distribution or copying of this e-mail and/or any
attachments thereto, is strictly prohibited.