Re: PING: Jeff Liebermann - Netgear WGT634U Storage Router Info!

On Mon, 05 Sep 2005 04:17:26 GMT, Jim Donald <jim_donald@xxxxxxxxxxx>
wrote:

>108Mbps is a bit of a marketing thing, huh?

Larger numbers are better, according to marketing. What they don't
tell you is that there's LOTS of overhead to Wi-Fi and you'll be lucky
to get about half that in thruput, and only at limited ranges. That's
why I always use "connection speed" and "thruput" to distinguish
between the numbers.

You might wanna download and read:
| http://www.intel.com/business/bss/infrastructure/wireless/deployment/hotspot.pdf
which is Intel's instructions on how to setup a wireless network.
Table 4-7 is a chart of range versus connection speeds. I don't agree
with their numbers, but they have the right idea. Divide the range in
ft by 1/3 to get a more realistic value.

>I thought as much. Yet
>50Mbps isn't too bad for me - or even 30Mbps. I had no idea that
>'connection speed' and 'data speed' would be different. When the tray
>icon said 'signal excellent - connected at 54Mbps' I just assumed that
>would also be the data throughput speed. Doh...;-)

Well, you're learning. It's common knowledge. See:
| http://www.tomsnetworking.com/Sections-article60-page5.php
It's a "walk test" comparing thruput with and without Super-G. Note
that the higher speeds are fairly well maintained throughout the
entire 300ft (center of graph) walk. I've done similar tests and
gotten similar results. The higher speed devices are generally better
overall, but not always at any given point or location.

>With regard to my problem. As a last resort (before doing anything
>major) I took my life in my hands and scaled a very shaky set of
>ladders to place the client antenna much higher up on the wall near
>the ceiling. I stuck the base on with some Blu-Tack rather than the
>supplied Velcro strip and now my speeds are around the correct levels
>you mentioned. Not 54Mbps - but between 15 and 30 most times.

Bingo. Now I know what was happening when your speed was dropping to
1Mbit/sec. You were dealing with indoor reflections and multipath.
These tend to be fatal to thruput. They reflected signals arrive at
inconvenient times (between packets), smear data, mangle timing, and
generally create packet errors. The access point defends itself by
slowing down, which increases the receiver sensitivity and reduces the
effects of multipath. You probably could have moved around a bit and
improved the situation.

>Also - would I be right in thinking that repositioning one's antenna
>and/or router even 'slightly' (by a mere few inches) can make a BIG
>difference to performance? Or is my improvement just a fluke?

Yes, antenna location and position is very critical to proper
operation. The idea is to get the best direct path. If you're
dealing with reflections, things will not work well. If you're in a
null, where two or more signals cancel, then moving 1/4 wavelength in
any direction will eliminate the effects. 1/4 wavelength is about
3cm.

>Otherwise your alternative advice is filed away for future reference.
>
>Further - I agree with you about the web server thing? One of THOSE in
>a router would be really cool.

It would be fairly easy to implement as the router already has a web
server in firmware used for configuration. Personally, I would prefer
to see the USB memory used for log file collection, authentication,
and SNMP based performance data collection, but that would probably be
to weird for a consumer product. Still, a "personal web server" would
be handy.

>P.S. Despite what it says in the WGT634U manual, with the right
>previously-set permissions, one CAN actually 'write' to any attached
>storage device over the Internet. I tried it and it works fine.

Yech. Passwords passed in the clear with no encryption. Easily
sniffed. No SFTP (secure ftp) or SCP (secure copy). No logging or
ftpaccess. Are you sure you wanna run such a crude ftp server?

Incidentally, on the password issue, one of my clueless customers
insisted on using the same password for everything. His POP3 password
for email got sniffed by someone. With the email address, he figured
out my clients eBay account name. The password worked. Fortunately,
his Paypal account used a different password. If you do play the
password game, kindly use a different password on everything,
especially easily sniffed passwords such as found on simple ftp
servers.


--
Jeff Liebermann jeffl@xxxxxxxxxxxxxxxxxxxxxx
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
.

Relevant Pages

  • Re: network troubleshooting
    ... Is there anyway I can test the line speed without setting up ftp? ... have FTP Service setup on the SBS Server and I dont want to break anything ... > The netmon tool that comes with SBS2003 can be useful isolating network ... > speeds, a simple test might be to set up an FTP site on the LAN interface ...
    (microsoft.public.windows.server.sbs)
  • Re: Silly query !
    ... > Connection Speeds - a Byte is normally 8 bits, ... > than FTP Transfers... ... > YouSendit.com Server?) ... > over an ADSL Connection that has a fairly high Contention Ratio ...
    (uk.people.silversurfers)
  • RE: Slow Network Speed over WAN
    ... that they'll fragment as they are shipped into the WAN. ... Windows NIC). ... Copies from the West Coast server to another in the same site happen at ... Those speeds are representative of either site (when testing against a test ...
    (microsoft.public.windows.server.networking)
  • RE: Slow Network Speed over WAN
    ... Have a look to make sure you are not fragmenting encrypted packets on either ... Windows NIC). ... Copies from the West Coast server to another in the same site happen at ... Those speeds are representative of either site (when testing against a test ...
    (microsoft.public.windows.server.networking)
  • Re: Spped of FreeWare FTP-Server under WinXP?
    ... Seems to me there are only a few things that would affect FTP speeds among ... To even test this you'd need a carefully controlled LAN ... Since FTP protocol is the same regardless of package, ... to the server which may slow down multiple file transfers or directory listings. ...
    (microsoft.public.windows.server.networking)