What are the advantages and disadvantages of using a managed switch over
an unmanaged switch and vice versa?
I'm trying to build a very robust network for home / small business use.
I'm not worried about managing a switch, just interested in
understanding why one would want a managed switch.
"Chip Thomas" <spamanot@gmail.net> wrote in message
news:m4bDj.4321$6H.1492@newssvr22.news.prodigy.net ...
>
> What are the advantages and disadvantages of using a managed switch over
> an unmanaged switch and vice versa?
>
> I'm trying to build a very robust network for home / small business use.
> I'm not worried about managing a switch, just interested in understanding
> why one would want a managed switch.
In my view, not saying this is 100% correct, managed switches are for large
enterprise networks which have a service level agreement in place between
the user community and the IT Department in case something goes down it
needs to be fixed within XX amount of time. Managed switches allow you to
remote into them and administrate them to a more robust degree than you get
with non-managed, which is why they are more expensive.
>What are the advantages and disadvantages of using a managed switch over
>an unmanaged switch and vice versa?
>I'm trying to build a very robust network for home / small business use.
>I'm not worried about managing a switch, just interested in
>understanding why one would want a managed switch.
In short, a managed switch allows to you control to a certain extent what
each port or group of ports is doing. You can turn ports on or off and do
things like seperate groups of ports into vlans so that you can have 3 or 4
seperate networks on a single switch (e.g. 3x 8 ports for a 24 port switch).
An umanaged switch just passes data between the appropriate ports without a
second thought - which in 99% of cases is what you want.
Usually managed switches are for large enterprises such as big businesses,
schools, ISPs, etc that need a great deal of control over their environment.
Skeleton Man wrote:
>> What are the advantages and disadvantages of using a managed switch over
>> an unmanaged switch and vice versa?
>
>> I'm trying to build a very robust network for home / small business use.
>> I'm not worried about managing a switch, just interested in
>> understanding why one would want a managed switch.
>
> In short, a managed switch allows to you control to a certain extent what
> each port or group of ports is doing. You can turn ports on or off and do
> things like seperate groups of ports into vlans so that you can have 3 or 4
> seperate networks on a single switch (e.g. 3x 8 ports for a 24 port switch).
>
> An umanaged switch just passes data between the appropriate ports without a
> second thought - which in 99% of cases is what you want.
>
> Usually managed switches are for large enterprises such as big businesses,
> schools, ISPs, etc that need a great deal of control over their environment.
>
> Chris
>
>
I don't need network segregation.
I do need high reliability / continuous availability. I don't need much
level of control.
One port into a router, the rest into end-user computers, servers,
wireless access points and other switches.
It seems that all I need is a robust unmanaged switch.
On Mon, 17 Mar 2008 04:43:00 GMT, Chip Thomas
<spamanot@gmail.net> wrote:
>
>I don't need network segregation.
>
>I do need high reliability / continuous availability. I don't need much
>level of control.
Well... you either need a managed switch to have ANY level
of control, or with an unmanaged switch you have NO control
over the switch.
>
>One port into a router, the rest into end-user computers, servers,
>wireless access points and other switches.
>
>It seems that all I need is a robust unmanaged switch.
What is robust supposed to mean? It's not like there's some
golden switch out there, any random switch you buy should
work or is defective and should be returned for refund (but
defect is rather rare unless considering a poorly designed
passive enclosure more prone to overheating if installed
where there is no ambient air movement and high room
temperature).
kony wrote:
> On Mon, 17 Mar 2008 04:43:00 GMT, Chip Thomas
> <spamanot@gmail.net> wrote:
>
>
>> I don't need network segregation.
>>
>> I do need high reliability / continuous availability. I don't need much
>> level of control.
>
> Well... you either need a managed switch to have ANY level
> of control, or with an unmanaged switch you have NO control
> over the switch.
>
>
>> One port into a router, the rest into end-user computers, servers,
>> wireless access points and other switches.
>>
>> It seems that all I need is a robust unmanaged switch.
>
> What is robust supposed to mean? It's not like there's some
> golden switch out there, any random switch you buy should
> work or is defective and should be returned for refund (but
> defect is rather rare unless considering a poorly designed
> passive enclosure more prone to overheating if installed
> where there is no ambient air movement and high room
> temperature).
I consider "robust" as meaning that it provides continuous service
without interruption or downtime.
Many home routers "work" but need periodic reset. They work, but I
would not want them in my system because they do not provide reliable
service. I do not consider them very robust from an operational
perspective. But the manufacturer will not accept them as being defective.
A gigabit switch that only delivers reduce throughput or needs to be
reset periodically because a port hangs is operational but hardly to be
considered robust. If its poorly designed but it operates even in a
reduced-effectiveness manner, the manufacturer won't cover it under
warranty. I'll never use an unmanaged 3Com switch again for that reason.
