RE: Virtual Ethernet - Part II

Our physical LAN is a 1GB connection, so the actual line speed should be
comparable.

Interestly, I tried changing the Maximum frame size on the VLAN to 1496
to match the LAN settings, and found that FTP speeds increased by 25 -
30%, while SAVRSTxxx speeds decreased by 50% or more.

VLAN utilization for FTP went up to 12% and VLAN utilization for
SAVRSTxxx commands dropped to 4%. During testing on a 100M connection
the utilization went over 70% while running a large FTP transfer.

In discussions of Virtual Ethernet I've seen usage of the terms
"Lightning Fast" and "Screams" So I was expecting the VLAN to be some
magnitude faster than a 1GB physical LAN over Cat 5 wire. Since the
highest VLAN line utilization I've seen so far is 20% I would expect
that some configuration could increase this utilization.

Regards,

Scott Ingvaldson
Senior IBM Support Specialist
Fiserv Midwest


-----Original Message-----
From: Luis Colorado [mailto:LuisC@xxxxxxxxxxxxxxxx]
Sent: Friday, January 25, 2008 8:06 AM
To: Midrange Systems Technical Discussion
Subject: RE: Virtual Ethernet - Part II

So, what was the line speed for the physical LAN connection?

-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Pete Helgren
Sent: Thursday, January 24, 2008 6:09 PM
To: Midrange Systems Technical Discussion
Subject: Re: Virtual Ethernet - Part II

Scott,

Here is what I got at a customer who has a "relatively" new 520:

Line Configuration:

Line description . . . . . . . . . : VETH0
Option . . . . . . . . . . . . . . : *BASIC
Category of line . . . . . . . . . : *ELAN

Resource name . . . . . . . . . . : CMN03
Online at IPL . . . . . . . . . . : *YES
Vary on wait . . . . . . . . . . . : *NOWAIT
Network controller . . . . . . . . : VETH0NET
Local adapter address . . . . . . : CEF3D0003002
Exchange identifier . . . . . . . : 056ADB0F
Ethernet standard . . . . . . . . : *ALL
Line speed . . . . . . . . . . . . : 1G
Current line speed . . . . . . . . : 1G
Duplex . . . . . . . . . . . . . . : *FULL
Current duplex . . . . . . . . . . : *FULL
Maximum frame size . . . . . . . . : 8996
Maximum controllers . . . . . . . : 40

I tried to find a big enough file to approximate what you are doing:

FTP over VLAN:

small file:
48362688 bytes transferred in 1.627 seconds. Transfer rate 29725.926
KB/sec big file
5515090944 bytes transferred in 294.812 seconds. Transfer rate 18707.176

FTP over network

small file
48362688 bytes transferred in 5.215 seconds. Transfer rate 9274.044
KB/sec.
big file
5515090944 bytes transferred in 599.517 seconds. Transfer rate 9199.229

Hope that is of some use.

Pete Helgren



Ingvaldson, Scott wrote:

I'm having a hard time believing that disk is the bottleneck here.
That would make the VLAN run at the same speed as the network, not

slower.

And I wouldn't be able to save this 60 GB library to a save file in
under 3 minutes.

My apples to apples comparison run today:

FTP over VLAN: 59521118976 bytes transferred in 4777.693 seconds.
Transfer rate 12458.129 KB/sec.

FTP over network: 59521118976 bytes transferred in 3808.596 seconds.
Transfer rate 15628.102 KB/sec.

Does anyone have any insight into the VLAN configuration or at least
some real world performance numbers?

Regards,

Scott Ingvaldson
Senior IBM Support Specialist
Fiserv Midwest


-----Original Message-----
From: Luis Colorado [mailto:LuisC@xxxxxxxxxxxxxxxx]
Sent: Thursday, January 24, 2008 9:25 AM
To: Midrange Systems Technical Discussion
Subject: RE: Virtual Ethernet - Part II

Well, if you think about it, 21577 Kbytes/sec is roughly equivalent to
169 Mbits/sec, meaning that your transfers through the network were
using less than 10% of your 2Gb/sec network. Years ago, with
10Mbit/sec and 100Mbit/sec network speeds, clearly the bottleneck was

the network.

However, with your 2Gb/sec connections, your bottleneck is now your
system (more specifically, your disk).

So, bottom line (and I don't know much about VLAN optimization), I
would say that this is as good as it gets.

-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Ingvaldson,

Scott

Sent: Thursday, January 24, 2008 10:00 AM
To: midrange-l@xxxxxxxxxxxx
Subject: Virtual Ethernet - Part II

OK, the VLAN appears to be working as designed but I still have some
questions.

The speed of our Anynet transfers has increased by a factor 4 - 8.
Transfers that were taking 3 - 4 hours now complete in 30 - 45

minutes.

As long as I do them one at a time. If I do more than one at a time
they slow considerably.

FTP transfers take the same amount of time using the VLAN as they did
running over the network. Reported transfer rates are in the range of

12000 - 14000 KB/sec. Strangely enough, I was able to hit 21577

KB/sec.

on a 2 GB transfer, not using the VLAN, doing an FTP to a Network
Server!

An iNav monitor shows the VLAN utilization to be 8 - 10% when doing
FTP and up to 20% for SAVRSTxxx commands. But if I run two
simultaneous SAVRSTxxx transfers the utilization drops to 15%, even if

the transfers are initiated from different LPARs.

Is there some tuning that can be done to the VLAN or is this as good
as it gets?

Regards,

Scott Ingvaldson
Senior IBM Support Specialist
Fiserv Midwest
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing

list To post a message email: MIDRANGE-L@xxxxxxxxxxxx To subscribe,
unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx Before posting, please take
a moment to review the archives at

http://archive.midrange.com/midrange-l.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing
list To post a message email: MIDRANGE-L@xxxxxxxxxxxx To subscribe,
unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx Before posting, please take a
moment to review the archives at http://archive.midrange.com/midrange-l.