Re: [WISPA] PTP Link Recommendations - The Future of WISPs
Butch, I do not disagree with your statements. I'm jsut saying readers may misinterperete the post if my statements were not added. If the intend is to do FDX, Using OFDM to accomplish it is one easy way to do it. The question that I was challenging is in what cases FDX is appropriate. It can be risky to rely on two channels in a noisy environment. What happens if those channels develop interference? What channels would you move to? Anyone can get a good link on day one, but what is the plan for preventing future disasters? I'm starting to justify FDX now that 10Mhz channels are becoming available. Two 10Mhz channels has less risk and more options than one 20 Mhz channel. When we first got into this business, we really thought we were smart. We were trying all kind of neat things. But at the end of the day, we learned reasons why the rest of the world wasn't doing it already. Its humbling for me to admit that publically, but I'm a smarter person for realizing it. We were doing a lot with OFDM in the early years. The idea was to use two 10mbps Trangos on two freqs to reach 20 mbps. Sometimes Full Duplex, Some times bonded Half Duplex. The end result is the noise floor got to high, and it was to hard to move channels around, when needing to cater to what channel was deployed adjacent, taking up two channels for a single link. Selecting FDX might have meant not serving a particular direction. The problem is that when a channel gets packet loss, OSPF doesn't know what to do, when its up and when its down. We were setting up OSPF so that they took different paths with full duplex immulated but if one of the links went down, the second path (by OSPF) was the other channel converting it to half duplex in the emergency state. In general we deliver packet loss less links. But its not just a factor of the technology to accomplish that in noisy areas. Its strategy of the WISP. Using more spectrum for a task than one needs to, can be wasteful, and give WISPs less options for selecting the channels that will allow them to use radios that will prevent the packet loss. Again, excellent arguements have been made on this list, specifically by Matt Liotta and Lonnie, on how using 10Mhz channels or GPS syncing with Full Duplex for a link can actually be more spectrum efficient than 20Mhz channels in Half Duplex. In real world I have not seen that yet, but in theory it all made sense. Its actually those debates that got me thinking to start doing more PtP links on my network and attempting spectrum re-use, that I am now more effectively accomplishing. My general rule is Do you need 20 mbps for the link? Do you demand Full Duplex? If the answer to both of these question is yes, and it may be for dominately VOIP applications, then it may very well be worth using OSPF for FDX deployments. But there is risk in doing it. Because if I really only need 10 mbps, or can survive with half duplex, I'd rather know that if any of my links encounter interference, that it only takes down half the customers, not all the customers, because I have customers spread out across more radios. Having two radios operating independantly using half duplex, allows redundancy on the fly, when needed. My end of the day conclusion was, if in Rural or Licensed, go for it, but otherwise I wouldn't do FDX unless doing it with 10Mhz channel size. The exception to this is that when 5.4Ghz gear is legal, there are many more available channels where it is less risky to take two channels for a link. 5.4Ghz will be the spectrum that revolutionizes FDX and PTP links in Urban/Suburban America. 5.4Ghz is almost useless in PtMP on small sectors. But it allows PTP links to go 7 miles with margin. Can you think of it now, 10Mhz channels on 100% clear fresh spectrum on day 1, thats 50 new channels, including polarity. Or 25 new channels using smart Dual Pol NLOS antennas. It will be the year of 2 ft dishes, to go the distance with low power spectrum. 2007 will be the year of smart routing. Not MESH as the world typically knows as MESH, but MESH as the definition, a network with two paths or more. PTP reduces latency over PTMP systems, allowing more hops to deliver the same QOS. Networks will be designed to go to building to buidling more often. WISPS will start to install two antennas as a requirement for every new install. Networks will be run layers on top of each other in parallel, so customers have fewest number of hops to the transit locations, but so more buildings can be served. In other words it won't be one large mesh, it will be many small mesh segments, with engineered PTP paths. This is not a new idea, just equipment options and costs have changed to allow it to become viable, and previous mass deployment and adoption, demanding it from the providers. The secret to it all is smart routing. It will come, or peoiple will leave routing and go back to layer2
RE: [WISPA] PTP Link Recommendations - The Future of WISPs
