re: No VOIP for PanamaNorwayGÇÖs Lyse Tele Builds Ethernet FTTH with Cisco Metro Network Lyse Tele, a power utility company that serves 110,000 customers in southwest Norway, is deploying direct fiber connections to residential customers in the Stavanger region of the country using a Cisco Metro Ethernet network. Lyse has connected 500 subscribers in the initial phase of the network rollout and is offering flat rate VoIP packages, video on demand, access to 40 channels of broadcast TV over a single IP-based Ethernet network. The deployment includes Cisco Catalyst 6509 switches in the core, Catalyst 4006 switches in the Ethernet distribution network, the Cisco AS5350 Universal Gateway, and Cisco 3640 Internet routers. Financial terms were not disclosed.
re: No VOIP for PanamaMost SP are bandwidth limited by the edge. Does that mean MPLS is irrelevant for most SP?? -------
Not at all :)
First off MPLS != TE. If carriers want to deploy L2 or L3 VPN services then an MPLS core makes sense. Sure, you can run a non-MPLS technology (UTI/L2TPv3 or virtual routers) or can run MPLS/GRE (to support draft-martini or RFC2547 without an MPLS core) but these solutions typically require more hardware ("tunnel PICs" etc.)
Secondly TE isn't only about bandwidth saving (so contrary to my earlier comment it *may* buy you something even if you have plenty of core bandwidth). For example using pre-signalled secondary paths you can get faster reconvergence than using an IGP.
re: No VOIP for PanamaHi BBBoy, If you bind enough DS0 into a ATM VCC, you may not need echo cancellation with ATM AAL1. Hence, AAL1 is a very low cost option in term of equipment/media gateway of transporting voice.. You waste bandwidth by using AAL1. However, you can get some back by doing DB-CES...
AAL2 save bandwidth but required more equipment/media gate processing power since it usually require echo cancellation and AAL2 SARing/Muxing require more processing power..
re: No VOIP for Panamadreamer 101 wrote: "Unless the explicit gioal is to make some equipment vendor rich, AAL2 chips is plenty and available. I would think that the jury is still out.."
Who are the leading AAL2 chip suppliers? I know of two, NMS has a group in Canada (used to me InnomediaLogic?) doing one, and also Agere. Any others? Who has the most market share?
I'm not sure that AAL2 is the current VoATM technology of choice. Sure, its standard on wireless, but the big US wireline implementations I have heard about are mainly AAL1, are they not?
I think of AAL2 as providing great trunking efficiency, but AAL1 as better on access lines and easier for switching.
Comments?
I am also curious about the Telus announcement of so-called VoIP on their LD network. There are no details in the articles cited here, but they are using Passport ATM switches. Could they really be doing VoIP not riding over any ATM layer? I find that hard to believe at this point. If its just trunking, they are incurring extra overhead for what benefit? Maybe some of our Canadian friends can help us out on this one.
By the way, this is one of the best and most educational threads I have every read on LR. Thanks to Dreamer, Packet Man, LGGS, Lucifer, Skeptic, etc. Keep it up!
Actually, a handset is doing VoIP in a Release 5 context doesn't have ATM implemented.
The Uu interface (the air interface) is defined as:
Application/IP/PPP/PDCP/MAC/RF
So, for VoIP (bearer) from the UE, we'd see:
Voice/vocoder/RTP/UDP/IP/PPP/PDCP/MAC/RF
ATM doesn't figure in until the UTRAN (the RNS and RNC), where inter-device transport is ATM. So, if you believe what the standards say, the handset wouldn't have a smidgen of ATM implemented.
Now, that doesn't mean that ATM won't be important. Quite the contrary, the UTRAN is ATM-based for transport, once you get to the land-based interfaces. Moreover, there's nothing preventing the service provider from implementing the IP core over an ATM transport layer. Infact, that approach would seem obvious if they're making the ATM investment in the UTRAN.
Just to totally confuse things, there are some vendors who are developing IP-based transport off of the RNC.
re: No VOIP for Panama"Even when 3G is all IP, the IP packet is chopped into cells via PPPmux in the RAN.. Do you call that an IP packet or not?? If someone is going to chop the IP packet into cells why do it via PPPmux as opposed to ATM cells?? Unless the explicit gioal is to make some equipment vendor rich, AAL2 chips is plenty and available. I would think that the jury is still out.."
