network limited to 50%, policy fault?

Every network activity, no matter if it's between servers or workstations very rarely passes 50% of use(combined or upload/download), it also doesn't matters the network utilization, it happens 24x7.
Some times it passes to 60/65%, but that's only between workstations
We've tried different brand switches, and the solution is the same.

After several tests, when i used a backup software that uses a propietary data move(not smb), the troguhtput bordered 99% no matter what switch i used.

I even tested this in a isolated network between a win2003 non-domain and a XP station and the same happened.

Now, is there some default hidden policy that murders network bandwidth in windows 2003(or XP for that matter)?

"Guillermo Lovato" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
Now, is there some default hidden policy that murders network bandwidth in
windows 2003(or XP for that matter)?

It isn't "murdered". That is the way it is supposed to be,...unless you
only ever want a maximum of two machines to ever use the LAN at one time
while the rest of the other wait till they are done. How would multiple
machines use the LAN at the same time if the first two that started
communicating used 100% of the bandwidth?

--

Phillip Windell [MCP, MVP, CCNA]
www.wandtv.com

<@.> wrote:
> "Guillermo Lovato" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
> Now, is there some default hidden policy that murders network
> bandwidth in windows 2003(or XP for that matter)?
>
> It isn't "murdered". That is the way it is supposed to be,...unless
> you only ever want a maximum of two machines to ever use the LAN at
> one time while the rest of the other wait till they are done. How
> would multiple machines use the LAN at the same time if the first two
> that started communicating used 100% of the bandwidth?

First off, a switched LAN gives 100Mb/s in each direction between any two
points.

Secondly, a server should provide data at the maximum speed it can to the
requests it has at the time.

Geoff

"Geoff Winkless" <usenet-at-geoff-dot-dj@[127.0.0.1]> wrote in message
news:427111c5$0$289$(E-Mail Removed) t...
> <@.> wrote:
> > "Guillermo Lovato" <(E-Mail Removed)> wrote in message
> > news:(E-Mail Removed)...
> > Now, is there some default hidden policy that murders network
> > bandwidth in windows 2003(or XP for that matter)?
> >
> > It isn't "murdered". That is the way it is supposed to be,...unless
> > you only ever want a maximum of two machines to ever use the LAN at
> > one time while the rest of the other wait till they are done. How
> > would multiple machines use the LAN at the same time if the first two
> > that started communicating used 100% of the bandwidth?
>
> First off, a switched LAN gives 100Mb/s in each direction between any two
> points.
>
> Secondly, a server should provide data at the maximum speed it can to the
> requests it has at the time.

It doesn't. I have never seen it do 100% with a single session on any LAN I
have ever seen with any OS.

--

Phillip Windell [MCP, MVP, CCNA]
www.wandtv.com

in my home, i have 2 XP machines with a switch(a cheap router/switch), and network efficiency is usually above 90%.
in my office, with large switchs and powerful computers, it's limited to 50%.

Phillip, i also said that between ANY 2 machines i get 50%(i.e. Server->PC2 , PC3->PC4 at the same time)

"Phillip Windell" <@.> wrote in message news:(E-Mail Removed)...
> "Geoff Winkless" <usenet-at-geoff-dot-dj@[127.0.0.1]> wrote in message
> news:427111c5$0$289$(E-Mail Removed) t...
> > <@.> wrote:
> > > "Guillermo Lovato" <(E-Mail Removed)> wrote in message
> > > news:(E-Mail Removed)...
> > > Now, is there some default hidden policy that murders network
> > > bandwidth in windows 2003(or XP for that matter)?
> > >
> > > It isn't "murdered". That is the way it is supposed to be,...unless
> > > you only ever want a maximum of two machines to ever use the LAN at
> > > one time while the rest of the other wait till they are done. How
> > > would multiple machines use the LAN at the same time if the first two
> > > that started communicating used 100% of the bandwidth?
> >
> > First off, a switched LAN gives 100Mb/s in each direction between any two
> > points.
> >
> > Secondly, a server should provide data at the maximum speed it can to the
> > requests it has at the time.
>
> It doesn't. I have never seen it do 100% with a single session on any LAN I
> have ever seen with any OS.
>
> --
>
> Phillip Windell [MCP, MVP, CCNA]
> www.wandtv.com
>
>
>

<@.> wrote:
> "Geoff Winkless" <usenet-at-geoff-dot-dj@[127.0.0.1]> wrote in message
> news:427111c5$0$289$(E-Mail Removed) t...
>> <@.> wrote:
>>> "Guillermo Lovato" <(E-Mail Removed)> wrote in message
>>> news:(E-Mail Removed)...
>>> Now, is there some default hidden policy that murders network
>>> bandwidth in windows 2003(or XP for that matter)?
>>>
>>> It isn't "murdered". That is the way it is supposed to be,...unless
>>> you only ever want a maximum of two machines to ever use the LAN at
>>> one time while the rest of the other wait till they are done. How
>>> would multiple machines use the LAN at the same time if the first
>>> two that started communicating used 100% of the bandwidth?
>>
>> First off, a switched LAN gives 100Mb/s in each direction between
>> any two points.
>>
>> Secondly, a server should provide data at the maximum speed it can
>> to the requests it has at the time.
>
> It doesn't. I have never seen it do 100% with a single session on any
> LAN I have ever seen with any OS.

