[Cerowrt-devel] 10GigE nics and SFP+ modules?

Dave Taht dave.taht at gmail.com
Tue Sep 9 12:54:56 EDT 2014

On Tue, Sep 9, 2014 at 9:26 AM,  <dpreed at reed.com> wrote:
> I agree with you if you have to install Cat6a and Cat7 structured wiring in
> a building! What a nightmare trying to find contractors who can meet spec on
> junction boxes, etc. and do the right testing.  Every connector on the path
> is problematic.
> But for "casual" home networking use or research lab experiments, copper is
> fine.  That was my point.  I think in a "rack-scale" datascenter networking
> situation. probably better to go with fiber, but copper costs are
> attractive.

I was absolutely delighted to see all the tutorials on this site, example:


Here I was thinking I'd have to go write another
http://www.rage.net/wireless/wireless-howto.html ...

I have been a big advocate of municipal fiber, but have not had a good
handle on the costs involved. I have generally heard it's 80% labor,
20% materials, and the cost of fiber cabling alone, at least at retail
is kind of intimidating.

I see a lot of tradeoffs being made with GPON that for example, single
mode fiber deployments such as amsterdam's didn't make.
- and kind of concerned about the "bumps in the wire" there.

and several costs there have dropped significantly, routerboard is
making a SFP+ capable
5 port switch for like 50 dollars, and things like this
router-you-can-drop-off-a-truck, developed in africa...


I'm not so much interested in 10GigE in the home, as 10GigE or 1GigE
on a small campus, like  what I have here, which is presently a dozen+
directional APs spread over runs of 126-380 meters, outdoors. If I
were to pursue FTTY (Fiber To The Yurt!), what would the costs and
tradeoffs be? What would a reasonably redundant network look like?
Could you go with a meshier topology? How would you repair things
after an earthquake?

It's an interesting set of theoretical questions. Unlike some, I'd not
have certain ISPs lobbying to stop me from trying, the telephone poles
are locally owned, ground burial is not a problem, either, and the
labor is cheap.

Admittedly, now that bug 442 is seemingly fixed, I can finally bring
the reliability of the wifi here back up to a saner level, and deploy
multiple exit nodes to comcast and elsewhere via hnetd, so upgrading
to fiber throughout the campus is purely speculative at the moment, a
nice distraction from thinking further on fixing wifi.

(my original intent remains: just fiddling with 10GigE on a small test
setup, to see packet patterns at high rates)

> On Tuesday, September 9, 2014 12:06am, "Joel Wirāmu Pauling"
> <joel at aenertia.net> said:
>> Note I know whilst you are talking copper. IME 10Gbit copper is
>> nothing but a hassle. You are lucky to get 30Metres out of a run
>> without Negotiation issues.
>> I encourage you strongly to look to using Optical LX style fibre if
>> you are moving to 10G. The cost of structured cabling and the length
>> of runs is a lot cheaper than Cat6a/7 runs needed for 10gbit copper.
>> On 9 September 2014 16:03, Joel Wirāmu Pauling <joel at aenertia.net>
>> wrote:
>> > Just a head's up I have had issues with the X5xx intel SFP+ optics
>> > interoperating with other vendors. This may not be an issue for your
>> > deployment.
>> >
>> > If you want good interop then Broadcom based optics seem to be the
>> > best bet at the moment.
>> >
>> > On 9 September 2014 12:09, Wes Felter <wmf at felter.org> wrote:
>> >> The Intel X710 just came out today; it's a little cheaper. (Note that
>> >> the
>> >> X710 is 10G and the XL710 is 40G because XL is 40 in Roman numerals.)
>> >>
>> >>
>> http://www.colfaxdirect.com/store/pc/viewPrd.asp?idproduct=2227&idcategory=0
>> >>
>> >> (I've never bought from Colfax, but they carry a lot of "enterprise"
>> >> equipment that isn't available on NewEgg/Amazon.)
>> >>
>> >> There are super-cheap optics and twinax available from China. You may
>> have
>> >> to hack the Intel driver since they're not "supported".
>> >>
>> >> http://www.fiberstore.com/c/10g-sfp+_63
>> >> http://www.fiberstore.com/c/10g-sfp+-cables_1115
>> >>
>> >> Or you could just get an account at the Snabb Lab.
>> >>
>> >> --
>> >> Wes Felter
>> >> IBM Research - Austin
>> >>
>> >>
>> >> _______________________________________________
>> >> Cerowrt-devel mailing list
>> >> Cerowrt-devel at lists.bufferbloat.net
>> >> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>> _______________________________________________
>> Cerowrt-devel mailing list
>> Cerowrt-devel at lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel

Dave Täht


More information about the Cerowrt-devel mailing list