Wired ethernet gigabit switch: Mikrotik or typical unmanaged switch

[FHTheron]

Does anyone have any performance information comparing small packet speeds through a mikrotik switch and a standard/typical retail switch?

Specifically looking at this:
http://routerboard.com/RB750GL
They claim 127.3mbps for small packets in fast path mode.

Comparing that too a standard switch like a D-Link DGS-1008D:
http://www.dlink.com/uk/en/business...s-1008d-8-port-10-100-1000mbps-gigabit-switch
The above (and many others) come with a standard specification of 1,488,000 packets per second which is (pps * (minimum_ethernet_frame + typical preamble and interpacket gap) * 8) == 1488000*(64+20)*8 == 999.936e6 which is probably where they get the number from in the first place.

I suppose there's very little reason why a typical unmanaged switch should not perform at the claimed packet rate, but still can't help but wonder why they range from R200 to R700 in price.

 

[Rickster]

Does anyone have any performance information comparing small packet speeds through a mikrotik switch and a standard/typical retail switch?

Specifically looking at this:
http://routerboard.com/RB750GL
They claim 127.3mbps for small packets in fast path mode.

Comparing that too a standard switch like a D-Link DGS-1008D:
http://www.dlink.com/uk/en/business...s-1008d-8-port-10-100-1000mbps-gigabit-switch
The above (and many others) come with a standard specification of 1,488,000 packets per second which is (pps * (minimum_ethernet_frame + typical preamble and interpacket gap) * 8) == 1488000*(64+20)*8 == 999.936e6 which is probably where they get the number from in the first place.

I suppose there's very little reason why a typical unmanaged switch should not perform at the claimed packet rate, but still can't help but wonder why they range from R200 to R700 in price.

Is that MB/s or mb/s?

 

[FHTheron]

Going by that table of theirs, it's Mbit/s as in 127.3e6 bits/s

Usually go with http://physics.nist.gov/cuu/Units/binary.html these days, but sometimes forget about the upper/lower case, which is probably why they recommend "bit" instead of just "b".

 

[Rickster]

Gigabit Ethernet is 1000 Mbit/s not a measly 127Mbit/s.

If you are looking for a good Gigabit switch, I would recommend this one, its quite cheap too.

The key difference between a managed and unmanaged switch is the
ability to configure the switch and to prioritize LAN traffic to ensure that the most important
information, as I have defined it, gets through.

Managed switches give you more control over your LAN traffic and offer
advanced features to control that traffic.
An unmanaged switch simply allows Ethernet devices to communicate with one another, such as a
PC or network printer, and those are typically what we call “plug and play.” They are shipped with a
fixed configuration and do not allow any changes to this configuration.

 

[Tinuva]

Does anyone have any performance information comparing small packet speeds through a mikrotik switch and a standard/typical retail switch?

Specifically looking at this:
http://routerboard.com/RB750GL
They claim 127.3mbps for small packets in fast path mode.

Comparing that too a standard switch like a D-Link DGS-1008D:
http://www.dlink.com/uk/en/business...s-1008d-8-port-10-100-1000mbps-gigabit-switch
The above (and many others) come with a standard specification of 1,488,000 packets per second which is (pps * (minimum_ethernet_frame + typical preamble and interpacket gap) * 8) == 1488000*(64+20)*8 == 999.936e6 which is probably where they get the number from in the first place.

I suppose there's very little reason why a typical unmanaged switch should not perform at the claimed packet rate, but still can't help but wonder why they range from R200 to R700 in price.

First of all, who looks at the speed you can achieve with 64byte packets? Honestly, if you want throughput, look at the 1518byte packets, which is what your network will be using when you transfer large files. With those packets, you get close to 1000Mbit/sec.
Small packets are good for latency intensive apps, likes games, which is when it will be used. No game, or VoIP app ever use more than 100Mbit/sec in a home environment where these devices will be in use.
So you are looking at the wrong information for your uses cases in a home or even small office environment.

