Mapped network drives (dis)appear willy nilly - why?

My homebuilt media centre PC is wired to my DG834G router and I connect to
that from my laptop over a wireless connection. Several of the
drives/folders on the server are shared and I have permanent drive mappings
to these defined on the laptop - e.g. M:Music, P:Pictures, S:Software,
V:Video etc..

But access to these drives seems to appear and disappear at random and I
haven't a clue why. Even when I cannot connect to a mapped drive I am still
able to connect to the server via RealVNC and internet access remains solid
but nothing I do will force the mappings to work - it just starts working
again when it is good and ready. Usually rebooting the laptop will bring the
connections back up but they will also eventually reappear when it suits
them.

The server and laptop are both running Windows XP Media Center Edition 2005.
I am running the standard Windows firewall on the laptop and server and have
run the home networking wizard to open up the necessary sharing ports. As I
am behind a NAT router SPI firewall I don't have any other blocking
software.

Also, file transfers do not seem to be that quick, only using a max of ~30%
of the wireless connection's 54Mbps bandwidth.

Any ideas? Thanks, Tim.

Tiny Tim wrote:
> My homebuilt media centre PC is wired to my DG834G router and I connect to
> that from my laptop over a wireless connection. Several of the
> drives/folders on the server are shared and I have permanent drive mappings
> to these defined on the laptop - e.g. M:Music, P:Pictures, S:Software,
> V:Video etc..
>
> But access to these drives seems to appear and disappear at random and I
> haven't a clue why. Even when I cannot connect to a mapped drive I am still
> able to connect to the server via RealVNC and internet access remains solid
> but nothing I do will force the mappings to work - it just starts working
> again when it is good and ready. Usually rebooting the laptop will bring the
> connections back up but they will also eventually reappear when it suits
> them.
>
> The server and laptop are both running Windows XP Media Center Edition 2005.
> I am running the standard Windows firewall on the laptop and server and have
> run the home networking wizard to open up the necessary sharing ports. As I
> am behind a NAT router SPI firewall I don't have any other blocking
> software.
>
> Also, file transfers do not seem to be that quick, only using a max of ~30%
> of the wireless connection's 54Mbps bandwidth.
>
> Any ideas? Thanks, Tim.
>
>
Autodisconnect on server after 15 mins of inactivity should be default
setting.
Try entering this at the cmd line

net config server /autodisconnect-1

should turn it off or use

net config server /autodisconnect:60

will disconnect after 60 mins of inactivity.


Are either of your computers doing any high CPU usage tasks when you get
your low data rate? If so then this can cause a reduction in transfer
rate.I am assuming you are using 11g cards in your computers,a lot of
laptops with built in wireless only have 11b capability and the max data
transfer for this 22Mbps.

nbt

"NBT" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> Tiny Tim wrote:
>> My homebuilt media centre PC is wired to my DG834G router and I connect
>> to that from my laptop over a wireless connection. Several of the
>> drives/folders on the server are shared and I have permanent drive
>> mappings to these defined on the laptop - e.g. M:Music, P:Pictures,
>> S:Software, V:Video etc..
>>
>> But access to these drives seems to appear and disappear at random and I
>> haven't a clue why. Even when I cannot connect to a mapped drive I am
>> still able to connect to the server via RealVNC and internet access
>> remains solid but nothing I do will force the mappings to work - it just
>> starts working again when it is good and ready. Usually rebooting the
>> laptop will bring the connections back up but they will also eventually
>> reappear when it suits them.
>>
>> The server and laptop are both running Windows XP Media Center Edition
>> 2005. I am running the standard Windows firewall on the laptop and server
>> and have run the home networking wizard to open up the necessary sharing
>> ports. As I am behind a NAT router SPI firewall I don't have any other
>> blocking software.
>>
>> Also, file transfers do not seem to be that quick, only using a max of
>> ~30% of the wireless connection's 54Mbps bandwidth.
>>
>> Any ideas? Thanks, Tim.
> Autodisconnect on server after 15 mins of inactivity should be default
> setting.
> Try entering this at the cmd line
>
> net config server /autodisconnect-1
>
> should turn it off or use
>
> net config server /autodisconnect:60
>
> will disconnect after 60 mins of inactivity.
>
>
> Are either of your computers doing any high CPU usage tasks when you get
> your low data rate? If so then this can cause a reduction in transfer
> rate.I am assuming you are using 11g cards in your computers,a lot of
> laptops with built in wireless only have 11b capability and the max data
> transfer for this 22Mbps.
>
> nbt

Thanks for the reply. Armed with your command line parameters above I found
this MS KB article -
http://support.microsoft.com/default...b;en-us;297684 - but that
says that even if the mapped drives disconnect they should reconnect without
problem. Mine simply refuse to re-establish connection. I don't see any red
Xs either. The drives look fine but just don't work. But then a few
hours/minutes later they have magically started working again - just not at
the time when I want them to.