I don't know if you are right or wrong Tom...only time will tell...but I'm clapping! -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tom DeReggi Sent: Friday, September 22, 2006 2:15 PM To: WISPA General List Subject: Re: [WISPA] PTP Link Recommendations - The Future of WISPs Butch, I do not disagree with your statements. I'm jsut saying readers may misinterperete the post if my statements were not added. If the intend is to do FDX, Using OFDM to accomplish it is one easy way to do it. The question that I was challenging is in what cases FDX is appropriate. It can be risky to rely on two channels in a noisy environment. What happens if those channels develop interference? What channels would you move to? Anyone can get a good link on day one, but what is the plan for preventing future disasters? I'm starting to justify FDX now that 10Mhz channels are becoming available. Two 10Mhz channels has less risk and more options than one 20 Mhz channel. When we first got into this business, we really thought we were smart. We were trying all kind of neat things. But at the end of the day, we learned reasons why the rest of the world wasn't doing it already. Its humbling for me to admit that publically, but I'm a smarter person for realizing it. We were doing a lot with OFDM in the early years. The idea was to use two 10mbps Trangos on two freqs to reach 20 mbps. Sometimes Full Duplex, Some times bonded Half Duplex. The end result is the noise floor got to high, and it was to hard to move channels around, when needing to cater to what channel was deployed adjacent, taking up two channels for a single link. Selecting FDX might have meant not serving a particular direction. The problem is that when a channel gets packet loss, OSPF doesn't know what to do, when its up and when its down. We were setting up OSPF so that they took different paths with full duplex immulated but if one of the links went down, the second path (by OSPF) was the other channel converting it to half duplex in the emergency state. In general we deliver packet loss less links. But its not just a factor of the technology to accomplish that in noisy areas. Its strategy of the WISP. Using more spectrum for a task than one needs to, can be wasteful, and give WISPs less options for selecting the channels that will allow them to use radios that will prevent the packet loss. Again, excellent arguements have been made on this list, specifically by Matt Liotta and Lonnie, on how using 10Mhz channels or GPS syncing with Full Duplex for a link can actually be more spectrum efficient than 20Mhz channels in Half Duplex. In real world I have not seen that yet, but in theory it all made sense. Its actually those debates that got me thinking to start doing more PtP links on my network and attempting spectrum re-use, that I am now more effectively accomplishing. My general rule is Do you need 20 mbps for the link? Do you demand Full Duplex? If the answer to both of these question is yes, and it may be for dominately VOIP applications, then it may very well be worth using OSPF for FDX deployments. But there is risk in doing it. Because if I really only need 10 mbps, or can survive with half duplex, I'd rather know that if any of my links encounter interference, that it only takes down half the customers, not all the customers, because I have customers spread out across more radios. Having two radios operating independantly using half duplex, allows redundancy on the fly, when needed. My end of the day conclusion was, if in Rural or Licensed, go for it, but otherwise I wouldn't do FDX unless doing it with 10Mhz channel size. The exception to this is that when 5.4Ghz gear is legal, there are many more available channels where it is less risky to take two channels for a link. 5.4Ghz will be the spectrum that revolutionizes FDX and PTP links in Urban/Suburban America. 5.4Ghz is almost useless in PtMP on small sectors. But it allows PTP links to go 7 miles with margin. Can you think of it now, 10Mhz channels on 100% clear fresh spectrum on day 1, thats 50 new channels, including polarity. Or 25 new channels using smart Dual Pol NLOS antennas. It will be the year of 2 ft dishes, to go the distance with low power spectrum. 2007 will be the year of smart routing. Not MESH as the world typically knows as MESH, but MESH as the definition, a network with two paths or more. PTP reduces latency over PTMP systems, allowing more hops to deliver the same QOS. Networks will be designed to go to building to buidling more often. WISPS will start to install two antennas as a requirement for every new install. Networks will be run layers on top of each other in parallel, so customers have fewest number of hops to the transit locations, but so more buildings can be served. In other words it won't be one large mesh, it will be many small mesh segments, with engineered PTP paths. This is not a new idea
Re: [WISPA] PTP Link Recommendations - The Future of WISPs
