Marek Isalski <ma...@faelix.net> writes:

> While there's a certain crowd who will look upon the cheap Latvian
> entry into the routing market with upturned noses, if you're prepared
> to deal with the few caveats (and smaller ecosystem for support),
> their price/performance can't really be beaten.  For every bug I've
> encountered with the MikroTiks over the last year, I've encountered
> just as many with any other vendor deployment - be it at a client, a
> customer, or a supplier.  And rather than worry about "advanced
> replacement" for parts, the entire router is affordable enough for you
> to have spares in stock.

We (hubs.net.uk) make extensive use of Mikrotik kit, even running BGP in
very non-trivial configurations with our member
networks. Price/(performance+features) is indeed difficult to beat.

But I would strongly advise against putting full tables into them,
however. We did some proper lab testing of this, and encountered some
serious bugs - confirmed with the vendor, and will not be fixed until
the next major release. These are things like recalculating the *entire*
FIB for every update (ever wonder why the BGP process keeps one CPU
pegged on the CCR? That's why.)

This kind of bug means it will appear to work, until there is
instability or churn around you, and your border routers melt. If I ran
a commercial network with a well-defined notion of "competitor", I would
encourage my competitors to adopt such a design. As it is, where we take
full tables it is BIRD on FreeBSD which is rock solid.

Cheers,
-w

------------------------------------------------+------------------------
William Waites        <wwai...@tardis.ed.ac.uk> : School of Informatics
Synthsys Centre for Mammalian Synthetic Biology : University of Edinburgh
------------------------------------------------+------------------------
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

Reply via email to