As for the slow network speeds, CPU% is trivial and not a factor. I've never
had really fast transfers. The server is an AMD 64 3000+ with 2*512MB RAM
and the client is a PIII 900 with 2*256MB RAM. The wireless card is a
Netgear WG511 802.11g card running at 54Mbps. An example of use is running a
simple filesync program to copy my mp3s, docs or whatever between the client
and server for backup purposes. There is no other heavy processing going on
and no P2P or anything like that.

Tiny Tim wrote:
8><..........................................
>>>Also, file transfers do not seem to be that quick, only using a max of
>>>~30% of the wireless connection's 54Mbps bandwidth.
>>>
>>>Any ideas? Thanks, Tim.
>>
8><.............................................. .


Just had a look at this http://tinyurl.com/5on9m and data throughput for
the WG511 has been tested at 20.2Mbps.

Just copied this:-
There is a common misunderstanding regarding the bandwidth, the data
rate, and the throughput of a wireless device:


Bandwidth refers to the raw data rate of the device.


Throughput refers to the actual amount of end user data that the
device can transfer in a given time interval.

The result of this misunderstanding is that wireless network users are
frequently disappointed in the wireless throughput (data transfer
speeds) that they experience.

from here http://tinyurl.com/2jnpu

"NBT" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> Tiny Tim wrote:
> 8><..........................................
>>>>Also, file transfers do not seem to be that quick, only using a max of
>>>>~30% of the wireless connection's 54Mbps bandwidth.
>>>>
>>>>Any ideas? Thanks, Tim.
>>>
> 8><.............................................. .
>
>
> Just had a look at this http://tinyurl.com/5on9m and data throughput for
> the WG511 has been tested at 20.2Mbps.
>
> Just copied this:-
> There is a common misunderstanding regarding the bandwidth, the data rate,
> and the throughput of a wireless device:
>
>
> Bandwidth refers to the raw data rate of the device.
>
>
> Throughput refers to the actual amount of end user data that the
> device can transfer in a given time interval.
>
> The result of this misunderstanding is that wireless network users are
> frequently disappointed in the wireless throughput (data transfer speeds)
> that they experience.
>
> from here http://tinyurl.com/2jnpu

My figures are taken directly from the Networking tab of Windows Task
Manager. It knows the connection is a 54Mbps connection and indicates that
throughput is at 30% of the 54Mbps max. Now in truth I don't know if it is
talking about raw bits and bytes or actual useful data. But either way it
seems that only 30% utilisation is a bit rubbish when the only wireless
bandwidth in use is from the laptop to the router, as the server is hard
wired. Signal strength is "Excellent" and the distance to the router is
about 8' with a direct line of site.

I wondered if maybe there are some network settings that could or should be
tweaked, like when you use DR.TCP to fiddle with RWIN and MTU etc. for
improving internet performance. Of course, I don't want to bugger up my
internet in trying to fix my LAN but I don't see how I can get anywhere near
100% utilisation. Perhaps it is simply that I never will.

"Tiny Tim" <_(E-Mail Removed)> wrote in message
news:4226485a$0$562$(E-Mail Removed)...
>
> My figures are taken directly from the Networking tab of Windows Task
> Manager. It knows the connection is a 54Mbps connection and indicates that
> throughput is at 30% of the 54Mbps max. Now in truth I don't know if it is
> talking about raw bits and bytes or actual useful data. But either way it
> seems that only 30% utilisation is a bit rubbish when the only wireless
> bandwidth in use is from the laptop to the router, as the server is hard
> wired. Signal strength is "Excellent" and the distance to the router is
> about 8' with a direct line of site.
>
> I wondered if maybe there are some network settings that could or should
> be tweaked, like when you use DR.TCP to fiddle with RWIN and MTU etc. for
> improving internet performance. Of course, I don't want to bugger up my
> internet in trying to fix my LAN but I don't see how I can get anywhere
> near 100% utilisation. Perhaps it is simply that I never will.
p.s. that translates to a maximum throughput, according to Task Manager, of
about 2MBps. If that's real data throughput then that doesn't seem too
impressive. If that is raw data with overheads etc then that is diabolical
throughput.

Tiny Tim wrote:
8><............................................... .................
> p.s. that translates to a maximum throughput, according to Task Manager, of
> about 2MBps. If that's real data throughput then that doesn't seem too
> impressive. If that is raw data with overheads etc then that is diabolical
> throughput.
>
>
My understanding is that the WG511 operates as a simplex unit(1Tx &
1Rx).While it is transmitting it cannot receive data therefore the max
useful data throughput it theoretically should achieve is 27Mbps (50% of
raw data rate).The CNET tests indicate that they achieved a useful
data throughput of about 19Mbps,see graph in link in earlier post, at
approx 8 foot.I assume this test was done without encryption as I am led
to believe this also reduces the amount of useful data output .If I
assume you are using encryption then I would expect you to be getting a
useful data throughput of 16Mbps (54 x 30/100 = 16.2Mbps) or better.

nbt

On Thu, 03 Mar 2005 09:22:07 +0000, NBT <(E-Mail Removed)>
wrote:

>Tiny Tim wrote:
>8><.............................................. ..................
>> p.s. that translates to a maximum throughput, according to Task Manager, of
>> about 2MBps. If that's real data throughput then that doesn't seem too
>> impressive. If that is raw data with overheads etc then that is diabolical
>> throughput.
>>
>>
>My understanding is that the WG511 operates as a simplex unit(1Tx &
>1Rx).While it is transmitting it cannot receive data therefore the max
>useful data throughput it theoretically should achieve is 27Mbps (50% of
> raw data rate).The CNET tests indicate that they achieved a useful
>data throughput of about 19Mbps,see graph in link in earlier post, at
>approx 8 foot.I assume this test was done without encryption as I am led
>to believe this also reduces the amount of useful data output .If I
>assume you are using encryption then I would expect you to be getting a
>useful data throughput of 16Mbps (54 x 30/100 = 16.2Mbps) or better.
>
>nbt

To look at it in a slightly different way, all wireless access is
equivalent to using a hub instead of a switch on a wired network.

As NBT says, you can only transmit or receive as the same path is used
for both. You might expect, therefore, that if data is flowing in one
direction, you can get the full throughput. (Un)fortunately, TCP/IP
provides for error checking and uses an acknowledgement to confirm
receipt of a valid packet. This means that control data has to pass
back across the link, preventing data being sent. Any collisions (both
ends trying to Tx at the same time) causes both to stop and wait a
"random" time before trying again - resulting in a short period when
no data is transferred in either direction.

Taken together with the time to validate and respond to a received
packet and the addition of packet wrapping to get the data to the
correct place, I would be surprised to see a useful throughput above
35-40%, even without repeats due to bit errors.

PeeGee
--
The reply address is a spam trap. If you need to reply directly, put the UK where it should be - first.

"PeeGee" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> On Thu, 03 Mar 2005 09:22:07 +0000, NBT <(E-Mail Removed)>
> wrote:
>
>>Tiny Tim wrote:
>>8><............................................. ...................
>>> p.s. that translates to a maximum throughput, according to Task Manager,
>>> of
>>> about 2MBps. If that's real data throughput then that doesn't seem too
>>> impressive. If that is raw data with overheads etc then that is
>>> diabolical
>>> throughput.
>>>
>>>
>>My understanding is that the WG511 operates as a simplex unit(1Tx &
>>1Rx).While it is transmitting it cannot receive data therefore the max
>>useful data throughput it theoretically should achieve is 27Mbps (50% of
>> raw data rate).The CNET tests indicate that they achieved a useful
>>data throughput of about 19Mbps,see graph in link in earlier post, at
>>approx 8 foot.I assume this test was done without encryption as I am led
>>to believe this also reduces the amount of useful data output .If I
>>assume you are using encryption then I would expect you to be getting a
>>useful data throughput of 16Mbps (54 x 30/100 = 16.2Mbps) or better.
>>
>>nbt
>
> To look at it in a slightly different way, all wireless access is
> equivalent to using a hub instead of a switch on a wired network.
>
> As NBT says, you can only transmit or receive as the same path is used
> for both. You might expect, therefore, that if data is flowing in one
> direction, you can get the full throughput. (Un)fortunately, TCP/IP
> provides for error checking and uses an acknowledgement to confirm
> receipt of a valid packet. This means that control data has to pass
> back across the link, preventing data being sent. Any collisions (both
> ends trying to Tx at the same time) causes both to stop and wait a
> "random" time before trying again - resulting in a short period when
> no data is transferred in either direction.
>
> Taken together with the time to validate and respond to a received
> packet and the addition of packet wrapping to get the data to the
> correct place, I would be surprised to see a useful throughput above
> 35-40%, even without repeats due to bit errors.
>
> PeeGee

Thanks for the replies. I can see where you're coming from. It still
disappoints me that a "54 Mbps ~ 5.4 MBps" connection will only transfer
data at 2MBps. But it sounds like that's all I can expect. I feel sorry for
those on an 11Mbps network :-(

"PeeGee" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> To look at it in a slightly different way, all wireless access is
> equivalent to using a hub instead of a switch on a wired network.

Communication between a few Ethernet stations connected to a hub will have a
combined throughput slightly less than the data rate. A few percent is lost
due to frame/packet overhead, and a little (remarkably little) due to
collisions.

For whatever reasons, the same does not apply to WiFi. 50% maximum makes
sense if the AP repeats everything it receives. This is like half-duplex
Ethernet where all communication goes via one station.

> As NBT says, you can only transmit or receive as the same path is used
> for both. You might expect, therefore, that if data is flowing in one
> direction, you can get the full throughput. (Un)fortunately, TCP/IP
> provides for error checking and uses an acknowledgement to confirm
> receipt of a valid packet. [...]

However, the sender does not usually wait for acknowledgement of the
previous packet before sending the next. This mechanism (the receive window
or RWIN) exists partly to eliminate the effect of round-trip time (including
that due to processing at both ends), and usually succeeds in doing so.

Alex