Hi Dreamer,
If you re-read my post. I stated that indeed the RAN is ATM-based, and, if you believe what 3GPP Rel 4 and Rel 5 tells you, the RAN will always be ATM-based, which means that, in a RAN context, ATM will be alive and well for a long time.
Looking back at my post, however, I realized I mis-spoke. When I said
"That's not to say that the jury is out and everything in the wireless world will go VoIP. At least one major wireless carrier is going VoAAL2 (what I consider REAL VoATM) in the RAN and core."
I meant to say
"That's not to say that the jury is IN and everything in the wireless world will go VoIP. At least one major wireless carrier is going VoAAL2 (what I consider REAL VoATM) in the RAN and core."
So I agree with you, the jury is still most certainly out.
As to AAL2 chips being available, you're absolutely right. What I had stated, however, was that I believed that IP handsets are more mature from a deployment perspective. In fact, any PC with a 3G wireless modem could do VoIP. These same devices, however, couldn't do VoATM with the same equipment.
re: No VOIP for PanamaIn you definition, what is a low speed leased line?? Is it DS1 and lower?? Is it multiple DS1?? MLPPP usually run over multiple DS1 lines.. Is DS3 high speed or low speed leased line?? There is no point arguing about low speed leased without knowing each other definitions. -------------------
Given that you are the one making the anti-IP arguement, its up to you to do your own research. If you don't know what speed/bandwidth causes problems, your original arguement has no basis.
And I have no interest in going out and gathering information for you if you can't even be bothered to make your own case.
3GPP Release 4.0 defines the Radio Access Network (RAN) as VoAAL2, so I'm not surprised to see a VoAAL2 phone. However, 3GPP Release 4 defines an IP-based core network. That means that the VoAAL2 call really only applies to the Iu-cs (the radio network part), in other words, if thecall is to remain in packet form, it will be converted to VoIP to traverse the core.
Also, 3GPP Release 5 defines an all-IP model, the Iu-cs is done away with and the Iu-ps (which was carrying the 3G data in Release 4) now also carries voice from VoIP handsets. Call control is via a SIP. Note, however, that the Iu-ps data link layer is ATM, in keeping with the ATM-based RAN architecture.
Given the delay of true 3G rollouts, and given that Release 5 defines an all-IP infrastructure, and given that IP-based handsets have much more airtime, I wonder if some carriers might skip the R4 step and jump straight to IP (why invest in the infrastructure when the standards are saying it'll be VoIP - if you believe what standards bodies say)
That's not to say that the jury is out and everything in the wireless world will go VoIP. At least one major wireless carrier is going VoAAL2 (what I consider REAL VoATM) in the RAN and core.
Also, since 3GPP defines the RAN as ATM-based, there will always be ATM there.
re: No VOIP for PanamaHi, Even when 3G is all IP, the IP packet is chopped into cells via PPPmux in the RAN.. Do you call that an IP packet or not?? If someone is going to chop the IP packet into cells why do it via PPPmux as opposed to ATM cells?? Unless the explicit gioal is to make some equipment vendor rich, AAL2 chips is plenty and available. I would think that the jury is still out..
re: No VOIP for PanamaHi Skeptic, In my definition, anything less than DS3 is low speed. Somebody trying to show that VoIP (cRTP) on MLPPP is more bandwidth efficient than VoATM and it is due to cell tax etc... In the sense of fairness, I am trying to show that MLPPP has its own fair share of fragment tax.. By the way, PacketMan example of VoIP efficiency is on a 768 Kbps leased line which won't you agree is a low speed line... Why is it okay to be Anti-ATM but it is not okay to be anti-IP?? Isn't this a bit one sided??
VoIP is useful on the network that you have a lot of bandwidth. With the connectionless nature of IP, you do not need to setup connection for VoIP calls. Bandwidth efficiency is never a virtue of VoIP..
re: No VOIP for PanamaHere is a few news clips regarding Telus:
"From where I'm sitting it brings the best decision from our leadership in terms of what we are going to do and what we're not going to do," says Pathak. "For example, I could have continued to invest in (Asynchronous Transfer Mode) infrastructure, but with the IP announcement I made it very clear we've stopped investing in the ATM side of the house." http://www.itbusiness.ca/index...
