Whats the diffference between a network Hub and a Switch ???

Maybe a stupid question, but I can't figure out the diff.

Thanks

Cris

Look here: http://duxcw.com/faq/network/hubsw.htm

bye

A hub, in its most basic form, simply repeats the incoming signal to all ports and does no processing.

A switch is a type of hub which reads the incoming packets and determined the appropriate port to forward the packet to without bothering the other ports.

Cheers fellas,

I have been trying to figure out why my home LAN is soooo slow and why the conflict light it always on while transfering to empeg. May have to invest in a switch me thinks.

Cris.

Well, most home LAN's won't benefit from a switch. The only time you'll see an improvement really is if you have two compleetly seperate connections - one from computer A to B, and another from computer C to D. Most home networks don't have enough active nodes for this situation to often arise. Also be aware most hubs have activity led's for each port, and a general colision light. Having the activity light on while transfering is compleetly normal.

A problem I have seen come up is 100/10 hubs/switches have issues going from 100 mbit links to 10mbit links. I suspect this is just a low quality hub problem, or maybe be just my imagination.

The empeg only runs on 10mbit (bits, not bytes), so the speed isn't blinding.

Matthew

depending on what OS you're using, you could check the collision count on each device.

That's not entirely true. If you had data going from A to C and B to C, you might see some improvements, as most switches will also try to avoid transmitting twice at the same time on a single port. Then again, most consumer-oriented are probably not quite that intelligent. So nevermind.

I don't know that much about networking, but I can tell you that my home network has impoved VERY much in speed since I'm using a router. Before this setup I simply connected my pc's with a crossover cable. Now I know a router is also something different (don't really know the difference between a router and a switch), but I suspect the difference in speed is comparable.

Yes, I'll conceed on the A to C and B to C, especialy if traffic is slightly intermittent as the buffering would be able to help.

As to Archeon, that I can't explain at all. Anyone have any ideas as to why that would happen?

Matthew

A router is in a whole different neighborhood than a switch or hub. Switches an d hubs provide basically the same service in different ways in that they connect together computers on the same LAN. They're basically cabling extensions. But a router actually transmits data from one LAN to another. It receives data that's not destined for it specifically and determines if the data needs to be passed on and, if so, to which network. To confuse issues more, many, if not most, of these home DSL/Cable routers also have an integrated hub or switch on the home side.

But a router itself cannot provide greater speed internal to a LAN. It's really just another computer on the LAN. It can only provide faster or more access to networks beyond the LAN.

So not knowing exactly how your network was set up before or now or knowing if your router has a hub built into it, it's hard to say why it might work better. I doubt that most of the router-built-in hubs are switching hubs, so it's unlikely to be something that a switch would improve. Maybe your network went from 10Mbps to 100Mbps. It's easily conceivable that your NICs didn't autonegotiate to 100Mbps when just using the crossover cable.

A router is in a whole different neighborhood than a switch or hub.

Right, but if they were referring to one of the new product classes of SOHO internet/DSL routers, it probably came with a 4-port 10/100 switch integrated into it.

I know I'm confusing the description even more by saying it, but I'm actually trying to simplify...

A Router and a Switch and a Hub are three different things, true, but there are times when you will find a product that is a Router and a Switch integrated together, or a Router and a Hub integrated together. If you have a Router with four extra ethernet plugs on it, you're probably looking at one of these combination devices.

It could also be that the xover cable was incorrectly built, i.e. out of polarity. This would WORK, but there would be problems.

Which is why I said ``To confuse issues more, many, if not most, of these home DSL/Cable routers also have an integrated hub or switch on the home side.''

I always use this story to illustrate just how dire the standard of lecturing is at my uni.

A lecturer asked that question to a class I was in last year. I answered with the usual reply about a hub broadcasting and a switch directing to specific ports.

His answer was 'er, no' with a smug grin.

Not quite sure what to say to that, I just looked at him, puzzled.

He went on the explain that switches were solely for connecting 10Mbit and 100Mbit networks together. 'Nothing more, nothing less' were his words.

In a later lecture, the same guy, talking about email, told us that POP3 stands for Point Of Presence 3. 'It's got nothing to do with post offices' he said. By that point everyone in the class had learned not to bother trying to correct him.

Interestingly, that just so happens to be what the switches do here where I work: Connect multiple 10-megabit hubs together.

