Internet for serious voice service ?

Discussion in 'VOIP' started by Pete Harris, Oct 14, 2003.

  1. Pete Harris

    Pete Harris Guest

    How good is the public Internet for serious phone service? Can you use
    it for business, or is it just for cheap phone calls?

    I hear a lot of conflicting reports.

    If we assume perfectly reliable connections to the ISPs at each end
    (no laughter, please), and that the ISPs themselves boast 100% uptime
    (same comment), how is the Internet itself as a medium for phone
    calls? How often do outages of 5 or more seconds occur? I picked 5
    seconds because that's probably about how long the party who is not
    talking will wait before hanging up.

    Even when there's no prolonged outage, how often does voice quality
    degrade into momentary unintelligibilty?

    Does distance make a big difference? Can we assume that SOHO workers
    calling their offices in the same geographical area will get much
    better performance than people calling across the USA or across
    Europe? Can even these close-by workers be satisfied, or do they need
    to sign up for special (and probably expensive) QoS services?

    - Pete
    www.bg.com/qphone
     
    Pete Harris, Oct 14, 2003
    #1
    1. Advertising

  2. Pete Harris wrote:

    > How good is the public Internet for serious phone service? Can you use
    > it for business, or is it just for cheap phone calls?


    IMHO only to be used for cheap phone calls and perhaps SOHO workers.

    > I hear a lot of conflicting reports.
    >
    > If we assume perfectly reliable connections to the ISPs at each end
    > (no laughter, please), and that the ISPs themselves boast 100% uptime
    > (same comment), how is the Internet itself as a medium for phone
    > calls? How often do outages of 5 or more seconds occur? I picked 5
    > seconds because that's probably about how long the party who is not
    > talking will wait before hanging up.


    The internet itself is fine. The problem occurs in the fact that routing
    is unreliable. One moment the call quality might be fine, than a router
    decides to die, and your route runs via China. Call quality might not be
    that good now. IMHO the 5 seconds marker you set is too high especially
    for business use. I think the accepted rule is that anything over 150 ms
    RTT causes quality to degrade. Anything over 500 ms RTT caused the
    'breakie-breakie-over' effect.

    > Even when there's no prolonged outage, how often does voice quality
    > degrade into momentary unintelligibilty?


    See above.

    > Does distance make a big difference? Can we assume that SOHO workers
    > calling their offices in the same geographical area will get much
    > better performance than people calling across the USA or across
    > Europe? Can even these close-by workers be satisfied, or do they need
    > to sign up for special (and probably expensive) QoS services?


    This depends mainly on your ISP. I live in the Netherlands and use a
    provider that has direct connect to NYIIX, so for me the service to the
    states would probably be fine. My neighbour uses a different ISP that
    needs peering to get to the states, his quality would not be as good.

    Next to this there are all sorts of factors. Here in the netherlands
    there are multiple DSL providers, one causes more lag on the DSL network
    than the other.

    Then you have things like jitterbuffers in phones, packetloss and echo
    cancellation. I think it would relatively save to *assume* that as long
    as you tested the 'internal' network of the provider (ie. all remote
    users and head office are connected to same provider) and found it's
    quality to be ok, that the call quality would probably be ok.

    There are several tools that can be used to estimate link
    characteristics and from that estimate the maximum call quality. I have
    used bing for this (with a 'b'). It measures link througput and RTT
    times based on a specified packetsize and is able to vary the packet
    size, output example :

    bing -u 8 -z -P -c 5 localhost www.bg.com
    BING localhost (localhost) and www.bg.com (207.158.238.114)
    44 and 108 data bytes (1024 bits)
    www.bg.com: 203.376Kbps 5.035ms 4.916992us/bit
    www.bg.com: 407.481Kbps 2.513ms 2.454102us/bit
    www.bg.com: 325.596Kbps 3.145ms 3.071289us/bit
    www.bg.com: 325.596Kbps 3.145ms 3.071289us/bit
    www.bg.com: 325.700Kbps 3.144ms 3.070313us/bit
    www.bg.com: 325.700Kbps 3.144ms 3.070313us/bit
    www.bg.com: 325.700Kbps 3.144ms 3.070313us/bit
    www.bg.com: 2.276Mbps 0.450ms 0.439453us/bit
    www.bg.com: 2.281Mbps 0.449ms 0.438477us/bit
    www.bg.com: 5.120Mbps 0.200ms 0.195313us/bit
    www.bg.com: 6.606Mbps 0.155ms 0.151367us/bit
    www.bg.com: 978.967Kbps 1.046ms 1.021484us/bit
    www.bg.com: 983.670Kbps 1.041ms 1.016602us/bit
    www.bg.com: 983.670Kbps 1.041ms 1.016602us/bit
    www.bg.com: 983.670Kbps 1.041ms 1.016602us/bit
    www.bg.com: 983.670Kbps 1.041ms 1.016602us/bit
    www.bg.com: 983.670Kbps 1.041ms 1.016602us/bit
    www.bg.com: 983.670Kbps 1.041ms 1.016602us/bit
    www.bg.com: 983.670Kbps 1.041ms 1.016602us/bit