Quigley said itGÇÖs important to note that Telus is taking baby steps towards all-IP infrastructure. "First itGÇÖs just going to be all their long-haul, inter-exchange stuff. TheyGÇÖre going to leave the core metro networks alone for now. ItGÇÖs not like everybodyGÇÖs going to be told, GÇÿYour IP phone is in the mail. Get ready to join us in the 22nd century.GÇÖ" Pathak said Telus would eliminate the ATM layer in its network, taking care to ensure ATM users are ported onto IP without problems. http://www.itworld.ca/portals/...
re: No VOIP for PanamaNGN 2002: VoIP Reality Check - The Provider Perspective The killer app on the Internet is voice, according to Ofer Gneezy, President & CEO, iBasis. Speaking at the Next Generation Networks conference, Gneezy said that packet telephony offers substantial efficiencies and cost savings in the network core. For international voice traffic, packet telephony is already huge. iBasis is one of top ten carriers of international voice traffic, measured in minutes, and all of it is VoIP. In the coming years, Gneezy believes that the international success of VoIP will extend to domestic long distance and local calls.
Henry Sinnreich, Distinguished Member of Engineering at WorldCom believes that Session Initiation Protocol (SIP) is the best solution for packet telephony. It is more flexible and provides better performance than VoIP or IP Centrix, and offers mobility, the integration of communications and data, and provides multimedia communications options. Commenting on the quality of packet telephony, Sinnreich said that errors and dropouts are an issue only for calls that access the Internet via low bandwidth connections. When true broadband connections are available at both ends of a call, even international calls that go through 20 router hops sounds perfect without customization of the network for voice. Sinnreich also commented that GÇ£flat billing is the nirvana.GÇ¥ Flat billing has been proven to be successful and is appealing to customers in voice and other telecommunications services. It also avoids all the complications of billing systems.
Guido Roda, Service Engineering Director at FastWeb, a new Italian service provider, described how his company uses FTTH to provide integrated voice, data, Internet, and video over IP to 100,000 customers. To enable voice services, FastWeb overlaps an H.323 infrastructure on its IP based transport backbone. The service provider has learned that multimedia services generate the most excitement, but commodity voice services are still customersGÇÖ benchmark for satisfaction. Voice service delivered by packet must meet existing voice expectations of toll quality and 100% availability for the service to be accepted. http://www.convergedigest.com Converge! Network Digest, 16-Oct-02
re: No VOIP for Panama====== Hi Packet Man, If MPLS Traffic Engineering is really good, why Equant (one of the leading MPLS SP) choose not to implement that.. http://www.telecommagazine.com... ======
because they are bandwidth limited at the edge rather than at the core - and so it wouldn't really buy them anything?
Have you got any idea or info on how many customers (i.e. how much traffic) is likely to be up and running over the IP network you're using?
It would be an interesting indication, since there's no doubt that it works great, except very minor issues on the quality of the voice, if there's plenty of bandwidth for each customer.
Hi All, For those interested in knowing that NTT DoCoMo 3G is implemented in ATM AAL2, the above URL provide a Google search that show all the articles relate to that...
re: No VOIP for PanamaHi Packet Man, Be specific.. What is the fragment size that you use when you have a 768Kbps line?? What is the fragment size that you use when you have a T1 line?? If it is less than 53 bytes, do not complain about ATM cell tax.. MLPPP has a bigger fragment tax than ATM..
re: No VOIP for PanamaHi Skeptic, In you definition, what is a low speed leased line?? Is it DS1 and lower?? Is it multiple DS1?? MLPPP usually run over multiple DS1 lines.. Is DS3 high speed or low speed leased line?? There is no point arguing about low speed leased without knowing each other definitions.
re: No VOIP for PanamaHaving the Government shutoff VoIP to protect the weak and dying LECs would be like the govt stopping email because the US Postal service is losing money. I don't believe the US Govt would go to this extent. The Panamanian Govt is clearly misguided...
It would seem that some industry players (telco, music, software) and the some govts are starting to take initiatives to block the ability to conduct "content transfer" of certain content types. While I am dead against unlawfull pirating of copyrighted material, I am fearful of what could become a more bigger problem to the Internet. If "things" on the Internet get blocked or blacked out, what do we as a society do? VoIP, and music swapping are just two things. Will someday they say "FTP is illegial, no more port 20/21". Sure we will go invent some other new whiz-bang protocol to transfer content, but as time goes on life and doing business on the Internet will be plagued with complicated software just to do simple commerce.