These instances of problems occur much more frequently than the
manufacturers would seem to admit. I can't tolerate this kind of junk
in my network.
On Mon, 17 Mar 2008 15:05:57 GMT, Chip Thomas
<spamanot@gmail.net> wrote:
>> What is robust supposed to mean? It's not like there's some
>> golden switch out there, any random switch you buy should
>> work or is defective and should be returned for refund (but
>> defect is rather rare unless considering a poorly designed
>> passive enclosure more prone to overheating if installed
>> where there is no ambient air movement and high room
>> temperature).
>
>I consider "robust" as meaning that it provides continuous service
>without interruption or downtime.
That's just about every switch made, unless as mentioned
previously it has a poor enclosure plus poor placement plus
high ambient temp.
>
>Many home routers "work" but need periodic reset. They work, but I
>would not want them in my system because they do not provide reliable
>service. I do not consider them very robust from an operational
>perspective. But the manufacturer will not accept them as being defective.
Agreed, routers are more troublesome, sometimes needing
firmware fixes and sometimes needing more thought and action
than just where they're located. These days I almost always
pop open routers, arctic epoxy a heatsink onto the network
processing chip, and drill a few addt'l vent holes into the
cover. So far that seems to work fairly well. Anyway,
routers are more complex than switches and a lot more
problematic. With switches that are unmanaged you have
reduced the logical error possiblity due to simplier
firmware and lower processing requirement as well, though
typically being higher priced switches they may have better
engineering for cooling (including a fan upon occasion) in a
hotter environment like a rack, more applicable if you'd
actually place one in a hot rack.
>
>A gigabit switch that only delivers reduce throughput or needs to be
>reset periodically because a port hangs is operational but hardly to be
>considered robust.
Ok, do you know of any that need this? IMO, that is very
uncommon even among the cheapest consumer switches.
>If its poorly designed but it operates even in a
>reduced-effectiveness manner, the manufacturer won't cover it under
>warranty. I'll never use an unmanaged 3Com switch again for that reason.
Many people use 3Com switches without issue, you might need
to look at how your use or environment differs if you need
something in particular to counter that. IOW, many people
would consider 3Com switches sufficiently robust, that they
work w/o issue.
>
>These instances of problems occur much more frequently than the
>manufacturers would seem to admit. I can't tolerate this kind of junk
>in my network.
Buy something from a seller with a good return policy and
put it through it's paces right away, so you have
opportunity to return for a full refund if it doesn't meet
your needs.
kony wrote:
> On Mon, 17 Mar 2008 15:05:57 GMT, Chip Thomas
> <spamanot@gmail.net> wrote:
>
>
>>> What is robust supposed to mean? It's not like there's some
>>> golden switch out there, any random switch you buy should
>>> work or is defective and should be returned for refund (but
>>> defect is rather rare unless considering a poorly designed
>>> passive enclosure more prone to overheating if installed
>>> where there is no ambient air movement and high room
>>> temperature).
>> I consider "robust" as meaning that it provides continuous service
>> without interruption or downtime.
>
> That's just about every switch made, unless as mentioned
> previously it has a poor enclosure plus poor placement plus
> high ambient temp.
>
Hardly the case nor is it my experience. Low quality electrolytic caps
can leak and fail over time. Though higher temperatures may accelerate
the process, the resulting failure is certain. Then there are the cheap
voltage regulators that can't tolerate much voltage variation or
overheat under high current draw. Do that too often and they go into
permanent retirement. (For an example of these failures, check the
reviews of the Netgear GS108)
>
>
>> Many home routers "work" but need periodic reset. They work, but I
>> would not want them in my system because they do not provide reliable
>> service. I do not consider them very robust from an operational
>> perspective. But the manufacturer will not accept them as being defective.
>
> Agreed, routers are more troublesome, sometimes needing
> firmware fixes and sometimes needing more thought and action
> than just where they're located. These days I almost always
> pop open routers, arctic epoxy a heatsink onto the network
> processing chip, and drill a few addt'l vent holes into the
> cover. So far that seems to work fairly well. Anyway,
> routers are more complex than switches and a lot more
> problematic. With switches that are unmanaged you have
> reduced the logical error possiblity due to simplier
> firmware and lower processing requirement as well, though
> typically being higher priced switches they may have better
> engineering for cooling (including a fan upon occasion) in a
> hotter environment like a rack, more applicable if you'd
> actually place one in a hot rack.
>
With routers, its hard to overcome the [often time] crap firmware.
These cheap devices fail enough to make it worth investing the few extra
bucks to avoid interruptions in service and the associated hassles and
reboots that may be necessary to recover. Heat may have little to do
with a router that can't handle the load.
