Page 1 of 1

RB153 or RB532?

Posted: Sat Nov 04, 2006 1:00 am
by danielillu
I'm going to install a backbone node with 3 SR5 (2 of them at about half or third of max power) and other one with 2 sr5+ sr2 (sr5 at less than max power) but my provider doesn't have RB532 stock, and I thought about using RB153 as they have 3 MiniPCI on board and I can get them in a couple of days, but...

how important are those hundred Mhz more in RB532?

One board (2sr5+sr2) is going to mangle + pcq Queue traffic, wds PtP over sr5 and distribute to my clients through sr2, all interfaces (wdss+sr2) bridged all together (I know is better to route them, not bridge, but I haven't designed that network).
the other one will have all sr5 at 5g turbo in wds PtP links with all wds bridged, too.
I usually don't use more than 20% CPU peak on existant RB532, but I don't have any using 3 radios.

My clients usually use (and abuse) aggresive p2p software most of time (i'm working on having them under control with queue and filter rules), and also there are VoIP lines because our voice provider think we don't generate enough money to them to install cables for us.

Thank you for your help.

PS: I'm not sure if this was the most correct place to post this topic. I'm sorry

Posted: Tue Nov 07, 2006 8:03 pm
by stealth
On ptp links I always use rb532 because speed is important. In fact I dont see any point why they developed rb153 using so slow cpu. It cant even drive effectivelly one atheros minipci, but they put 3 minipci slots and 5 ethernets.
Look at these test
There is test of two ptp links running on the table.
3mbps FD, they must be kidding. RB532 can do as fast as 16+mbps FD in the same configuration and test condition.
This board was developed by some PR manager. People screams for faster routerboard with intel XScale 533Mhz cpus, not so slow board.

Posted: Wed Nov 08, 2006 6:30 am
by ericsooter
I think it was intended for the 153 to be more of a drop in replacement for the Linksys type AP's in indoor environments. I've personally used them in this scenario when building out indoor AP's for businesses; and they work very well. As for outdoor QOS and high traffic use, they are probably not optimal.

Posted: Wed Nov 08, 2006 10:17 am
by cmit
There is test of two ptp links running on the table.
Apart from that I, too, don't see the real usage scenario for the RB153:
I looked onto the two test systems, and they do have a wireless RX level of around -20 dBm. I'm quite sure that this is overdriving the RX part of the CM10 in there.
I suspect you could get away a bit better when lowering that so that they are in the -50's or -60's range...

Best regards,
Christian Meis

Posted: Wed Nov 08, 2006 10:42 am
by stealth
This is website of one of czech mikrotik distributor. I am not really even think about buying rb153. CPUs runs on 100%, so overloading of rx amplifier should not be issue of low bandwidth in this particular test.

Posted: Wed Nov 08, 2006 11:05 pm
by danielillu
Thanks to everyone with your comments and that url.
I'll get rb532 only, then, I'll go for sure devices.

If its true (what I don't put in doubt) what you and that url says, I agree all of you about why they released that configration. as all advantatges of having 3 miniPCI and 5 Ethernet on board goes away when CPU can't move all that gear fluently.

Thank you for your great help.