Second, the price difference come in by the fact that the Mikrotik runs a complete router-os, far more advance than a basic unmanaged switch which does nothing more than that. Mikrotik can do BGP, OSPF for routing, VPNs which include pptp/ipsec/l2tp. It can show traffic flowing through the layer3 interfaces, very cool for debugging. It has a web proxy, dns proxy, can do pppoe dialup, multiple of those, with traffic load balancing and all kinds of advance things. It is like a mini computer on top of the switch.

I love my Mikrotik, it is a RB751G. Have my adsl router in bridge mode, do the pppoe dialup from my mikrotik, including vpns to work. Route only work traffic over the vpns, and the rest of my traffic over the normal adsl connection. Makes life far simpler when I get a call from work and need to check up on something. Saves lot of time. I also do traffic shaping so that my downloads can flow 24/7 without affecting game play, streaming, browsing, voip ect. It literally is a complete routing management device for me at home. It is also a small switch at the location it is in the house, which is at the TV, for the ADSL, HTPC and then a link to the other side of the house where I have a proper 24 port unmanaged gig switch with more pcs and the file server ect.

Hope that helps you understand the price difference, as well as the mistake you made with the speed performance for the switching chip.

 

[FHTheron]

First of all, who looks at the speed you can achieve with 64byte packets? Honestly, if you want throughput, look at the 1518byte packets, which is what your network will be using when you transfer large files. With those packets, you get close to 1000Mbit/sec.
Small packets are good for latency intensive apps, likes games, which is when it will be used. No game, or VoIP app ever use more than 100Mbit/sec in a home environment where these devices will be in use.
So you are looking at the wrong information for your uses cases in a home or even small office environment.

That's the perspective reminder I needed, thanks. I'd still be interested in a real world comparison though, just in case it matters one day...

For the price difference I was not considering mikrotik as part of that range, I realise they are far more complex and powerful than a simple switch and it then follows that they would be more expensive. I was wondering more about this sort of difference:
Tenda SG80 and Asus GX-D1081 V3. The Asus switch is almost double the price. The Asus has a larger MAC Address Table and faster switch fabric, but as you pointed out, in typical use cases this should not really matter all that much if the basics get the job done.

I suppose another way of asking it is who is the target consumer for the Asus GX-D1081 V3? Apart from the fact that it looks nicer and has possible brand loyalty, the larger buffer and faster switch fabric got me wondering what is is meant to be used for. I imagine having constant high throughput between several ports (large packets) might start to show a difference, but still not enough to really matter.

I'm thinking of replacing my ADSL router with exactly that mikrotik setup (modem in bridged mode and lots of fun on RouterOS) later, for now I just need to get one room on a wire and for that I think I'll start with the cheapest 5-port switch I can find, which would be this odd shaped switch that will probably refuse to sit straight at the edge of a table.

 

[Asha'man X]

Here's a funny thing: depending on how you configure a managed switch, there are times when an unmanaged switch can be a little faster. The reason for this is that the unmanaged switch works on layer 2. Basically, it's a case of get packet, amplify packet, send packet out correct port that will match the MAC address in the packet.

Managed switches usually operate at layer 3 and above, which can add overhead when you start adding things like VLAN's, Quality of Service, Routing and so on. It's unlikely you'll notice any difference unless the switch is heavily loaded, or you have tons of features enabled that overwhelms the processor in the switch.

 

[Tinuva]

Here's a funny thing: depending on how you configure a managed switch, there are times when an unmanaged switch can be a little faster. The reason for this is that the unmanaged switch works on layer 2. Basically, it's a case of get packet, amplify packet, send packet out correct port that will match the MAC address in the packet.

Managed switches usually operate at layer 3 and above, which can add overhead when you start adding things like VLAN's, Quality of Service, Routing and so on. It's unlikely you'll notice any difference unless the switch is heavily loaded, or you have tons of features enabled that overwhelms the processor in the switch.

Actually, you will find that the layer2 and layer3 functions run on different chips, especially on Mikrotiks.
Layer3 may get slowed down, but layer2 will stay fast.

Vlans operate at layer2 btw, and doesnt slow down switching. Just keep in mind, that if you route between 2 vlans, the packets go via layer3 routing which may seem slow. That said, why would you need vlans at home?

Switching is always fast, routing can get slow.