It's not only about geolocation. This is not related to Netflix, but.. I had a problem getting to Google (everything Google) from home a month or two ago for about two hours. But it worked if I remote desktop'd into my PC at the office. The primary DNS at the NOC goes out through GTT. The secondary (which I hit at home) sits on AT&T. Same address space. A query for google.com from one server gave me a completely different result than the other. So they're responding based on what network you're coming in on as well. Both did get me to Google in Chicago though. My guess is different load balancers. Netflix could be doing the same thing, I don't know.

On 3/23/2015 12:05 PM, Ken Hohhof wrote:
I also don’t know how important it is that a CDN be “near” you.
By definition, you’re probably talking a sustained download, either a video stream or some kind of large file download. And most of the time you will see 4 parallel TCP connections. I really don’t think latency matters once you start the download. What does matter is server balancing. If your DNS server has correct geoIP but Netflix chooses to send your customers to a server in Dallas, maybe their Chicago servers are overloaded or undergoing maintenance. Do you really want to second guess their decisions? About all you can do is make sure your DNS server is in the right place according to the geolocation database services, and let the content provider decide what IP address to hand out to your customers and how to route that IP (they may use geoIP info to decide the routing, not the DNS). Now, if your DNS server appears to be in a whole wrong part of the world, that may have dramatic effects, like totally different content being available because Netflix thinks your customer is in Europe or Asia.
*From:* Mike Hammett <mailto:af...@ics-il.net>
*Sent:* Monday, March 23, 2015 11:55 AM
*To:* af@afmug.com <mailto:af@afmug.com>
*Subject:* Re: [AFMUG] testing DNS server performance
Pardon the mess, I'm on a laptop with a damn touchpad.



-----
Mike Hammett
Intelligent Computing Solutions
http://www.ics-il.com

------------------------------------------------------------------------
*From: *"That One Guy" <thatoneguyst...@gmail.com>
*To: *af@afmug.com
*Sent: *Monday, March 23, 2015 11:43:50 AM
*Subject: *Re: [AFMUG] testing DNS server performance

used the wrong term
Ignore the term
Take cache out of thyne mouth
now, being a windows dick, I dont have torch
I want to simply be able to verify that appropriate CDNs are being utilized
namebench is still running, I dont know what its output is going to be
This cant be a new thing, I see threads occasionally about content being problematic in that users are getting less than desirable CDNs, it always seems to boil down to DNS, I just want a tool that will tell me where the content is coming from. (in a perfect world, it would display on a map with a quality indicator to that CDN, I dont have any expectation that that component of the tool would exist) On Mon, Mar 23, 2015 at 11:34 AM, Chuck McCown <ch...@wbmfg.com <mailto:ch...@wbmfg.com>> wrote:

    Not generic.  You have to use the one they provide.  And they will
    not give to you unless you are doing some like 4tB per month.
    *From:* Ken Hohhof <mailto:af...@kwisp.com>
    *Sent:* Monday, March 23, 2015 10:33 AM
    *To:* af@afmug.com <mailto:af@afmug.com>
    *Subject:* Re: [AFMUG] testing DNS server performance
    I was not aware you could cache Netflix streams with a generic
    caching server.  Not only due to DRM, but also Netflix app
    switches streams dynamically to match video quality to connection
    speed. Plus first the customer authenticates to Netflix server,
    chooses what content to watch, etc.
    *From:* Chuck McCown <mailto:ch...@wbmfg.com>
    *Sent:* Monday, March 23, 2015 11:23 AM
    *To:* af@afmug.com <mailto:af@afmug.com>
    *Subject:* Re: [AFMUG] testing DNS server performance
    Say a new movie is on Netflix.  Or latest season of cards.
    Everyone is going to want to watch it.  So 1000 simultaneous
    backbone streams to Netflix vs 1000 simultaneous streams to the
    caching server in your NOC.  I choose the latter.

    ------------------------------------------------------------------------
    *From: *"Ken Hohhof" <af...@kwisp.com <mailto:af...@kwisp.com>>
    *To: *af@afmug.com <mailto:af@afmug.com>
    *Sent: *Monday, March 23, 2015 11:08:27 AM
    *Subject: *Re: [AFMUG] testing DNS server performance

    I don’t understand how the caching server is going to help with
    CDNs. Actually, with so much Internet content now being either
    dynamic HTML or streaming, I wouldn’t think caching would be worth
    it, unless you are talking about something like a Netflix
    OpenConnect appliance. Maybe you can cache software updates, I’m
    not sure about that.
    *From:* That One Guy <mailto:thatoneguyst...@gmail.com>
    *Sent:* Monday, March 23, 2015 10:35 AM
    *To:* af@afmug.com <mailto:af@afmug.com>
    *Subject:* Re: [AFMUG] testing DNS server performance
    Geographically close CDNs. I want to make sure we are getting
    content from Illinoisish rather than california for netflix, since
    all that matters is netflix
    On Mon, Mar 23, 2015 at 10:30 AM, Josh Baird <joshba...@gmail.com
    <mailto:joshba...@gmail.com>> wrote:

        For performance, look at queryperf which I think is provided
        by ISC/bind.
        I'm not sure what you mean by "we are getting good CDNs and
        the like," though.
        Josh
        On Mon, Mar 23, 2015 at 11:27 AM, That One Guy
        <thatoneguyst...@gmail.com <mailto:thatoneguyst...@gmail.com>>
        wrote:

            Im bringing live our first caching server today. Is there
            a good tool for comparing queries between DNS servers.
            Im not all that concerned about speed since we are so
            small there wont be a huge amount of benefit I would
            think. Im primarily wanting to make sure we are getting
            good CDNs and the like
-- If you only see yourself as part of the team but you don't
            see your team as part of yourself you have already failed
            as part of the team.



-- If you only see yourself as part of the team but you don't see
    your team as part of yourself you have already failed as part of
    the team.



--
If you only see yourself as part of the team but you don't see your team as part of yourself you have already failed as part of the team.

Reply via email to