But they also connect directly to 100-megabit cards in the servers, and a few 100-megabit-capapble workstations plug straight into the switches. I mean, the teacher was wrong, but he did describe one valid use for a switch.

'Nothing more, nothing less' were his words.

Scary. This reinforces my feeling that you should always strive to take morning classes -- before the instructors get into their cups!

So, you *did* write this up on the course eval, I hope?

My router is the US Robotics 8000 home router. And yes, it's got a 4-port hub built-in.

When the computers were connected with a crossover cable, they used the same NIC's (3Com Fast etherlink 10/100BAse-TX cards) as they are now. I also tried 3 different networking cables. The speed stayed the same. The network did work at 100Mbps, at least that's what I told win2000 to do. (I didn't leave the speed setting to "auto", I actually told win2000 to use 100Mbps/full duplex - didn't do much good for the speed though).
But ever since the router is on my network the speed has improved dramatically. Probably even doubled or more. Why this is so I can't explain.

Ah, that's why, then. The 4-port thing in the US Robotics 8000 is 100-megabit switched.

(According to one review of the USR8000 I found on the web. The official USR site is rather unclear on the point.)

I can understand why he didn't realise what switches actually were - he is an IT management consultant, not a geek. Like most non-tenchical people, he knows what things are used for, now how they work. It's just unfortunate that he is only teaching half the story.

Luckily I do have plenty of other very good lecturers.

There are also multi-speed hubs that can do the same thing without switching. I don't really know how, but there are some.

Regardless, the term for his definition is a bridge. Any device that connects two dissimilar media into one network is a bridge.

Any device that connects two dissimilar media into one network is a bridge.

With this constrained definition, it would probably meet the definition of a translational bridge. I don't think that there's any standard associated with the definition of translational bridging, so while a Token Ring-to-Ethernet bridge would clearly fall under that definition, not sure that a 10Mbps-to-100Mbps speed-sensing Ethernet bridge would.

All I mean to say is that a device does not have to connect dissimilar media to meet the definition of a bridge. There have been plenty of single-speed, single media, same-frame-format bridges. Switches are just evolved intelligent multi-port bridges, making definitions progressively more muddy.

All I mean to say is that a device does not have to connect dissimilar media to meet the definition of a bridge.

To argue semantics (), just because there's a more precise definition does not mean that the less precise definition is incorrect. For example, if I said that any fruit that produces orange juice is a citrus fruit, I would be correct, despite the fact that it would be more precise to say that it was an orange.

Which is not to say that you're incorrect or anything.

just because there's a more precise definition does not mean that the less precise definition is incorrect.

No disagreement. I just got nervous when you said "the term for his definition is a bridge" (I should have included that), with the possible implication that it was exclusive.

So if you graft grapefruit braches onto an orange tree, does it become translational citrus?

So if you graft grapefruit braches onto an orange tree, does it become translational citrus?

Something like that.

In reply to:

---

Any device that connects two dissimilar media into one network is a bridge.

---

This is why i've always been annoyed by DSL companies and their insistance that they give you a DSL Modem, not a DSL bridge. Every time I talked to tech support I'd complain about my DSL bridge and then go back and say modem. Finaly I've convinced myself to use their wrong terminology just to simplify things.

Now, the strange thing is that a month ago I was sitting in my EE class and the professor started describing bridged ethernet exactly as I'd always thought of switched ethernet. The text confirmed that he wasn't just making it up as he went along. The basic reason that I came up with for this is that no traffic goes over a bridge that doesn't need to, just like a switch.

Matthew

Unfortunately, you can also have bridges that are not so intelligent and just forward everything along. Like the default bridging setting of Ascend Maxes, which, when I initially dealt with them, tried to send all of the traffic on a 100Mbps ethernet segment down a 64kbps ISDN line. That was fun.

Something like that.

HA! We can always count on you to reach into that hat!

In reply to:

---

I always use this story to illustrate just how dire the standard of lecturing is at my uni.

---

Our HCI lecturer last term, describing how poor most UIs were, said that the car handbrake was a typical example of poor UI, because 'people pull them up more to stop on hills. If it will hold on the flat with 2 clicks, you only need to put 2 clicks on it, no matter how steep the hill'.

This was the guy who also said, later on, that the brake was the left hand pedal and the clutch the middle pedal.