    --- localhost statistics ---
    bytes out in loss rtt (ms): min avg max std dev
    44 31 31 0% 0.454 0.547 2.141 0.297
    52 31 31 0% 0.420 0.594 1.188 0.310
    60 31 31 0% 0.421 0.650 5.448 0.908
    68 31 31 0% 0.420 0.476 1.153 0.180
    76 31 31 0% 0.419 0.428 0.454 0.009
    84 31 31 0% 0.418 0.427 0.457 0.009
    92 31 31 0% 0.421 0.426 0.433 0.003
    100 31 31 0% 0.421 0.427 0.453 0.006
    108 31 31 0% 0.419 0.426 0.433 0.003

    --- www.bg.com statistics ---
    bytes out in loss rtt (ms): min avg max std dev
    44 30 30 0% 140.977 144.011 146.784 1.299
    52 30 30 0% 141.190 143.408 149.122 1.912
    60 30 30 0% 141.810 142.883 145.470 1.019
    68 30 30 0% 141.475 143.622 150.335 2.247
    76 30 30 0% 141.314 142.676 145.788 0.893
    84 30 30 0% 141.829 143.172 148.107 1.407
    92 30 30 0% 142.314 144.084 148.763 1.322
    100 30 30 0% 142.121 144.629 147.807 2.167
    108 30 30 0% 141.983 145.825 152.416 2.737

    --- estimated link characteristics ---
    host bandwidth ms
    www.bg.com 983.670Kbps 140.522


    Arnold
     
    Arnold Ligtvoet, Oct 14, 2003
    #2
    1. Advertising

  3. Pete Harris

    root Guest

    Pete Harris wrote:

    > How good is the public Internet for serious phone service? Can you use
    > it for business, or is it just for cheap phone calls?
    >


    What exactly is your term for "cheap phone calls"? Do you mean poor audio
    quality (not to be confused with jittering), audio jittering, delays?

    It has been more than six months since I have run a small network of
    VoIPBlaster (VB) devices for a Point-2-Point VoIP telephone conversation
    through my cable modem connection in the US. So, my VB device as well as
    the phonebook server are always online. The other VB parties use a 56Kbps
    dialup modem to connect to the Internet and they are located in S.E. Asean
    countries. Honestly, the audio quality from VB devices is crystally clear,
    i.e. no noise and acoustic/line echoes, with an exception of some delays (~
    1/2 sec or even less) due to Internet connection between the two points.

    > I hear a lot of conflicting reports.
    >
    > If we assume perfectly reliable connections to the ISPs at each end
    > (no laughter, please), and that the ISPs themselves boast 100% uptime
    > (same comment), how is the Internet itself as a medium for phone
    > calls? How often do outages of 5 or more seconds occur? I picked 5
    > seconds because that's probably about how long the party who is not
    > talking will wait before hanging up.
    >


    If you have experienced a delay of about 5 sec, definitely the Internet
    connection between your point to the other party is to blame.

    > Even when there's no prolonged outage, how often does voice quality
    > degrade into momentary unintelligibilty?
    >


    The audio quality does not degrade for a P2P VoIP connection since it is a
    digitized signal. If you are referring to the delays and echoes, that's
    another story (at least that's my opinion).

    > Does distance make a big difference? Can we assume that SOHO workers
    > calling their offices in the same geographical area will get much
    > better performance than people calling across the USA or across
    > Europe? Can even these close-by workers be satisfied, or do they need
    > to sign up for special (and probably expensive) QoS services?
    >
    > - Pete
    > www.bg.com/qphone


    Definitely. The further the two points (in term of routing), the higher the
    delays. These days, I can say all cheap calling cards you can buy on the
    market are based on VoIP. Some Long Distance Telephone companies (I am
    talking about the big giants) are also using the VoIP to cover their costs
    to compete with the emerging calling card companies. That's why sometimes
    one will experience less delays when calling the same number.