Or will it simply be futile for these organizations to even try blocking? After some time will they simply give up?
Here's another thought? The Internet is here to stay, no changing that. Will modern society become so dependant on the Internet that any attempt by a govt/industry to stop certain evolution will simply have the long term consequence of putting that company/industry at risk of becoming uncompetitive with major impact to the economy.
The telco industry is just one example. What do you think? How long can a govt prevent certain services from being avaialble on the Internet? How long can the Internet go before becoming 100% regulated? Will it ever get there?
re: No VOIP for Panamadreamer101 wrote: Do you know that when you run VoIP on Multilink PPP, all (data plus voice )packets are fragmented?? Hence, it suffered the same inefficiency that everyone that complain about like ATM cell tax. The only difference is that it is not 53 bytes cell. It is a different size cell.
My comment: I am very aware of the potential problems that could occur when running VoIP on very slow WAN links. However unlike you, I know how to set the configs properly. When done properly, only the large packets containing data are fragmented down into smaller pieces. When done properly the packets containing voice are not fragmented. When you set the threshold too small everything is fragmented. Knowing what you are doing is key. Dreamer101 you are wrong. As skeptic said you are either ignorant or dishonest.
I did a search, by author dreamer101, on all of your posts here in LR. Regardless of the subject, topic or thread you claim ATM is the only way, and IP and/or MPLS do not work, and ATM is why you are laughing your way to the bank. I don't have to confess that I am an IP QoS flag waver everyone knows that, but you will never read me saying ATM don't work. At least present to the readers stuff that we can truly debate. And what did you post a few days ago....a cell phone that is doing VoATM or VoIPoATM? I am understaing you right? This cell phone terminates ATM inside of it? Is that true? Seriously is that true? I am not waisting my time relying to any of your ignorant/self-serving posts. I'm starting a new thread about this article, no need to reply dreamer/boobymax101.
re: No VOIP for PanamaDo you know that when you run VoIP on Multilink PPP, all (data plus voice )packets are fragmented?? Hence, it suffered the same inefficiency that everyone that complain about like ATM cell tax. The only difference is that it is not 53 bytes cell. It is a different size cell. ---------------------
Do you know that you misread that article and took a feature that was designed for optimization of voice/data on low-bandwidth WAN interfaces and generalized it to all VOIP and all multi-link PPP.
Did you just not read the following paragraph or did you deliberatly ignore it:
-------------------------------------- If you try to run VoIP with PPP over a low bandwidth leased line, you run the risk of encountering massive latency and jitter due to large data packets that seize the leased line, "freezing out" all other traffic until these large packets are sent across. This is where the concept of packet fragmentation comes into play. Real-time frames can then be interleaved among the fragmented data frames. -----------------------------------
re: No VOIP for Panama You know, you are at a point where you are either very ignorant or very dishonest.
Did you even bother to read the title of that cisco article:
"VoIP with PPP Over Low Bandwidth Leased Line"
If you try to do voice and data, over really slow-speed WAN serial interfaces, you are going to have huge latency and jitter problems with voice if you don't have a small fixed packet size to keep the data packets under control. This isn't an IP problem, its a problem with combining voice/data on low bandwidth links in general.
The real answer is that its foolish to combine voice and data traffic over these sorts of links in the first place. But for people who want to do so, there is a solution.
Your line of reasoning here suggests that network protocols should be designed "down" to support the oldest and slowest devices in existance. I'm glad that IP isn't optimized to carry voice and data over slow-speed serial modems. But I dont understand why that bothers you.
re: No VOIP for PanamaHi PacketMan, Do you know that when you run VoIP on Multilink PPP, all (data plus voice )packets are fragmented?? Hence, it suffered the same inefficiency that everyone that complain about like ATM cell tax. The only difference is that it is not 53 bytes cell. It is a different size cell.
re: No VOIP for PanamaHi All, On one hand the IP folks, complaining about ATM cell tax. On the other hand, the IP folks is doing SARing with VoIP over MLPPP...
<<< When using Multilink PPP, the ip precedence command causes data packet fragmenting. This is to minimize voice packet delay. Setting the precedence bit in an IP header gives priority to the voice packets over data frames or data packets. >>
Or, is it one of those thing as long as IP folks doing it, it is a good thing but it is done via ATM it is bad??