All these mods you do.... Is it really worth the time or effort? Seems
sorta like putting $1200 worth of tires on a used Yugo. I found it
easier to use better quality equipment that costs a few dollars and is
better quality. Your solution voids warranties and makes the devices
only marginally serviceable. And when those cheap caps or regulators
ultimately fail (which they will - just a matter of time), you're back
to square one without a warranty.
But even your efforts do not address problems with low-quality board
components. And when those cheap caps or regulators ultimately fail
(which they will - just a matter of time), you're back to square one
without a warranty.
Its easier and often cheaper to find a used Cisco router on eBay. Set
it up and forget it. Also the 3Com OfficeConnect Cable/DSL Router seems
to be pretty good. The OfficeConnect gigabit switches don't seem to
measure up to the router or their specs.
I run all of my equipment in a well-ventilated environment with a
ceiling fan moving the air around. No racks. No equipment cabinets.
Ambient temps usually around 65F or less in winter, 76F during summer.
I think you've started to address my question regarding managed vs.
unmanaged. If a managed switch will have fewer failures (because of the
extra engineering and more bullet-proof firmware), then that extra
reliability becomes part of the business decision on a switch purchase.
If an unmanaged switch of similar construction using similar firmware
(without the management options) and has roughly the same uptime
reliability for 60% of the price of a managed switch, the purchase
decision looks pretty clear.
>
>> A gigabit switch that only delivers reduce throughput or needs to be
>> reset periodically because a port hangs is operational but hardly to be
>> considered robust.
>
> Ok, do you know of any that need this? IMO, that is very
> uncommon even among the cheapest consumer switches.
3Com OfficeConnect gigabit. Doesn't matter if its a 5-port or 8-port
switch. A gigabit switch that seems to run at the MOST around 250mbs.
>
>> If its poorly designed but it operates even in a
>> reduced-effectiveness manner, the manufacturer won't cover it under
>> warranty. I'll never use an unmanaged 3Com switch again for that reason.
>
> Many people use 3Com switches without issue, you might need
> to look at how your use or environment differs if you need
> something in particular to counter that. IOW, many people
> would consider 3Com switches sufficiently robust, that they
> work w/o issue.
>
I though3Com was supposed to be pretty good also. I looked hard.
But 3Com performance didn't match the specs. After about three months
of reduced we started swapping out components until I identified the
3Com switches as the bottleneck. When the 3Coms were swapped out for
Netlink, D-Link and / or Cisco gigabit switches, the throughput jumped
dramatically. This happened at each point where the switches were
replaced.
If you check the more recent reports about 3Com OfficeConnect gigabit
switch performance, my observations are pretty consistent with others.
>
>> These instances of problems occur much more frequently than the
>> manufacturers would seem to admit. I can't tolerate this kind of junk
>> in my network.
>
> Buy something from a seller with a good return policy and
> put it through it's paces right away, so you have
> opportunity to return for a full refund if it doesn't meet
> your needs.
"Chip Thomas" <spamanot@gmail.net> wrote in message
news:lMSDj.4522$6H.4067@newssvr22.news.prodigy.net ...
> kony wrote:
>> On Mon, 17 Mar 2008 15:05:57 GMT, Chip Thomas
>> <spamanot@gmail.net> wrote:
>>
>>
>>>> What is robust supposed to mean? It's not like there's some
>>>> golden switch out there, any random switch you buy should
>>>> work or is defective and should be returned for refund (but
>>>> defect is rather rare unless considering a poorly designed
>>>> passive enclosure more prone to overheating if installed
>>>> where there is no ambient air movement and high room
>>>> temperature).
>>> I consider "robust" as meaning that it provides continuous service
>>> without interruption or downtime.
>>
On Tue, 18 Mar 2008 12:16:04 -0500, Chip Thomas
<spamanot@gmail.net> wrote:
>kony wrote:
>> On Mon, 17 Mar 2008 15:05:57 GMT, Chip Thomas
>> <spamanot@gmail.net> wrote:
>>
>>
>>>> What is robust supposed to mean? It's not like there's some
>>>> golden switch out there, any random switch you buy should
>>>> work or is defective and should be returned for refund (but
>>>> defect is rather rare unless considering a poorly designed
>>>> passive enclosure more prone to overheating if installed
>>>> where there is no ambient air movement and high room
>>>> temperature).
>>> I consider "robust" as meaning that it provides continuous service
>>> without interruption or downtime.
>>
>> That's just about every switch made, unless as mentioned
>> previously it has a poor enclosure plus poor placement plus
>> high ambient temp.
>>
>
>Hardly the case nor is it my experience. Low quality electrolytic caps
>can leak and fail over time.
True, but it just means you'd replace the equipment
periodically and have spares on hand. Plus, the caps that
leak and fail might've been defective in gear a few years
ago but today the more common culprit is they're running so
hot from the issues mentioned previously, casings poorly
engineered for cooling.
