CWNP Wireless Certification & Wireless Training

Your Account |

Search:

Free Stuff

Mesh has never been easier

Add to:

Written by Devin Akin

Saturday, 19 January 2008

For a long time, I avoided mesh like the plague. It was non-standard, and everyone had their own radically different take on how things should be done. After what I'd call a slow start, Firetide has really made some snazzy moves over the last year. I just spent 8 hours in the lab going through the latest available indoor mesh hardware and code. What can I say? It just works.

The coolest part is that it works like you'd expect it to work if you were building a mesh system. It's intuitive and flexible (generally attributes that don't go too well together). Firetide's East Coast SE, Mike Graham, is the Mesh Master and an all-around RF guru. I'm reasonably sure he understands the details behind every single feature in the entire platform. :) I got to play with their dual-radio 6000 series mesh nodes and their 4000 series APs. The Hotview management software discovers and configures these devices easily and quickly. No hickups, which was impressive.

While Firetide apparently spends most of its time on rooftops (deploying outdoor systems), their indoor systems work nicely and are a great fit for a number of vertical markets. Their entire system acts like one big Ethernet switch once deployed. It's extremely slick. I wouldn't hesitate to deploy other vendors' split-MAC WLAN systems right over the top of it. They have a bonding feature where both radios in each mesh node can be on a different channel (whether in the same band or not) and form dual links to every other mesh node. This allows for double throughpu, with the current caviat that a single flow (based on source/destination pair) doesn't seem to use all of a mesh link's bandwidth. It takes 2 flows to see traffic across both mesh links because of how the load-balancing works.

Their radios can do anything from 2.4 to 4.9 to 5.x GHz. Nice.

Comments (6) Add Comment

Subscribe to this comment's feed

...
written by Slipshod, January 23, 2008

Etherchannel over the air! Not surprised about the caveat on the loadsharing - that's very typical of Ethernet loadsharing algorithms. They are typically hashing src&dst L2 and/or L3 addresses. It works well with real clients, but gets kind of rough with an overlay WLAN product because all the client traffic on one AP would count as the same flow.

Sounds like a nice system. Mesh is definitely getting to be very mature, and upcoming 3-radio systems go a long way towards mitigating multi-hop bandwidth contention issues. Too bad the muni market has largely imploded; I was looking forward to ubiquitous access in the major metro areas.

...
written by JARPUnator, January 28, 2008

It is really difficult to implement Mesh.

...
written by FarVision, January 28, 2008

Mesh discussion without the word Meraki in it? Terrible!

...
written by Frank, January 29, 2008

So, what's your take on the enterprise WLAN vendors doing the same, most notable Aruba, Cisco, and Trapeze?

...
written by Devin Akin, January 29, 2008

WLAN infrastructure vendors such as Aruba, Cisco, Trapeze, Meru, Ruckus, and others that are implementing mesh as an extension of their Split-MAC architecture are doing something completely different than Firetide is doing at this point. Firetide's mesh can be thought of as one big Ethernet switch, whose guts are wireless. I'm not sure we'll see alot of head-to-head competition between these two completely different architectures since they suit such different needs. I haven't played with anyone's Split-MAC mesh except Ruckus's, and theirs is currently "light mesh" - meaning one hop from a portal AP. It worked well. I'm looking forward to playing with others.

...
written by Slipshod, January 29, 2008

Most of the split-MAC mesh seems to be an add-on instead of an end unto itself. You usually don't see the all-mesh vendors going up against the WLAN folks - it's a different set of customers. Cisco is probably the one spending the most effort on Mesh (1500 seems like a good outdoor AP, and the DOCSIS support in the 1520 is interesting in it's niche), but even so the code is a mess with an entirely seperate code branch for mesh and lots of caveats on roaming and interoperability of mainline software.

Write comment