A recent post from Siklu's, Danny BenSimhon, grabbed my attention with claims of multi-gigabit for crazy distances: 5, 8, 10 miles, “and more”. I commented that those distances shocked me. I want to believe, but I'm unconvinced. Here are my thoughts on the subject, and I'm curious about yours, so please chime in at the end.
Siklu makes an awesome radio, which I know by hands-on experience.
In my opinion, there isn't a better 80GHz radio on the market. Yet Siklu's marketing about their “Overbuild” solution is a serious stretch to me.
For starters, Overbuild is a great concept, but hardly a new idea. Essentially, Siklu is taking an 80GHz radio with kick-ass throughput, and enabling it to be paired with lower frequency, lower throughput radios that aren't so susceptible to rain. In that way, you get 2GB from the Siklu radio for extended distances, backed by a safety net for weather.
IgniteNet has been doing something along the same lines, pairing a blazing fast 60GHz radio with a longer range 5GHz backup. And before that, many of us were already combining lower and higher frequency radios in the field, except integration happened at the customer's switch, rather than within the radio apparatus.
The problem with Siklu's Overbuild, is that it fails to perform miracles.
Granted, miracles are a lot to ask, but look at the below charts. If you're rubbing your eyes raw, stop. Your vision is fine.
Here's what stands out to me. First, the crazy distances for 80GHz—5, 8 and 10 miles—purportedly at a constant 2GB for longer than 51-weeks a year. Isn't that only possible where cacti are indigenous? Elsewhere, extended 80GHz paths will frequently drop (in modulation) to speeds well below 2GB.
Also, these charts say that Miami, with its tropical monsoon climate, is better for 80GHz than New York. Having had installations in both states, I have a hard time accepting this, regardless of what Siklu's link budget calculator says.
Let me tell you a related tale about a horrible installation.
I was consulting on a project for the city of Providence, Rhode Island. One of their paths was 2.93 miles long and I specified a redundant set of 18GHz radios.
A Siklu competitor then learned of the opportunity and sold the customer on an 80GHz solution. I protested to no avail, except that I made the vendor vouch for their uptime numbers (“link availability”) in the bid docs.
Flash forward to the install. I expected the link to be problematic in rain, but I was stunned by just how bad it was. Despite the extended 2' dish, a rock solid pipe mount and optimal alignment, the link crashed from the first week. Modulation drops were constant, even in piddling rain. The customer amassed hundreds of pages of error logs and called repeatedly for field support.
Months later, after the radio vendor put us through a gauntlet of idiotic questions and repeated firmware installations, I reminded them that their reputation was on the line in the bid docs, and made them take their radios back (4 of them). Hutton Communications made the exchange for two 18GHz DragonWave links, after which the customer finally had a rock solid connection and I could get on with my life.
Today, I'm designing a network for a WISP that wants the highest customer satisfaction in semi urban communities.
To that end, I'm specifying Siklu's multi-gig radios (50-60 links), and restricting all paths to one and a quarter miles (1.25) from tower base stations. Think that's too conservative?
Look at the heroes at Everywhere Wireless in Chicago. Keegan Bonebrake, EVP of Strategy, told me that they keep 80GHz paths to eight-tenths of a mile (.8). Think he's wrong? Tell it to the 100 out of 100 customers that gave Everywhere Wireless a perfect, 5-star rating. These guys are making wireless look good!
Siklu's Overbuild is a good idea, but applications are limited and I'd be careful not to overdo it. The way I would sell it, is to promise a customer that they'll get 300 to 800 Mbps *all the time*, based on whatever the lower frequency radios do, but that remarkably, they'll also get 1-2 GB from the Siklu radio on most days. What's not awesome about that?
On the other hand, if the customer demands a consistent 2GB on 5, 8 and 10 mile paths for 360 some-odd days a year, then unless you can see Las Vegas, don't gamble on 80GHz.
Why do I care?
I work for WISPs and want to see them succeed. Moreover, I want our industry to succeed. I've had it up to my eyeballs with misperceptions about wireless having weather problems, and Overbuild threatens to prolong that fairy tale. Truth is, a well planned and installed fixed wireless link will never go down in rain, except where power is lost.
Siklu, if I'm wrong, educate me, and I'll follow up with another article where I eat my words. I'd love to revisit my network design and go from 1.25 mile tower radiuses to 1.5 or even 2-miles. It would make a huge difference. Right now, I don't dare do it.