I don't have a S-31DLC20D to test, but Cisco 3560/3750 switches can support 3rd party SFPs if you enter the following commands:

service unsupported-transceiver
no errdisable detect cause gbic-invalid

The first allows non-Cisco SFPs and the 2nd stops SFPs being put into the err-disable state.

With 700 devices in the same layer 2 network every single device will receive & have to process every single broadcast frame, so splitting them up via VLANs would reduce this broadcast overhead. Ideally you'd have 50-100 devices maximum in the same layer 2 network. However I doubt an RB1200 is fast ...

Your ASA configuration is pretty straightforward, nothing that the RB1100 can't do. However what speed connection is the ASA handling?

I've just checked the download site & SWOS 1.7 is now there, including for the RB250GS, however the change list is minimal: What's new in 1.7: *) updated web look *) fixed problem - setting link speed (without autoconfiguration) did not work (allways 10Mbits was used) Given that there's still a few ...

Layer 2 only - with VLANs, port mirroring & some simple access list capabilities.

I've been running RouterOS 5.23 & RouterBoot 3.02 for a few days on a RB1200 without any problems, however it's not doing a huge amount currently - just NAT, Firewall & L2TP/IPsec VPN.

Does this mean SwOS is going to be updated soon? It's been ages since 1.6 was released & there's a number of issues with it that have been flagged up in the forum, for example http://forum.mikrotik.com/viewtopic.php?f=17&t=49646

There's several threads where people have discovered the latency increase, but I've not seen an satisfactory explanation yet. This thread has links to a few more related discussions: http://forum.mikrotik.com/viewtopic.php?t=56779

The 10-15ms latency increase seems to be a byproduct of the RB1200s limited IPsec acceleration, some further insights here http://forum.mikrotik.com/viewtopic.php?t=53026 & here http://forum.mikrotik.com/viewtopic.php?f=2&t=56504

I've experienced similar problems with ether9 & 10, they're connected to the CPU differently (via PCI-e apparently) & this seems to hamper performance. Therefore I'd avoid using ether9 & 10 if possible.

I don't see a NAT exemption on the Cisco config either, normally in the ACL for the ip nat inside you add deny statements from the internal network to the far end internal network. Also if you put an ACL on the outside interface you'd need to allow UDP 500, 4500 & ESP via appropriate permit statemen...

Forgot to say that it's running RouterOS v5.19. I've rebuilt the configuration from scratch & experience the same issue, weirdly if I move the connection to the cable modem to ether10 queuing only uses ~1.5% CPU time for the same throughput.

I recently upgraded from an RB750G to an RB1200 & have ported the configuration across - it's very simple just 5 ports in a switch group, the port connected to the cable modem picks up an address via DHCP & some firewall rules. However the CPU utilisation is surprisingly high, for example when pushi...