[1st-mile-nm] Agave Broadband News

[John Brown]

________________________________
From: John Brown
Sent: Thursday, December 08, 2011 4:20 PM
To: Kirk Kirkpatrick; 1st-mile-nm at mailman.dcn.org
Subject: RE: [1st-mile-nm] Agave Broadband News

I responded to the Press Release post.  My comments:   (Note that I don't go out as Mr. Kirkpatrick did and say that he is a stinky poo-poo head)

1. Document that MVDDS is not NEW technology as the Press Release leads one to believe.
    Mr. Kirkpatrick comments further support that the technology is not "NEW" but has been around for at least 2 decades

2. Much of the data with respects to what can the license holder do or not do is cut and past from the FCC website and the electronic version of the license.
    The data is fairly easy to get to and if people want URL's to the specifics I'm happy to provide.

3. Wikipedia stated that HyperCable was from the 80's.  If they are wrong, then someone should fix it.

4. The technology is ONE-WAY, pursuant to the FCC license.  Technically speaking this means that the end subscriber may not transmit on the licensed MVDDS  frequency.
    As I stated and as Mr. Kirkpatrick stated in response and as the FCC license states, to use this technology as a TWO-WAY solution will require a different path away from the end subscriber.
    Yes, this could be a 3G Cellular Phone based path, it could be a WiFi, or something else in the unlicensed 2.4 or 5.x or other frequency blocks.
    I caution that if it is in the 2.4 or 5.x blocks that the noise floor in ABQ is pretty HIGH, and that current City of Santa Fe ordinances prohibit the placement of antennas without prior approval by the City.
    This COULD lead to frequency congestion and thus packet loss.  Same holds true for the 3G return path solution.

5. I also caution that focusing on the downlink speed is only a partial answer.  TCP requires a reasonable return path.  If that return path has delay, congestion, packet-loss, etc then
    there will be problems with the effective download speed perceived by the end user.   Since the packet flow is ASYMMETRICAL, the asymmetry of the path will have to be
    carefully managed.  In general this is the BDP (Bandwidth Delay Product).

6. BDP in part causes packet buffer sizing issues.  Packets need to be buffered until the ACK is received back from the end subscriber.  If that ACK is lost, then the packet
   must be resent.   Larger buffers require more RAM on the networking devices, which in turn causes a higher BOM (Bill of Materials) cost for the equipment.   The larger the pipe the larger the buffers have to be, unless the
   return path is really really good.

7.  It would interesting to know what the supported MTU is for the MVDDS technology on the RF interface.  What overhead exists in addition to the IP header.  Is there any SAR performed on the payload.

8. The single license issue..  The FCC has presently granted a single license for a single transmitter.  Based on that public government data, my statement that should this transmitter fail, users would loose this service, is a reasoned statement.   Yes, Mr. Kirkpatrick is correct that they could request and be granted additional transmitter licenses by the FCC.  But today, they have a single license for a single transmitter.

9. "Shared bandwidth"  In looking at the bandwidth delivered to a subscriber, Cable and RF (cable is RF as well) are shared.  DSL, T1, DS3 are point to point.  Being that there is a dedicated path between the specific end-subscriber and the first level aggregation equipment.  Cable / RF based systems are shared.  Certain Fiber based systems, ala PON are also shared.  The "sharing" is in some cases done with TDM.

10.  I don't see a screen shot at the URL posted by Mr. Kirkpatrick for the real-world bandwidth in ABQ.
       Here is a recent one from one of our residential subscribers http://www.speedtest.net/result/1626433361.png

11. It will be interesting to run pathchar, iperf and other similar tools on this type of network.

12.  According to the MDSA website "http://www.mdsamerica.com/index.cfm/technology/mdsa-mvdds-gigaband/" I would like clarification on the ".. to avoid fixed IP addresses."  Does this prevent the end subscriber from having a static publicly routed IP address, without any form of NAT/PAT.  If yes, then this would prevent customers from having their own server(s) onsite.

Cheers...


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www2.dcn.org/pipermail/1st-mile-nm/attachments/20111208/b44220e3/attachment.html>