Why do some routers have problems with filesharing?

Why do some routers have problems with filesharing?

I keep reading that they do have problems - but why?
What is the problem?

I would have thought that once you had an Internet connection you could
do more or less anything with it!
--
Chris

On Sun, 10 Jul 2005 20:41:53 +0100, Chris <nospam@[127.0.0.1]> wrote:

> Why do some routers have problems with filesharing?

They don't, its the users who have the problem of not understanding how
routers work

> I keep reading that they do have problems - but why?
> What is the problem?

Firewalls and lack of knowledge on how they work

> I would have thought that once you had an Internet connection you could
> do more or less anything with it!

You can, you just have to read the manual

On Sun, 10 Jul 2005 20:41:53 +0100, Chris <nospam@[127.0.0.1]> wrote:

>Why do some routers have problems with filesharing?

Routers translate addresses from the internet side to your local
network side. When a filesharer connects to your internet address eg
62.1.1.1, your router has to route that traffic to your PC, which will
have a very different address eg 192.168.0.1. So it performs 'network
address translation' - NAT.
Some software, notably P2P software, doesn't like this as it wants to
connect directly to your PC, without address translation.

>I keep reading that they do have problems - but why?
>What is the problem?

The P2P s/w is badly written.

>I would have thought that once you had an Internet connection you could
>do more or less anything with it!

Apparently you can't make biscuits with it, or have sex with it. So
I'm told.

Mark McIntyre <(E-Mail Removed)> wrote:
> On Sun, 10 Jul 2005 20:41:53 +0100, Chris <nospam@[127.0.0.1]> wrote:
>
>>Why do some routers have problems with filesharing?
>
> Routers translate addresses from the internet side to your local
> network side. When a filesharer connects to your internet address eg
> 62.1.1.1, your router has to route that traffic to your PC, which will
> have a very different address eg 192.168.0.1. So it performs 'network
> address translation' - NAT.
> Some software, notably P2P software, doesn't like this as it wants to
> connect directly to your PC, without address translation.
>
>>I keep reading that they do have problems - but why?
>>What is the problem?
>
> The P2P s/w is badly written.

This is blatantly wrong.

On a properly configured router, filesharing should 'just work'.
However, many routers are basically not exhaustively tested these days.
This means that on demanding applications, such as P2P, which may involve
traffic from tens of thousands of hosts, and dozens of simultaneous
active connections, they can fall over.
I've got two seperate routers, one will just stop routing packets at
all, and fall over in a sulk every so often, and the other has the
management software crash, so the web interface breaks.
They both only do this when P2P is used, with a large number of maximum
connections.

The P2P software is doing nothing wrong, it simply assumes, not unreasonably
that it can make and accept hundreds of connections/hour, without problems.

On Sun, 10 Jul 2005 22:01:06 +0100, Sniffer <(E-Mail Removed)> wrote:

>On Sun, 10 Jul 2005 20:41:53 +0100, Chris <nospam@[127.0.0.1]> wrote:
>
>> Why do some routers have problems with filesharing?
>
>They don't, its the users who have the problem of not understanding how
>routers work

I think I have a reasonable idea how routers work but still spent the
best part of yesterday afternoon trying to get 2 laptops and my
desktop to see each other wirelessly via a new Netgear DG834G. All the
machines could access the router config screen (no adsl) and the
router was not set with any mac security or other 'restrictive'
settings. The same machines could all see each (wirelessly) through
my Netgear AP / Q-Tec router?
>
>> I keep reading that they do have problems - but why?
>> What is the problem?
>
>Firewalls and lack of knowledge on how they work

Windows F/W and ZA wer off on all machines durig the tests.
>
>> I would have thought that once you had an Internet connection you could
>> do more or less anything with it!
>
>You can, you just have to read the manual

I can't think of the last time I 'read the manual' and have managed so
far? Based on the general assumption that most kit defaults to the
most friendy settings and ignoring the fact that M/Softs network
browsing is crap (I was a Netware Instructor so seen how it can work
<g>) there are times when kit just does it's own thing.

Like my Q-Tec router .. you set the WiFi off, the WiFi LED goes off,
it says WiFi is of on the status screen but you can then 'see' an SSID
that you have no control over and connect to the thing with no
security restrictions ..?

All the best ..

T i m

Mark McIntyre wrote:
>
>
> >I would have thought that once you had an Internet connection you could
> >do more or less anything with it!
>
> Apparently you can't make biscuits with it, or have sex with it. So
> I'm told.