You started out claiming that's "the way it is supposed to be" based on a
poor understanding of LAN technology. Now you're simply relying on
observational evidence?

Can I suggest next time you don't post authorititively on something about
which you know little?

Geoff

"Geoff Winkless" <usenet-at-geoff-dot-dj@[127.0.0.1]> wrote in message
news:42774b91$0$298$(E-Mail Removed) t...
> > It doesn't. I have never seen it do 100% with a single session on any
> > LAN I have ever seen with any OS.
>
> You started out claiming that's "the way it is supposed to be" based on a
> poor understanding of LAN technology. Now you're simply relying on
> observational evidence?
>
> Can I suggest next time you don't post authorititively on something about
> which you know little?

Then bless us with the truth.

Why do I see what I see? Why does everyone else see what they see? Why is
the original poster having the issue he thinks he has? How would two
machines on Ethernet receive communication from more than one other machine
at the same time if the first session they took part in used 100% of the
bandwidth and saturated the line between it and the switch?

Be useful,..be constructive.


--

Phillip Windell [MCP, MVP, CCNA]
www.wandtv.com

<usenet-at-geoff-dot-dj@[127.0.0.1]> wrote:
> <@.> wrote:
>> How
>> would multiple machines use the LAN at the same time if the first two
>> that started communicating used 100% of the bandwidth?
>
> First off, a switched LAN gives 100Mb/s in each direction between any
> two points.

Re-reading this I realise that I oversimplified .

Of course there's no way each point-to-point connection could have 100Mb
since there's only 1 cable (!!): I should have said that a properly designed
switch must be capable of arbitrating a full duplex 200Mb to each point...
the point being that two machines using 100% will not stop other machines on
the LAN from talking.

Geoff

<@.> wrote:
> "Geoff Winkless" <usenet-at-geoff-dot-dj@[127.0.0.1]> wrote in message
> news:42774b91$0$298$(E-Mail Removed) t...
>>> It doesn't. I have never seen it do 100% with a single session on
>>> any LAN I have ever seen with any OS.
>>
>> You started out claiming that's "the way it is supposed to be" based
>> on a poor understanding of LAN technology. Now you're simply
>> relying on observational evidence?
>>
>> Can I suggest next time you don't post authorititively on something
>> about which you know little?
>
> Then bless us with the truth.
>
> Why do I see what I see? Why does everyone else see what they see?
> Why is the original poster having the issue he thinks he has?

I'd be tempted to say network card problem except the OP states that non-SMB
traffic achieves 99%, which means the hardware and network topology can be
pretty much discounted as a reason altogether.

Just trying now by (SMB) copying a large file I've achieved around 85%
utilization between two XP workstations and a 2k server on the network here
and 77% between two 2k3 servers, all of which are under other load.

> How
> would two machines on Ethernet receive communication from more than
> one other machine at the same time if the first session they took
> part in used 100% of the bandwidth and saturated the line between it
> and the switch?

It is down to the switch to arbitrate when another machine sends data to
that port.

Geoff

"Geoff Winkless" <usenet-at-geoff-dot-dj@[127.0.0.1]> wrote in message
news:4277a813$0$305$(E-Mail Removed) t...
> > Why do I see what I see? Why does everyone else see what they see?
> > Why is the original poster having the issue he thinks he has?
>
> I'd be tempted to say network card problem except the OP states that
non-SMB
> traffic achieves 99%, which means the hardware and network topology can be
> pretty much discounted as a reason altogether.
>
> Just trying now by (SMB) copying a large file I've achieved around 85%
> utilization between two XP workstations and a 2k server on the network
here
> and 77% between two 2k3 servers, all of which are under other load.

There is this known issue involving SMB.....

321169 - Slow SMB Performance When You Copy Files from Windows XP to a
Windows 2000 Domain
http://support.microsoft.com/?id=321169

........but I think this would be seen in a more extreme case then what he is
experiencing and is probably not his problem. I am still not totally
convinced that he even has a problem. He might also have an issue with the
quality of the cables if they aren't "factory terminated" or if there are
punchdown blocks involved that may be wired & punched too "sloppy".

--

Phillip Windell [MCP, MVP, CCNA]
www.wandtv.com