    --
    root/administrator
     
    root, Oct 14, 2003
    #3
  4. Pete Harris

    RC Guest

    Simple answer is no QoS (quality of service). Might be great one second and
    just plain gone the next. This doesn't work for business customer calls, ok
    for internal calls or calling home from school. Things like that.

    Physical distance in meaningless, routed distance affects delay. I've done a
    trace route to sites on the same ISP and literally next door, only to find
    that the data was going through Delaware (I'm in Boston). When I was doing
    video conferencing (before VoIP, remember Picture Tel) we used 250ms as a
    max delay before most people would notice. This was one-way but included all
    the processing. IP pings don't include the audio processing so would have to
    be much less. Around 250 round trip.




    "Pete Harris" <> wrote in message
    news:...
    > How good is the public Internet for serious phone service? Can you use
    > it for business, or is it just for cheap phone calls?
    >
    > I hear a lot of conflicting reports.
    >
    > If we assume perfectly reliable connections to the ISPs at each end
    > (no laughter, please), and that the ISPs themselves boast 100% uptime
    > (same comment), how is the Internet itself as a medium for phone
    > calls? How often do outages of 5 or more seconds occur? I picked 5
    > seconds because that's probably about how long the party who is not
    > talking will wait before hanging up.
    >
    > Even when there's no prolonged outage, how often does voice quality
    > degrade into momentary unintelligibilty?
    >
    > Does distance make a big difference? Can we assume that SOHO workers
    > calling their offices in the same geographical area will get much
    > better performance than people calling across the USA or across
    > Europe? Can even these close-by workers be satisfied, or do they need
    > to sign up for special (and probably expensive) QoS services?
    >
    > - Pete
    > www.bg.com/qphone
     
    RC, Oct 14, 2003
    #4
  5. Pete Harris

    root Guest

    RC wrote:

    > Simple answer is no QoS (quality of service). Might be great one second
    > and just plain gone the next. This doesn't work for business customer
    > calls, ok for internal calls or calling home from school. Things like
    > that.
    >
    > Physical distance in meaningless, routed distance affects delay. I've done


    Is that so? Try to do a VoIP with someone who uses a high-orbit satelite
    (orbiting @ about 230,000 KM from earth) and see if that will make a
    difference in delays. AFAIK, a round-trip on a high-orbit satelite
    introduces a total of about 540msec delays. Now, if one is in the US and
    the other one is on the other side of the globe, there may involve two
    high-orbit satelite links, i.e. source -> 1st satelite -> 2nd satelite ->
    destination, which definitely will introduce more delays. Fortunately,
    Internet links that use a satelite connection mostly involve with low-orbit
    satelites; therefore, the delays on up/down-links are pretty small even
    though the P2P connection is half the globe away.

    > a trace route to sites on the same ISP and literally next door, only to
    > find that the data was going through Delaware (I'm in Boston). When I was


    If you are in Boston using ISP A and your neighbor uses ISP B, there is a
    chance that your (VoIP) connection to your neighbor involves lots of hops
    and it may also route through Los Angeles, CA. At any rate, each hop will
    also introduce some delays.

    > doing video conferencing (before VoIP, remember Picture Tel) we used 250ms
    > as a max delay before most people would notice. This was one-way but
    > included all the processing. IP pings don't include the audio processing
    > so would have to be much less. Around 250 round trip.
    >
    >
    >
    >
    > "Pete Harris" <> wrote in message
    > news:...
    >> How good is the public Internet for serious phone service? Can you use
    >> it for business, or is it just for cheap phone calls?
    >>
    >> I hear a lot of conflicting reports.
    >>
    >> If we assume perfectly reliable connections to the ISPs at each end
    >> (no laughter, please), and that the ISPs themselves boast 100% uptime
    >> (same comment), how is the Internet itself as a medium for phone
    >> calls? How often do outages of 5 or more seconds occur? I picked 5
    >> seconds because that's probably about how long the party who is not
    >> talking will wait before hanging up.
    >>
    >> Even when there's no prolonged outage, how often does voice quality
    >> degrade into momentary unintelligibilty?
    >>
    >> Does distance make a big difference? Can we assume that SOHO workers
    >> calling their offices in the same geographical area will get much
    >> better performance than people calling across the USA or across
    >> Europe? Can even these close-by workers be satisfied, or do they need
    >> to sign up for special (and probably expensive) QoS services?
    >>
    >> - Pete
    >> www.bg.com/qphone