>Though higher temperatures may accelerate
>the process, the resulting failure is certain.
I would say not just accelerate with modern equipment but
rather the sole cause within the first 6-8 years or so,
after which the 'lytics may have increased their dielectric
leakage enough to be a secondary source of self-heating.
>Then there are the cheap
>voltage regulators that can't tolerate much voltage variation or
>overheat under high current draw.
The whole site should not be subject to voltage variations
large enough to interfere with any but the worse of cheap
switches' PSUs. That kind of problem would tend to
interfere with more than just the switches.
I don't feel they would overheat too much either unless the
ambient temps were unreasonably high.
>Do that too often and they go into
>permanent retirement. (For an example of these failures, check the
>reviews of the Netgear GS108)
I don't have any 108s, but I do have a GS105 and a couple
GS605 that've been working fine for, I can't remember
exactly, maybe 2 years average now with never a single reset
or power cycling necessary (I take that back, I recall the
GS605 needs one power cycling when a gigabit system is
initially plugged directly into it or else it defaults to
running in 100Mb mode). I did take their top covers off and
drill a bunch of holes for better ventilation, I just accept
that is necessary for longest life from any switch that
doesn't have a (typically noisey) fan in it.
However, these are fairly cheap switches, if the goal is
reliable and you want to buy cheap, buy a backup switch or
two.
The main thing I would stress with any passively cooled
switch is don't assume any amount of forethought in picking
a certain brand or model will be enough, actively check the
temps and drill a bunch of holes in the top to improve
cooling. Same goes for routers, maybe even moreso.
>With routers, its hard to overcome the [often time] crap firmware.
>These cheap devices fail enough to make it worth investing the few extra
>bucks to avoid interruptions in service and the associated hassles and
>reboots that may be necessary to recover. Heat may have little to do
>with a router that can't handle the load.
Routers are faster and better at handling load today than
they used to be. I've had P2P downloads of 'nix ISOs that
pegged the router at near 100% load over thousands of
connections and with a heatsink and ample casing holes they
didn't heat up - and this amount of load was only due to
custom firmware (DD-WRT) with QOS prioritizing other
traffic, otherwise the load was only 2 or 3 dozen %.
Manufacturers do learn from mistakes made in the past when
it comes to firmware, IMO it's less likely a source of
problems today.
>
>All these mods you do.... Is it really worth the time or effort? Seems
>sorta like putting $1200 worth of tires on a used Yugo.
Sure it's worth it. Takes about 5 minutes to take the lid
off a case, drill a few holes, wipe the debris away and
screw it back on. Result - trouble free networking for
nearly free (those Netgear switches mentioned above were
about $15 each after a rebate when purchased. If you can't
spend 5 minutes, even less if doing more than one at a time,
how much time have you already spent looking for some ideal
that you may never find without such simple and quick
modifications?
>I found it
>easier to use better quality equipment that costs a few dollars and is
>better quality.
.... and yet you are claiming so many problems - problems I
don't have! You should learn instead of rejecting what
works.
>Your solution voids warranties and makes the devices
>only marginally serviceable.
? You're trying hard to make excuses, how much do you need
a warranty on a low cost switch? After shipping and
considering network downtime, it's value is just about zero,
not worth bothering to seek and attain an RMA, spend time
and money to ship off somewhere. Certainly more of a PITA
than just putting some holes in the case.
What's "marginally servicable" supposed to mean? What kind
of servicing do you anticipate for a switch that works ok?
>And when those cheap caps or regulators
>ultimately fail (which they will - just a matter of time), you're back
>to square one without a warranty.
As mentioned above, avoiding problems then not caring about
a warranty is doubly nice.
>
>But even your efforts do not address problems with low-quality board
>components. And when those cheap caps or regulators ultimately fail
>(which they will - just a matter of time), you're back to square one
>without a warranty.
They address problems fine. Regulators don't just randomly
fail, eliminate the cause. Focus on heat like I wrote all
along instead of thinking you can ignore it and just buy
some ideal device that wouldn't be effected by heat.
If I'm back to square one, I'm ok with that since square one
was spending only a tiny bit of money on a device that
works (still). You're making things hard on yourself, if
there weren't a better way I wouldn't have written a thing
about it.
>
>Its easier and often cheaper to find a used Cisco router on eBay. Set
>it up and forget it. Also the 3Com OfficeConnect Cable/DSL Router seems
>to be pretty good. The OfficeConnect gigabit switches don't seem to
>measure up to the router or their specs.
.... and yet if it were true you wouldn't be here asking what
to get. Sorry but you are assuming based on limited
information and rejecting the thing known to change the
situation. Heat Management.
Do whatever you like, I'm quite happy with the results of my
suggestions.
Managed vs. unmanaged switches 
Linear Mode
