Well, first of all RFC3021 made back in December of 2000 says you can
do a /31 on PTP links just fine ;)

Secondly, if you're in the situation where you have a single flow that
needs more than 1Gbps, your options are very limited. I agree it's not
the best solution if for whatever reason that you have no switches at
the site, and I'd also say than even with hardware offloading LAG/LACP
really sucks balls over wireless due to the changing throughput of the
links... but yeah.

It does work, but again - there really needs to be a layer2 or layer3
based protocol to handle this kind of thing. Stupid Performant
Networks / Accedian patents :/

On Sat, Apr 2, 2016 at 4:03 PM, Eric Kuhnke <[email protected]> wrote:
> I'm sure that works but I'm not going to build an OSPF /30 router-to-router
> 1 to 10GbE backbone link that looks like:
>
> router -> switch -> fiber riser to roof -> radio -> air -> radio -> fiber
> riser to roof -> switch -> router
>
> introduce a switch in the path between two routers just because?
>
>
> On Fri, Apr 1, 2016 at 9:12 PM, Josh Reynolds <[email protected]> wrote:
>>
>> You don't do it on a router, you bond it with a switch and do the
>> layer 4 hashing on it's ASIC, then uplink it via SFP+.to your router.
>>
>> On Fri, Apr 1, 2016 at 5:45 PM, Eric Kuhnke <[email protected]> wrote:
>> > I am not aware of any notable backbone-ready router platform (something
>> > that
>> > speaks BGP+OSPF+MPLS, takes full tables and can carry multiple VRFs)
>> > that
>> > does layer-4 hashing in ASICs/hardware...   Routers are not intended to
>> > be
>> > load balancers.
>> >
>> > It's not a microwave problem, the same issue is seen if you portchannel
>> > two
>> > 1Gbps fiber connections between routers.
>> >
>> > On Fri, Apr 1, 2016 at 3:26 PM, Josh Reynolds <[email protected]>
>> > wrote:
>> >>
>> >> That's not exactly true, it depends on the hashing method. If your
>> >> equipment can do Layer 4 hashing, then you absolutely can maximize the
>> >> aggregate throughput via single stream.
>> >>
>> >> Sadly, LAG/LACP as a whole like this sucks terribly for
>> >> wireless/microwave.
>> >>
>> >> On Fri, Apr 1, 2016 at 5:05 PM, Eric Kuhnke <[email protected]>
>> >> wrote:
>> >> > Looking at the 2200FX it has regular SFP ports, so at 2Gbps (500 MHz
>> >> > channel
>> >> > 32QAM?) FDD, you can only achieve 2Gbps by doing an 802.3ad between
>> >> > two
>> >> > routers. In which case no single customer data stream will exeed
>> >> > 1Gbps.
>> >> > Not
>> >> > quite the same thing as having a radio with a SFP+ 10GbE interface.
>> >> >
>> >> > That said it will probably be less costly than the equivalent options
>> >> > from
>> >> > Bridgewave, E-Band or perhaps SIAE.
>> >> >
>> >> > On Fri, Apr 1, 2016 at 2:59 PM, Peter Kranz <[email protected]>
>> >> > wrote:
>> >> >>
>> >> >> Siklu has both a high power and low power 2Gbps radio now. The high
>> >> >> power
>> >> >> unit is not on their website but due next month and its really much
>> >> >> more
>> >> >> than the cheap one honestly.
>> >> >>
>> >> >>
>> >> >>
>> >> >> Peter Kranz
>> >> >> www.UnwiredLtd.com
>> >> >> Desk: 510-868-1614 x100
>> >> >> Mobile: 510-207-0000
>> >> >> [email protected]
>> >> >>
>> >> >>
>> >> >>
>> >> >> From: Af [mailto:[email protected]] On Behalf Of Eric Kuhnke
>> >> >> Sent: Wednesday, March 30, 2016 1:55 PM
>> >> >> To: [email protected]
>> >> >> Subject: Re: [AFMUG] Russians made a 10 Gbps radio?
>> >> >>
>> >> >>
>> >> >>
>> >> >> Without violating any manufacturer's NDA, this is the tip of an
>> >> >> iceberg,
>> >> >> the Russians just happen to be publishing info a tiny bit earlier
>> >> >> than
>> >> >> the
>> >> >> other 80 GHz radio manufacturers. There are 10GbE 256QAM FDD radios
>> >> >> in
>> >> >> development and field testing from all of the significant players in
>> >> >> the
>> >> >> industry.
>> >> >>
>> >> >> I would expect the 'new' 5Gbps to 10Gbps radios to ship in the US at
>> >> >> a
>> >> >> price point under $20k per link including antennas, which will
>> >> >> occupy
>> >> >> the
>> >> >> price tier previously occupied by the 1Gbps high-powered 80 GHz
>> >> >> stuff
>> >> >> that
>> >> >> began shipping 3.5 to 4 years ago. Then you have the lower powered
>> >> >> less
>> >> >> expensive 80 GHz stuff like Siklu which has a max Tx power of +8 or
>> >> >> +10
>> >> >> and
>> >> >> will come down further from its already-low price under $10k/link.
>> >> >>
>> >> >>
>> >> >>
>> >> >> On Wed, Mar 30, 2016 at 9:07 AM, Mike Hammett <[email protected]>
>> >> >> wrote:
>> >> >>
>> >> >> Putting 2 and 2 together, from a FB post, it looks like they're 20k
>> >> >> GBP
>> >> >> per link. That sounds simply awesome if they are.
>> >> >>
>> >> >>
>> >> >>
>> >> >> -----
>> >> >> Mike Hammett
>> >> >> Intelligent Computing Solutions
>> >> >>
>> >> >> Midwest Internet Exchange
>> >> >>
>> >> >> The Brothers WISP
>> >> >>
>> >> >>
>> >> >>
>> >> >> ________________________________
>> >> >>
>> >> >> From: "Eric Kuhnke" <[email protected]>
>> >> >> To: [email protected]
>> >> >> Sent: Monday, March 28, 2016 5:42:25 PM
>> >> >> Subject: [AFMUG] Russians made a 10 Gbps radio?
>> >> >>
>> >> >> http://www.elva-1.com/news_events/a40107
>> >> >>
>> >> >> http://www.elva-1.com/products/a40106
>> >> >>
>> >> >> http://www.elva-1.com/data/files/Datasheets/2016_02_24_PPC-10G.pdf
>> >> >>
>> >> >> 2000 MHz wide channel and 256QAM for 10 Gbps in the FDD 71-86 GHz
>> >> >> bands.
>> >> >> Question is...  What's the Rx level needed for that, and how quickly
>> >> >> does it
>> >> >> drop off with rain?
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >
>> >> >
>> >
>> >
>
>

Reply via email to