Please stop saying one thing and do something else..
re: No VOIP for PanamaHi Packet Man, What VoIP is doing to WAN is a waste of bandwidth.. The best and most efficient way of transporting voice is actually compression and silence suppression on TDM mode. Netrix can transport 10 channels of toll quality voice on 64Kbps leased line...
(1) I like this answer. I nver looked at ATM being used this way. Is this being done today?
(2) Don't quite agree with this one. For WAN links with limited bandwidth you can use cRTP (compressed RTP/UDP/IP) over PPP. See my post regarding my little test on a 768Kbs WAN link. I could of gotten more than 6 calls out of it, but I gave some bandwidth to regular data as the customer wanted.
(3) Okay. So you think that of the carriers today that are doing VoPacket, they all have ATM (VoIPoATM or VoATM) in their network. Not one has 100% IP and 0% ATM (VoIPoPOS)?? I have some research to do.
You said: On the other side of the coin, ATM makes no sense as the "end-to-end" network for all services. In the Enterprise, VoIP works very well, ATM adds complexity and cost that isn't need.
It's not either IP or ATM. They are different things; both have applications. IP is the application protocol, ATM is definitely an infrastructure protocol.
I say: I agree with you all the way. I have never said ATM doesn't work, I can't say it cause it does work. I just like to try to give some arguement to those who say you have to have ATM period, cause IP QoS don't work period. Take dreamer101 for example. He is an ATM bigot. You read every one of his posts in LR and he is always putting down IP and MPLS. I think he is BoobyMax's brother. If I hired him to run my staff we'd be putting in $200 dollar ATM NICS and IP/ATM stacks in PC, and then buying ATM switches with ELAN software. And remember those IP routers we don't need? Well I guess I will have to buy one with an ATM interface and connect it to his 100% ATM network so I can get my ATM based ELAN (VLANS) to pass routed IP between the two.
Where do you see MPLS in all of this? Hey you got any good WWW links for some great reading on MPLS?
I am still waiting for someone to give me a concrete example where ATM QoS provided a solution where the newer IP QoS tools could not.
A few come to mind:
1. Circuit Emulation -- replacement of cross connects with a packet network. This is a HUGE network invesment.
2. WAN Voice with limited bandwidth (also a very large market) -- needs short packets to manage latency and jitter, IP overhead is too high to be practicable.
3. PSTN Voice -- it turns out to very expensive to engineer an IP (or MPLS) network to carry Voice the with 99.999% availability and MOS 4.0 voice quality required by the PSTN. Any carrier doing this seriously uses ATM as the Network layer to provide QoS to packet voice traffic, whether it's ATM or IP.
On the other side of the coin, ATM makes no sense as the "end-to-end" network for all services. In the Enterprise, VoIP works very well, ATM adds complexity and cost that isn't need.
It's not either IP or ATM. They are different things; both have applications. IP is the application protocol, ATM is definitely an infrastructure protocol.
Opinion: companies that focus on IP will claim that IP will solve all network problems, whereas companies with ATM oproducts only evangelise ATM.
Hi Skeptic, Equant had bought some Cosine box which is a Virtual Router based IP VPN box. Perhaps, the truth about MPLS RFC 2547 VPN being scalable and suitable for large scale deployment is begining to come up..
re: No VOIP for PanamaHi Skeptic, One of the greatest hype/BS/illusion that MPLS is selling is that by doing TE with MPLS, IP do not have to be over-provisioned and drive the ISP into chapter 11. It seem a great irony to me that one of the greatest MPLS flag waver out there is actually admitting that MPLS TE is not worth the trouble. MPLS had failed in its original mission..
re: No VOIP for PanamaHi Skeptic, There is two ways to look at this.. Is it they won't because they choose to or they won't because is it too hard or it is near impossible??
re: No VOIP for PanamaIf MPLS Traffic Engineering is really good, why Equant (one of the leading MPLS SP) choose not to implement that.. -------------------- Because, based on the article, they dont even seem to understand what MPLS TE is for and they believe that over-provisioning (and buying more routers) is the universal solution to all network problems. When their network reaches a certain size, they will run into the problems everyone else hits and then change their minds. ---------------- (from the article) So rather than do traffic engineering in the core, Equant prefers to over-provision backbone capacity. ----------------
re: No VOIP for PanamaYour comment: Hi Packet Man, If MPLS Traffic Engineering is really good, why Equant (one of the leading MPLS SP) choose not to implement that..
