Worth upgrade to 100Mbs Switch ?

Our local school runs a 17 PC Peer-Peer Newtwork through a 10Mbs hub. All
PC's are low spec (P266mhz etc)
Our headmster has been told to upgrade the hub as this will increase network
speed .

1) Do you think its worth the money. (Yes even £100 is a lot from school
budget)
2) Would NIC in each PC need to be changed to 100Mbs (more ££'s)?

tia for any advice

I personally think the money could be spent on a new pc.

In article <3f1c570b$0$18487$(E-Mail Removed)>,
(E-Mail Removed) says...
> Our local school runs a 17 PC Peer-Peer Newtwork through a 10Mbs hub. All
> PC's are low spec (P266mhz etc)
> Our headmster has been told to upgrade the hub as this will increase network
> speed .
>
> 1) Do you think its worth the money. (Yes even £100 is a lot from school
> budget)
> 2) Would NIC in each PC need to be changed to 100Mbs (more ££'s)?
>
> tia for any advice
>
> I personally think the money could be spent on a new pc.
>
>
It depends very much on the usage patterns of the network, and what they
hope to achieve by upgrading the hub. If there's a lot of traffic
evenly spread across the network then a switch will help, but if a lot
of the traffic is going to a file or application server, web proxy or
whatever then the bottleneck will still exist. Duplex operation will
help some, but not necessarily very much. Has anyone analysed the
network traffic before coming to this conclusion? Or inventoried the
existing network cards? It's unlikely that many, if any, are 100Mbps -
if the "busy" machines have flow-control 100Mbps NICs then an
autosensing store-and-forward switch with flow control should further
improve performance, but as I said it does depend on the pattern of use.

On Tue, 22 Jul 2003 00:36:26 +0100, Rob Morley <(E-Mail Removed)>
wrote:

>In article <3f1c570b$0$18487$(E-Mail Removed)>,
>(E-Mail Removed) says...
>> Our local school runs a 17 PC Peer-Peer Newtwork through a 10Mbs hub. All
>> PC's are low spec (P266mhz etc)
>> Our headmster has been told to upgrade the hub as this will increase network
>> speed .
>>
>> 1) Do you think its worth the money. (Yes even £100 is a lot from school
>> budget)
>> 2) Would NIC in each PC need to be changed to 100Mbs (more ££'s)?
....
>It depends very much on the usage patterns of the network, and what they
>hope to achieve by upgrading the hub. If there's a lot of traffic
>evenly spread across the network then a switch will help, but if a lot
>of the traffic is going to a file or application server, web proxy or
>whatever then the bottleneck will still exist. Duplex operation will
>help some, but not necessarily very much. Has anyone analysed the
>network traffic before coming to this conclusion? Or inventoried the
>existing network cards? It's unlikely that many, if any, are 100Mbps -
>if the "busy" machines have flow-control 100Mbps NICs then an
>autosensing store-and-forward switch with flow control should further
>improve performance, but as I said it does depend on the pattern of use.

I've just ugraded my home network (2 w*ws boxes, one freebsd) from 10
to 100Mb. Frankly, the improvement was disappointing; I regularly do
dumps across this network, and the speed has improved only by perhaps
50% -- I think because of slow w*ws file access, plus cpu limits for
compressing the data. On the other hand, I used to run a network of
diskless Sun w/stations some years ago actually swapping across a 10Mb
network -- it just about worked -- and the company's entire network of
perhaps 500 machines worked on a 10Mb lan, segmented by bridges to
localise traffic where possible; it worked, even though we were very
heavily into nfs.

Also, (and this may be a either a problem or a solution!), don't
forget that a switch will prevent any machine monitoring network
traffic between others on the network, which is allowed by a hub.

As already said, you need to check your requirements (now and future)
carefully before stretching any budget.

Btw, you can buy an 8-port switch for £25 inc vat, and network cards
for a fiver (no frills like WOL, but I assume that needn't be a
problem)

--
Please use the corrected version of the address below for replies.
Replies to the header address will be junked, as will mail from
various domains listed at www.scottsonline.org.uk
regards. Mike Scott Harlow Essex England.(unet -a-t- scottsonline.org.uk)

Mike Scott wrote:
> On Tue, 22 Jul 2003 00:36:26 +0100, Rob Morley <(E-Mail Removed)>
> wrote:
>
>
>>In article <3f1c570b$0$18487$(E-Mail Removed)>,
>>(E-Mail Removed) says...
>>
>>>Our local school runs a 17 PC Peer-Peer Newtwork through a 10Mbs hub. All
>>>PC's are low spec (P266mhz etc)
>>>Our headmster has been told to upgrade the hub as this will increase network
>>>speed .
>>>
>>>1) Do you think its worth the money. (Yes even £100 is a lot from school
>>>budget)
>>>2) Would NIC in each PC need to be changed to 100Mbs (more ££'s)?
>
> ...
>
No, no, no!!

My wife's school is in exactly the same situation.
You may have no choice when the LEA start making more and more stringent
support conditions.
Cynically, I think its similar to the total Microsoft monopoly being
imposed by DFES after they took a £2m cash bribe from Mr Gates himself -
remember him coming grinning out of Downing Street 3 years ago?

Think about the bottlenecks. Even if you've got (or will get) a 2Mb/s
broadband link, that's only 1/5 the speed of your existing 10Mb/s
ethernet. Round here they're proposing groups of 4-6 primary schools
sharing each 2Mb/s link. BT's ADSL suffers from over-sharing itself, so
you're unlikely to get your full 2Mb/s or anywhere near it, except at
02:00 am!

Even if your 17PCs are mostly using a few popular websites which reside
in your local server's cache, access to the disc to read the cache is
unlikely to exceed 50Mb/s within the fastest small servers and thats for
big files, not the thousands of little graphic buttons you need for
webpages.

100Mb/s switches do give an improvement, but for 17PCs on a small server
(I mean single 2GHz processor, single hard disc) it will be much more
like a 50% improvement, for internal work only, than a tenfold increase.
More like a 5% improvement in Internet speed.

I wonder how the LEAs really justify their pressure for this
expenditure, or have they been duped too?

There are only three computer viruses which are known to infect humans:
the first is called "Cynicism" the second "Altzheimers" and I can't
remember for the life of me what the third is called....

--
Have fun,
Mike
--
http://fonehelp.co.uk - PC support, no fix, no fee!

In article <3f1c570b$0$18487$(E-Mail Removed)>,
"Richard Ward" <(E-Mail Removed)> writes:
> Our local school runs a 17 PC Peer-Peer Newtwork through a 10Mbs hub. All
> PC's are low spec (P266mhz etc)
> Our headmster has been told to upgrade the hub as this will increase network
> speed .
>
> 1) Do you think its worth the money. (Yes even £100 is a lot from school
> budget)
> 2) Would NIC in each PC need to be changed to 100Mbs (more ££'s)?
>
> tia for any advice
>
> I personally think the money could be spent on a new pc.

What is the problem you are trying to solve?
It's not worth spending any money on an upgrade to the hub unless
a) you actually have some identified problem,
b) the cause of which is the hub,
c) nothing else would also need upgrading to solve it,
d) it's the most pressing problem which is solvable given the
available resources.

You need to satisfy these points first.

--
Andrew Gabriel