Pascal Favre Pascal.Favre at gmx.net
Thu Dec 8 13:20:58 EST 2005

Here is a SYSTRAN Translation of my previous mail:

Mail to Devolo:

I got mine on Tuesday 05.07.2005 DL on Highspeed starter package.
The installation ran problem-free. In differently areas I have measurements
through-taken. Results are ernuechternd.

The consideration to the purchase was the building of a file server in the cellar for
Video photographs. A client in the bedrooms (3 recording channels and 1
Playing channel) and next year a client in the living room (1 recording channel
and 1 playing channel). For each channel becomes a transfer throughput of 0.6MB
needed. Totally 3.6MB/sec. Content I would already be with 2.4MB/sec.

I have my tests with a Linux (2.8Ghz, into different areas in EEC
and cellars) and a Solaris Sparc Ultra-10 440Mhz (in the office) accomplished.
When transportation minutes, which are to be used in the employment, have
I NFS with different write/read buffer sizes and udp/tcp alternatively

The following instruction was adapted in each case and on the Linux computer
mount t NFS o "rsize=4096, wsize=4096, nfsvers=3, UDP" sol:/export/store
/ mnt/sttest
dd if=/dev/zero of=/mnt/sttest/a bs=64k count=1024

The measurements were based on two different days.
Here the result:

Two computers at a SWITCH (without HighSpeed!)
10.8MB to 11.5MB

rsize=4096 - 10.8MB
rsize=65536 - 11.5MB

No large difference is to be constituted between UDP and tcp.

Resembles plug socket: Signal rate 72Mbit
Bedroom: 62Mbit

rsize=2048 - 2.8MB
rsize=4096 - 3.4MB, 3.3MB (second day)
rsize=8196 - 2.8MB

rsize=4096 - 2.8MB, 3,0 MT
rsize=8196 - 2.8MB, 3,2 MT
rsize=16384 3.2MB, 3,4 MT
rsize=32768 3.3MB, 3,4 MT
rsize=65536 3.3MB, 3,3 MT

Both (plug socket/bedrooms) were identical, despite 10Mbit
Signal difference.
TCP is to before courses.
UDP has only an optimum with rsize/wsize of 4096.
The maximum is an information flow-rate of 3.4MB.

Cellar location 1: Signal 11Mbit
Values between 320KB and 400KB, dependent on tcp/udp and buffer size

Cellar location 2: Signal 20Mbit
Values between 350KB and 480KB

Cellar location 3 (auxiliary building): 9Mbit
No measurements accomplished.

Repetitions resulted in in the case of same attitude comparable values.
Rule measure is always somewhat positive.
Unfortunately results are my project a file server in the cellar too with this
platzieren completely burst!

Living room: 44Mbit - day 1
1.3MB - 1.6MB

Living room: 46Mbit - day 2
1.1MB - 1.2MB

Even with a higher signal value, worse results (-25%).
On the second day eineWaschmachine which possibly ran to this result
contributed (Tumbler/Trockenraum did not run).

Also tests with ftp were made. There here the buffer sizes not
to be optimized can, were somewhat worse test results.
In one their company QA/REPORTS (she cannot find unfortunately any longer) I could
read that only 50Mbit can be used, this enspricht only 60% of
85Mbit. They become no injustice with your interpretation (similarly how
this with WLAN to be made) to have. But as a customer I fill
been deceptive.
With a SWITCH of 100Mbit I receive with an optimal buffer size
Attitude 11.5MB (departure Overhead ~5% considers).
Nevertheless of 50Mbit I expect one into the most optimal constellation
Information flow-rate from for instance 6MB (TCP/UDP/IP Overhead taken off).
Only half (3.4MB) was maximum reached.
In the dwelling 1.2MB were measured as maximum in one of the rooms,
which corresponds only to 20% (relative to 50Mbit).
In the cellar one can count this at the most on 400KB, corresponds to about 6%
(relative to 50Mbit).

In the appendix you find to two files with results of test, however without
They found some critical ones words, hope here I you become these
to understand constructional can.

My desire to receive I do not have 3.6MB also into the best test
reached. The maximum was 3.4MB.
An employment in the cellar is completely impossible (~400KB).
Only one file server remains in the office as alternative. But I do not need
DLAN Highspeed in the bedroom. A Ethernet connection exists already.
Planned for close test year a client in the living room, but with one
Information flow-rate of ~1.2MB is scarcely limited this.

Since the cellar with different walls and soils is bent, is WLAN
no guaranteed alternative for me. At the most the new MIMO Technology
is promising. This was to be set the reason on DLan Highspeed. I
are however very entaeuscht over the achievement.


Mail from Devolo support

the data transmission rates indicated in the Informer cannot be converted indeed directly to the data transmission rates the 

>From the data transmission rate angezeiten there log data for TCP takes, Ethernet and DL on approx.. 65% in requirement, soda for 
the pure data communication normally approx.. 35% in accordance with the announcement in the Informer are available. Their mentioned 
20% are indeed less, than would be to be expected.

This could be connected with the following interference sources in electricity mains:

DL on/DL on USB:
Contact strips with switch and/or Ueberspannungschutz/master Slave circuit
Transmission over several phases
viele/starke consumers in the proximity DL on
(computer, monitor, Notebooknetzteile, lamps with dimmer)
furnished network bridge
furnished Firewalls

If you liked to return the devices nevertheless, this would be unfortunately not direct only over your dealer possible, over us.

Yours sincerely

Dirk Heckel

Its Devolo support team

Please you answer to this Mail without further modification
the reference field. This simplifies the processing of the Mails.
Also please always quote the entire email process.

 Service center service LINE: 01805 338656 (0.12 euro/minute)
Lock beautiful outer E-Mail: support at devolo.de
 STR. 6-8 InterNet: http://www.devolo.de
 D-52072 Aachen

More information about the mythtv-users mailing list