My comment: I have never mentioned MPLS once in any of my posts. I know very little about it and thus offer no opinions. I just keep defending IP QoS from all these "ATM only, IP QoS no work" flag wavers. Please do not change the subject, although we I guess we have all ignored the true subject of this post.
re: No VOIP for PanamaYour comment: Please show us an example how to use IP DSCP QOS to setup end to end QOS???
My comment: I gave two examples already. Once in this article I gave an example of how I used DSCP to ensure packets containing voice got through the network that had WAN links running at 96%, with FTP taking the performance hit. We ran a 20 minute test with 6 phones calls up on a network with 768Kbps WAN links. Our DMS toll staff who provided the ears, gave two thumbs up. Also in a previous post a few weeks ago I actually posted some sample config from one of my routers doing IP DSCP QoS.
Your comment: If you have an ATM switch that is routing capable, you could always mapped IP DSCP to a particular ATM VCC and guaranteed end to end QOS for an IP application.
My comment: Cool! I didn't know you could do that. But you still need the IP QoS to interface with the applications! I guess what I should of been saying is that IP QoS is well rounded enough and mature enough now, that ATM is no longer needed. I'm sure ATM and IP QOS can solve the same problem, there are usually two ways to skin a cat. I just can't stand to hear someone state you got to have ATM QoS because IP QoS does not work, because frankly its not true. Maybe three or fours years ago, but certainly not today. ATM is dead.
So rather than keep challenging me to give more and more examples, how about once and for all if someone from the "pure ATM only" arena give me an example of where ATM QoS solved a problem that todays IP QoS could not.
Man, I'm starting to sound like a broken record.
Hey LR editiors: Has someone done this in a lab and produced a report? If so please post so either me or the ATMers can keep quiet.
re: No VOIP for PanamaHi Packet Man, If MPLS Traffic Engineering is really good, why Equant (one of the leading MPLS SP) choose not to implement that.. http://www.telecommagazine.com...
re: No VOIP for PanamaHi Packet Man, Please show us an example how to use IP DSCP QOS to setup end to end QOS??? If you have an ATM switch that is routing capable, you could always mapped IP DSCP to a particular ATM VCC and guaranteed end to end QOS for an IP application.
re: No VOIP for PanamaEither a troll or a joke, right? Quick answer: ATM has End-to-end QoS (Service Classes, Traffic Contracts, Connection Admission Control...) while IP has DSCP (if you're lucky) with per-hop management of forwarding behavior.
No I wasn't joking. ATM has end-to-end QoS only at Layer 2. ATM does not provide QoS to the application layer, unless the application itself has been specifically written to work with ATM. 99% of the software being written is being written to work with IP not ATM. So you still need IP in there.
With that said, I am still not sure how one can say ATM is better than IP DSCP QoS? If IP DSCP QoS can gaurentee (which it does, you just have to set it up) end-to-end QoS (done on a per hop basis I agree), I've yet to see how ATM is better than IP QoS. I agree there might be additional network engineering required in IP QoS, to make it work right, but ATM is no walk in the park either.
I've worked on very large ATM networks, and was a pro-ATM tech head myself a couple of years ago. However since changing jobs and becoming very saturated with the new IP QoS thats available I am pretty much converted. I like ATM, I think part of my 'issue' is overlaying a packet-switched network onto a packet-switched network. I preferr the KISS (keep it simple stupid) rule and use IP/PPP.
I am still waiting for someone to give me a concrete example where ATM QoS provided a solution where the newer IP QoS tools could not.
Can you elaborate on how the QoS is better in ATM? How is ATM QoS better than IP QoS? Please tell me.
Either a troll or a joke, right?
Quick answer: ATM has End-to-end QoS (Service Classes, Traffic Contracts, Connection Admission Control...) while IP has DSCP (if you're lucky) with per-hop management of forwarding behavior.
re: No VOIP for Panamawhoever trys to do that will go bankrupt sooner than you can see the visible QoS public Internet...... There is no free lunch....
Why do you say that? I can see it being hard for all the ISPs of the world to come to an agreement on some sort of universal IP QoS plan, but certainly not impossible. The framework is in place. Its called Differentiated Services Code Point (DSCP). There are 6 bits used thus 64 classes of service can be defined. The IETF has already created a framework called Per Hop Behavior (PHB). For example the PHB called EF or DSCP 46. Using that frame work the Tier 1 Backbone players could come up with some sort of universial Internet QoS plan. They could then push that plan out the the Tier 2 and Tier 3 Players.