On Fri, 22 Sep 2006, Tom DeReggi wrote: I'm jsut saying readers may misinterperete the post if my statements were not added. The parts that you added, though, were specifically related to the radio portion of the network (and you were correct there), but it has nothing to do with what we did (now 3 times) with OSPF. It was that part that I responded to. If the intend is to do FDX, Using OFDM to accomplish it is one easy way to do it. OFDM isn't full duplex, unless I'm sorely mistaken. The question that I was challenging is in what cases FDX is appropriate. It can be risky to rely on two channels in a noisy environment. What happens if those channels develop interference? What channels would you move to? Anyone can get a good link on day one, but what is the plan for preventing future disasters? This is actually a better question. The fact is, that MOST people who brag on their 10, 20, 50Mbps infrastructure don't need anywhere near that. In 2 of the cases where I built this type of setup, there was a real need that a FDX implementation solved. These were not ISPs, but a corporate install where they were doing VoIP among other things. The third one was an ISP and they had 2 links up and running (one in 5.8 and another in 2.4) and wanted to find a way to utilize these better (they were previously just bridged and STP was running). Two 10Mhz channels has less risk and more options than one 20 Mhz channel. This is obviously true and very spectrum conscious. Its humbling for me to admit that publically, but I'm a smarter person for realizing it. :-) I have a hard time admitting that I _could_ have _possibly_ been wrong. ;-) (I guess that's just human nature.) I tend to agree (a little) with the vision that you posted, but that's also a bit off the topic at hand. I guess you get the right to stray from the topic, being the long-timer that you are. :-) -- Butch Evans Network Engineering and Security Consulting 573-276-2879 http://www.butchevans.com/ Mikrotik Certified Consultant (http://www.mikrotik.com/consultants.html) -- WISPA Wireless List: wireless@wispa.org Subscribe/Unsubscribe: http://lists.wispa.org/mailman/listinfo/wireless Archives: http://lists.wispa.org/pipermail/wireless/
Re: [WISPA] PTP Link Recommendations - The Future of WISPs
If the intend is to do FDX, Using OFDM to accomplish it is one easy way to do it. OFDM isn't full duplex, unless I'm sorely mistaken. OOPs typo. MEant OSPF. Tom DeReggi RapidDSL Wireless, Inc IntAirNet- Fixed Wireless Broadband - Original Message - From: Butch Evans [EMAIL PROTECTED] To: WISPA General List wireless@wispa.org Sent: Friday, September 22, 2006 5:31 PM Subject: Re: [WISPA] PTP Link Recommendations - The Future of WISPs On Fri, 22 Sep 2006, Tom DeReggi wrote: I'm jsut saying readers may misinterperete the post if my statements were not added. The parts that you added, though, were specifically related to the radio portion of the network (and you were correct there), but it has nothing to do with what we did (now 3 times) with OSPF. It was that part that I responded to. If the intend is to do FDX, Using OFDM to accomplish it is one easy way to do it. OFDM isn't full duplex, unless I'm sorely mistaken. The question that I was challenging is in what cases FDX is appropriate. It can be risky to rely on two channels in a noisy environment. What happens if those channels develop interference? What channels would you move to? Anyone can get a good link on day one, but what is the plan for preventing future disasters? This is actually a better question. The fact is, that MOST people who brag on their 10, 20, 50Mbps infrastructure don't need anywhere near that. In 2 of the cases where I built this type of setup, there was a real need that a FDX implementation solved. These were not ISPs, but a corporate install where they were doing VoIP among other things. The third one was an ISP and they had 2 links up and running (one in 5.8 and another in 2.4) and wanted to find a way to utilize these better (they were previously just bridged and STP was running). Two 10Mhz channels has less risk and more options than one 20 Mhz channel. This is obviously true and very spectrum conscious. Its humbling for me to admit that publically, but I'm a smarter person for realizing it. :-) I have a hard time admitting that I _could_ have _possibly_ been wrong. ;-) (I guess that's just human nature.) I tend to agree (a little) with the vision that you posted, but that's also a bit off the topic at hand. I guess you get the right to stray from the topic, being the long-timer that you are. :-) -- Butch Evans Network Engineering and Security Consulting 573-276-2879 http://www.butchevans.com/ Mikrotik Certified Consultant (http://www.mikrotik.com/consultants.html) -- WISPA Wireless List: wireless@wispa.org Subscribe/Unsubscribe: http://lists.wispa.org/mailman/listinfo/wireless Archives: http://lists.wispa.org/pipermail/wireless/ -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.1.405 / Virus Database: 268.12.8/455 - Release Date: 9/22/2006 -- WISPA Wireless List: wireless@wispa.org Subscribe/Unsubscribe: http://lists.wispa.org/mailman/listinfo/wireless Archives: http://lists.wispa.org/pipermail/wireless/