I would assume Kraftee and the boys are working on the former, while a
spammer hosted in deepest China is working on the latter.
<g>
Bill

On 11 Jul 2005 00:01:44 GMT, Ian Stirling <(E-Mail Removed)>
wrote:

>Mark McIntyre <(E-Mail Removed)> wrote:
>> On Sun, 10 Jul 2005 20:41:53 +0100, Chris <nospam@[127.0.0.1]> wrote:
>>
>>>Why do some routers have problems with filesharing?
>>
>> Routers translate addresses from the internet side to your local
>> network side. When a filesharer connects to your internet address eg
>> 62.1.1.1, your router has to route that traffic to your PC, which will
>> have a very different address eg 192.168.0.1. So it performs 'network
>> address translation' - NAT.
>> Some software, notably P2P software, doesn't like this as it wants to
>> connect directly to your PC, without address translation.
>>
>>>I keep reading that they do have problems - but why?
>>>What is the problem?
>>
>> The P2P s/w is badly written.
>
>This is blatantly wrong.
>
>On a properly configured router, filesharing should 'just work'.
>However, many routers are basically not exhaustively tested these days.
>This means that on demanding applications, such as P2P, which may involve
>traffic from tens of thousands of hosts, and dozens of simultaneous
>active connections, they can fall over.
>I've got two seperate routers, one will just stop routing packets at
>all, and fall over in a sulk every so often, and the other has the
>management software crash, so the web interface breaks.
>They both only do this when P2P is used, with a large number of maximum
>connections.
>
>The P2P software is doing nothing wrong, it simply assumes, not unreasonably
>that it can make and accept hundreds of connections/hour, without problems.


Agreed. Given that many P2P apps are opening multiple simultaneous
connections to multiple IP addresses, so using NAT faces a number of
problems, inlcuding

mapping ports against IP addresses (port forwarding) (can fail when
more than 1 internal host uses the same port. DC is a prime example of
this, as one has to define a router/firewall rule that maps all
traffic on ports 411 and 412 to a single IP address.

NAT sessions use memory, and a lot of sessions can require more memory
than the router has, or overwhelm the routers processor, especially
when using packet inspection features, resulting in dropped packets or
even dropped sessions.

On Tue, 12 Jul 2005 22:56:36 +0100, PiLGRiM <(E-Mail Removed)>
wrote:

>On 11 Jul 2005 00:01:44 GMT, Ian Stirling <(E-Mail Removed)>
>wrote:
>
>>Mark McIntyre <(E-Mail Removed)> wrote:
>>>
>>> The P2P s/w is badly written.
>>
>>This is blatantly wrong.

FWIW some P2P s/w needs to have unnatted access, for reasons I can
only assume are associated either with fsckwittedness or venality of
the writers.

>on demanding applications, such as P2P, which may involve
>>traffic from tens of thousands of hosts, and dozens of simultaneous
>>active connections, they
(cheap routers)
>can fall over.

Agreed, this is definitely one reason. My original remark stands - if
the P2P s/w were properly written, it'd not try to initiate dozens of
simultaneous connections to consumer-grade equipment.


>>The P2P software is doing nothing wrong, it simply assumes, not unreasonably
>>that it can make and accept hundreds of connections/hour, without problems.

Personally I think thats unreasonable. Consumer routers were never
designed to take that sort of load. Its like buying a maestro and
expecting to be able to use it to carry gravel by the ton.

>mapping ports against IP addresses (port forwarding) (can fail when
>more than 1 internal host uses the same port. DC is a prime example of
>this, as one has to define a router/firewall rule that maps all
>traffic on ports 411 and 412 to a single IP address.
>
>NAT sessions use memory, and a lot of sessions can require more memory
>than the router has, or overwhelm the routers processor, especially
>when using packet inspection features, resulting in dropped packets or
>even dropped sessions.

Absolutely.

On Wed, 13 Jul 2005 21:39:52 GMT, Mark McIntyre
<(E-Mail Removed)> wrote:

>On Tue, 12 Jul 2005 22:56:36 +0100, PiLGRiM <(E-Mail Removed)>
>wrote:
>
>>On 11 Jul 2005 00:01:44 GMT, Ian Stirling <(E-Mail Removed)>
>>wrote:
>>
>>>Mark McIntyre <(E-Mail Removed)> wrote:
>>>>

>>>The P2P software is doing nothing wrong, it simply assumes, not unreasonably
>>>that it can make and accept hundreds of connections/hour, without problems.
>
>Personally I think thats unreasonable. Consumer routers were never
>designed to take that sort of load. Its like buying a maestro and
>expecting to be able to use it to carry gravel by the ton.

Um, I have to differ here. Its not unreasonable to make and drop
hundreds of connections per hour, after all even at the rate of making
one connection every second thats only 3600 connections, so "hundreds
per hour" work out at about one connection every 3 seconds.

Now, hundreds of connections per secod would be unreasonable, or at
least verging on it. However, given that even plain ADSL can go to 8mb
bandwidth, then what defines a resaonble number of connections/second
for a consumber grade ADSL router, given that if it is fit for purpose
it has to be able to cope with an 8mb interface... even stating "near
wire speed routing" is a copout, unless one defines "near wire
speed"...

In article <(E-Mail Removed)>,
(E-Mail Removed) says...
<snip>

> then what defines a resaonble number of connections/second
> for a consumber grade ADSL router,
>
>

The amount of memory that the router has available for it's NAT table
defines the number of connections it can cope with.
--
Mugwump

Reply to 'usenetmail{at}discworld{dot}org{dot}uk