Why do you think implementing IP QoS on the Internet would be a financial burden on the ISPs?
re: No VOIP for PanamaLucifer scirbed : F" or a space-fller topic like "No VoIP for Panama", this board is getting very contentious and cantankerous. I don't understand why there is such strongly held opinion on VoIP by Optical and IP guys"
-LR Editorial guys : you've already got a wireless, optical, storage site: I reckon there's a good case for a Voice site also ?
re: No VOIP for PanamaYou're changing the subject ..... you kept telling us that VoATM QOS is better than VoIP QOS. I asked you who even makes VoATM applications, and I think I even said not VoIPoATM.
Regardless, cool phone ! Can you give me some WWW links as to where I can find more info on that phone? I am interested.
re: No VOIP for Panama" Can you elaborate on how the QoS is better in ATM? How is ATM QoS better than IP QoS? Please tell me. "
ATM has better QoS since it has cell tax, otherwise ATM would have been dead long time ago. Such cell tax is making ATM QoS predictable in terms of traffic shaping,policing,bandwidth calcaulation,jitter,latency control..... all the mathematics.
re: No VOIP for Panamadreamer101 wrote: you will see that you can get good QOS and lower bandwidth with VoATM over AAL2 than VoIP.. There is a table in the document showing the bandwidth usage with each option..
My comment: Hmmm nice article, not bad actually. Just one thing though. Who makes a VoATM telephone set? Who makes Softphones that interface at ATM? Who makes Telephony application software that intergrates directly into the ATM later? I've only heard of VoIP sets! Which means if you wanna do Voice over Packet (the network could be enterprise or the Internet) and use these cool new IP sets you will need to do VoIP anyway. If your network is ATM based you will have to do VoIPoATM, and you end up wasting way more bandwidth than pure VoIP. I suppose you could keep buying those antique analog phones and use VoATM gateways. But half the fun is the cool features that come with the IP sets. Stuff like LDAP, DNS, HTTP/XML, MobileIP, IP Voice Mail, IP Auto-Attendant, IP Centrex, etc is all being incorpated into....you guessed it.....VoIP features. ATM is dead.
If you want to be able to take advantage of Voice over Packet in every way possible, pure VoIP end-to-end (ATM not needed) is the only way to go.
Yes I know there is little or no IP QoS on the public Internet today, but that is starting to change too.
If a carrier has a large deployment of ATM then I can see it making business sense, otherwise IP. Even some carriers who have large ATM networks are starting to announce a migration plan to IP only.
Hi All, Choose the first document in this search and you will see that you can get good QOS and lower bandwidth with VoATM over AAL2 than VoIP.. There is a table in the document showing the bandwidth usage with each option..
re: No VOIP for PanamaFor a space-fller topic like "No VoIP for Panama", this board is getting very contentious and cantankerous. I don't understand why there is such strongly held opinion on VoIP by Optical and IP guys, although I suspect it has to do with the fact that Voice over IP engineering requires knowledge of voice and IP, two skills rarely present in one person, and the fact that to implement VoIP in the WAN is difficult -- let no-one tell you otherwise -- and this does make JC look a little foolish with his comment -- build it for IP and voice travels for free...but I digress.
For one point of view with some facts and based on experience, look no further than:
I did read the boards on dslreports.com before I ordered the service a few months ago to see if Vonage was on the up-and-up. I just surfed there now, and I do see some complaints related to a service outage. So far I haven't been affected in NJ, though I did get an e-mail from them giving me a small rebate (I think it was $10 bucks) because of an outage.
In anycase, I have a cellphone for backup. I'm still a big fan of the service.
I've been using residential VoIP service in my apt. for approximately four months now through Vonage, and I can tell you I couldn't be happier with the service. I got sick of having to pay >$40/mo. for a Verizon landline which I hardly used on top of long distance charges. Now I pay $22/mo. for all my local and long distance calls plus I get call waiting, caller ID, call forwarding, LNP, voicemail, a call log, and can even listen to my voicemails on the Internet.
The quality of the calls is slightly less than toll quality, but to most people I talk to, the difference is indiscernible. In fact, most people are shocked when I tell them I use IP phone service.