    --
    root/administrator
     
    root, Oct 15, 2003
    #5
  6. Pete Harris

    shido Guest

    I'll set you up with a free account using our service and let you make your
    own decisions about voip. When it is done properly you dont need ASR's or
    worry about echo or sound quality. If its from NuFone you can expect only
    the very best. You can use your favorite softphone or ip phone or you're
    using Asterisk we can set you up with an I)nter A)sterisk eX)change account.
    Normally our US and Canada rates are around 2.9 cents / minute , since we
    use paypal ( ) turn up times vary from 30 seconds to 2
    minutes. If you request to port an 800# or if you need a new 800# it can
    vary from 2 minutes to 4 or 5 days.

    --
    Greg Merriweather
    The NuFone Network

    519-251-8225 x 3000


    IM:
    "Pete Harris" <> wrote in message
    news:...
    > How good is the public Internet for serious phone service? Can you use
    > it for business, or is it just for cheap phone calls?
    >
    > I hear a lot of conflicting reports.
    >
    > If we assume perfectly reliable connections to the ISPs at each end
    > (no laughter, please), and that the ISPs themselves boast 100% uptime
    > (same comment), how is the Internet itself as a medium for phone
    > calls? How often do outages of 5 or more seconds occur? I picked 5
    > seconds because that's probably about how long the party who is not
    > talking will wait before hanging up.
    >
    > Even when there's no prolonged outage, how often does voice quality
    > degrade into momentary unintelligibilty?
    >
    > Does distance make a big difference? Can we assume that SOHO workers
    > calling their offices in the same geographical area will get much
    > better performance than people calling across the USA or across
    > Europe? Can even these close-by workers be satisfied, or do they need
    > to sign up for special (and probably expensive) QoS services?
    >
    > - Pete
    > www.bg.com/qphone
     
    shido, Oct 15, 2003
    #6
  7. Pete Harris

    RC Guest

    "root" <,> wrote in message
    news:rD4jb.564197$cF.241978@rwcrnsc53...
    > RC wrote:
    >
    > > Simple answer is no QoS (quality of service). Might be great one second
    > > and just plain gone the next. This doesn't work for business customer
    > > calls, ok for internal calls or calling home from school. Things like
    > > that.
    > >
    > > Physical distance in meaningless, routed distance affects delay. I've

    done
    >
    > Is that so? Try to do a VoIP with someone who uses a high-orbit satelite
    > (orbiting @ about 230,000 KM from earth) and see if that will make a
    > difference in delays. AFAIK, a round-trip on a high-orbit satelite
    > introduces a total of about 540msec delays. Now, if one is in the US and
    > the other one is on the other side of the globe, there may involve two
    > high-orbit satelite links, i.e. source -> 1st satelite -> 2nd satelite ->
    > destination, which definitely will introduce more delays. Fortunately,
    > Internet links that use a satelite connection mostly involve with

    low-orbit
    > satelites; therefore, the delays on up/down-links are pretty small even
    > though the P2P connection is half the globe away.


    In your satalite example routed distance is the same as physical distance,
    so of cource you get increased delay. Odly enough you picked a problem I've
    had to work with, video conferening over 2 sat hops. The delay ended up as
    over 450ms one, including high level encrypted signal. With such a large
    delay the vid-conf units would drop the connection. Easy enough to fix, just
    bumpted up the time-out on signal aknowlegment. But with almost a second
    roud trip delay it was considered impractical (the Admeral didn't like it).

    >
    > > a trace route to sites on the same ISP and literally next door, only to
    > > find that the data was going through Delaware (I'm in Boston). When I

    was
    >
    > If you are in Boston using ISP A and your neighbor uses ISP B, there is a
    > chance that your (VoIP) connection to your neighbor involves lots of hops
    > and it may also route through Los Angeles, CA. At any rate, each hop will
    > also introduce some delays.


    The pint was that even if you use the same ISP, you have no garrentee that
    you routed distance isn't much more then you physical distance.