A.

so while a Token Ring-to-Ethernet bridge would clearly fall under that definition,

Ummm... how did Hobbits get into this discussion?

Ohhh.. that's Token Ring, not Tolkien Ring. My mistake.

tanstaafl.

Ohhh.. that's Token Ring, not Tolkien Ring. My mistake.

Tolkien Ring Network

Ohhh.. that's Token Ring, not Tolkien Ring. My mistake.

Heh. Well, hobbits are nowhere near as deterministic as Token Ring, but they'll be with us a lot longer!

multi-speed hubs

As I understand it, these are simply switches that switch between the two speeds -- imagine switches, but with only two ports on them -- the 10Mbps port and the 100Mbps port. These are then "hubbed out" to the other ports.

I could be wrong, though.

So if you graft...

Actually, most orange trees are grafted onto lemon root stocks, anyway...

More interestingly: Apparently, citrus is citrus is citrus. If you take an orange and plant it, you could end up with a lemon tree. If you take the lemon and plant it, you could end up with a grapefruit tree, and so on...

Check out "Oranges", John McPhee, ISBN 0141182032

If you take an orange and plant it, you could end up with a lemon tree.

I've never heard of this. What's your source? It sounds farfetched to me, considering that for this to happen, the plants in question would have to spontaneously change species at some point during their life cycle.

It sounded farfetched to me, too, when I first heard it, which is why I bought that book. It doesn't cover it in too much detail, but I did find these two references:

Preface, page ix:

"Citrus does not come true from seed. If you plant an orange seed, a grapefruit might spring up. If you plant a seed of that grapefruit, you might get a bitter lemon."

Page 35:

"Up from those lime seeds came sweet orange trees, tangerines, limequats, citrons..."

OK, a little work with google once I got the term-of-art, "come true from seed", seems to reveal that not all species of citrus can do this, but that some of them can.

I actually have one of those multi-speed hubs and it has no documentation warning about only being able to plug certain speeds into certain places. In fact, I can move my empeg (the only 10Mbps device I have on that network) to any port and not have a problem. Now it's still possible that there's something like what you suggest going on behind the scenes, but, if so, it's totally transparent to the user.

In reply to:

---

Now it's still possible that there's something like what you suggest going on behind the scenes, but, if so, it's totally transparent to the user.

---

Exactly. Dual-speed hubs can tell when a connection is 100Mbit - there's usually a light next to the port.

Obviously. But that doesn't mean there's any switching going on, which is the part of Roger's statement I was refuting. And the fact that it can detect whether it's 10Mbps or 100Mbps would seem to me to be 90% of the way towards not needing a switch anyway (assuming switching is not a priority to the product).

Citrus does not come true from seed.

(avoids temptation to mark quotation with [lime] tags)

The wonderful Oxford Companion To Food goes on about this, too. Basically, in many highly-bred plants, the borderlines between species are not as clear-cut as "this is a giraffe, but that's a hamster". If you think about (say) limeness as a recessive gene that's carried by many oranges, then there will be a certain number of lime trees every time two oranges pollinate.

The same, IIRC, happens with brassicas, where a huge variety of cultivars (cabbages, kale, brussels sprouts, oilseed rape, cauliflowers, broccoli/calabrese) have been bred from the relatively small genetic raw material of the seakale. "Although consumers can easily distinguish broccoli and cauliflowers," says OCF, "botanists have great difficulty".

Peter

PS. I read somewhere that all the world's domesticated hamsters are descended from one pregnant female caught one day in Egypt. Anyone know whether that's true?

Yes, it is totally transparent, the ports are auto-sensing, but there's still a (virtual) two-port switch in there.

In particular, from Netgear's website (talking about the DS108):

Each unit has a built-in self-learning bridge which provides the communications link between the 10 and 100 Mbps network devices. The intelligent bridge automatically manages network traffic such that 100 Mbps traffic does not unnecessarily crowd the 10 Mbps network segment and 10 Mbps traffic does not crowd the 100 Mbps segment.

Personally, I'd get a switch (I have a pair of Netgear FS108 switches at home). They're not much more expensive, and performance is better.

Well, there you go. I'd generally get the switch, too. (I got my dual-speed hub for free.)

domesticated hamsters

Actually, Syria:

http://www.geocities.com/Heartland/Hills/1327/guide1.html
http://www.doghause.com/funfacts05.html