    >
    > > doing video conferencing (before VoIP, remember Picture Tel) we used

    250ms
    > > as a max delay before most people would notice. This was one-way but
    > > included all the processing. IP pings don't include the audio processing
    > > so would have to be much less. Around 250 round trip.
    > >
     
    RC, Oct 31, 2003
    #7
  8. Pete Harris

    Gavin Scott Guest

    root <,> wrote:
    > Is that so? Try to do a VoIP with someone who uses a high-orbit satelite
    > (orbiting @ about 230,000 KM from earth)


    Er, which would be what exactly? Geostationary orbit is something
    like 35,900 KM form the surface, so if your provider's stats are
    orbiting the Moon then I'd suggest a change :)

    G.
     
    Gavin Scott, Oct 31, 2003
    #8
  9. Gavin Scott wrote:

    > root <,> wrote:
    >> Is that so? Try to do a VoIP with someone who uses a high-orbit satelite
    >> (orbiting @ about 230,000 KM from earth)

    >
    > Er, which would be what exactly? Geostationary orbit is something
    > like 35,900 KM form the surface, so if your provider's stats are
    > orbiting the Moon then I'd suggest a change :)
    >
    > G.


    I believe that is a low orbit satelite.

    --
    root/administrator
     
    root/administrator, Nov 4, 2003
    #9
  10. Pete Harris

    Hank Karl Guest

    On 14 Oct 2003 00:41:53 -0700, (Pete Harris) wrote:

    >How good is the public Internet for serious phone service? Can you use
    >it for business, or is it just for cheap phone calls?

    The answer is very subjective and depends on your demands as well as
    what the internet can provide. You may pay the 2.9 cents/minute for a
    US domestic call because you think it sounds slightly better than
    VoIP. You may also have a free VoIP service like Free World Dialup
    because some of your customers will accept a lower voice quality if
    they can call you for free. And you may find that calls to some
    numbers over the PSTN are so bad that VoIP sounds better anyway.

    >
    >I hear a lot of conflicting reports.

    Quality of VoIP calls can vary depending on the route and other
    traffic present on the network. And the route between two given
    points changes from time-to-time, it can even change during a call.

    >If we assume perfectly reliable connections to the ISPs at each end
    >(no laughter, please), and that the ISPs themselves boast 100% uptime
    >(same comment), how is the Internet itself as a medium for phone
    >calls? How often do outages of 5 or more seconds occur? I picked 5
    >seconds because that's probably about how long the party who is not
    >talking will wait before hanging up.
    >
    >Even when there's no prolonged outage, how often does voice quality
    >degrade into momentary unintelligibilty?

    Complete outages are probably very rare. Poor voice quality is much
    more common. http://www.voiptroubleshooter.com/voipt_usersymptoms.htm
    lists many of the problems you may experience over the public
    internet, and will give you an idea of what causes the problems.

    RFC 3611 (RTCP XR) specifies quality reporting for VoIP, but its just
    out. When this RFC becomes more widely implemented, it will be easy
    to measure the quality of your calls which will make it easier to
    resolve the problem. ( See
    http://www.voiptroubleshooter.com/voipt_netsymptoms.htm )

    >Does distance make a big difference? Can we assume that SOHO workers
    >calling their offices in the same geographical area will get much
    >better performance than people calling across the USA or across
    >Europe? Can even these close-by workers be satisfied, or do they need
    >to sign up for special (and probably expensive) QoS services?


    You have to look at distance as the number of hops between two points,
    not the number of miles between two points. Each router in the chain
    adds delay and jitter and may drop packets, and otherwise inject
    problems into the data stream. If you and your associate are in
    adjoining apartments, but on different ISPs, your call may be routed
    through more devices than a call on the same ISP that goes to another
    town.
    >- Pete
    >www.bg.com/qphone
     
    Hank Karl, Nov 5, 2003
    #10
    1. Advertising

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Sven Holm
    Replies:
    0
    Views:
    1,513
    Sven Holm
    Sep 7, 2003
  2. Sven Holm
    Replies:
    0
    Views:
    1,422
    Sven Holm
    Sep 9, 2003
  3. Sven Holm
    Replies:
    5
    Views:
    4,506
    Arnold Ligtvoet
    Sep 12, 2003
  4. henry
    Replies:
    1
    Views:
    517
    Michael S. Cooper
    Nov 8, 2003
  5. Giuen
    Replies:
    0
    Views:
    995
    Giuen
    Sep 12, 2008
Loading...

Share This Page