I'm not a telecomm person, so I can't tell you how they employ QOS (if at all), but I can tell you it works...well. The only requirement is that you have a broadband connection with 90kpbs of upstream bandwidth and a router. Vonage supplied me with the Cisco voice router that sits on top. Set up was a snap.
The best part is that I don't have to keep changing my phone number when I move. I've moved four times in the last five years and changing phone service/numbers is a pain to say the least, and, even though I live in NJ, I got a Manhattan area code for my number since that's where most of my friends/family live.
Verizon take heed because when the word gets out how well this works, it's game over for the incumbents. For all the naysayers, just do a search on the web for "Vonage" and see if you can find a post from one disgruntled customer.
re: No VOIP for PanamaThis math is open for discussion:
1.544Mbps / 11.2Kbps (G.729 on cRTP VoIP) = 137 calls.
1.544Mbps / 26.4 Kbps (G.729 on VoIP with no cRTP) = 58 calls.
1.544Mbps / 67.2Kbps (G.711 on VoIP with cRTP) = 22 calls.
1.544 Mbps / 82.4Kbps (G.711 on VoIP no cRTP) = 18 calls.
This is based on 0% being used for data and video.
In large scale networks VAD will increase the numbers of calls, but can make planning trickery. Also one should take away 20 to 25% of the WAN circuit (in this case 1.544Mbps) to allow for routing protocol overhead, etc. Monitor the network and them maybe bring that number down if safe to do so. A small remote office could use static routes and thus free up bandwidth to get another call or two plus some for data.
re: No VOIP for PanamaIP overhead is less than ATM overhead, so how can you say VoIP is 2X when compared to ATM? VoIPoATM would be wastefull. G.711 VoIPoPPP is 82.4Kbps, and G.711 VoIPoATM is 106Kbps! G.729 is 26.4 / 42.4 respectivly. Toss in cRTP and the numbers become G.711 67.2K and G.729 11.2K for VoIP.
Trunk utilization needs to be less than 40% to control jitter? Says who? I did a stress test for a project and had the WAN link at 94% utilization, and that was on a 768Kbs with 30 simulteanous FTP sessions, and a couple of Citrix streams going across the same link. I had six G.711 voice calls up and all sounded perfect for the 20 minute test. I also used staff from our DMS100 toll group to provide the ears for the calls and got two thumbs up.
Mind you the FTP wasn't the fastest, but that how the customer wants it.
re: No VOIP for Panamadreamer101 wrote: I can get more phone calls on a T1 line and better QOS with VoFR..
...actually, you can get only 24 simultaneous phone calls only at one time on your T1 (30 if you had the sense to configure it as an E1 and use those 6 extra channels)........can anyone IP-clueful calculate what that 1.5/2MB bandwidth would get you in terms of simultaneous VoIP calls for comparable quality (this should bring some interesting answers - BobbyMax, where are you when your logic is needed ?)
re: No VOIP for PanamaHi Packet Man, You could make anything to work if you waste enough bandwidth on it.. If you have direct fibre between two phones, you can run any protocols and it will work.. Basic questions, how much bandwidth it take to deliver a phone calls?? If it is VoATM, it is X.. If it is VoIP, it is 2X due packet size.. The trunk utilization need to be less than 40% to control jitter. Hence, it is another 2.5 times.. The bottomline, it is 5X.. As a carrier, unless you get bandwidth for free.. Why are you using 5X bandwidth to deliver a phone call with lower QOS than VoATM or TDM?? It is commodity services.. For enterprise, VoIP on LAN make sense.. VoIP on WAN -> You must be kidding.. I can get more phone calls on a T1 line and better QOS with VoFR..
re: No VOIP for PanamaYour comment: It seems very typical that a Canadian carrier would - true to form - buy from a Canadian equipment provider (let's keep those loonies flying in tight formation, eh?). Also not unusual to have a legacy ATM core (hence the Nortel Passport). Finally, I can't see them ignoring the eventuality of VOIP. Seems like the usual progression. Am I off-base?
Canadian carrier also had a bunch of Voice over packet (IP and/or ATM) Softswitches made by a company called Telica out of Mass.
I'm not sure if they are still using it or if they sent it back. But still maybe ATM today, it will be VoIPoATM tommorrow, and then VoIP only later. Where MPLS fits into this I can't comment.
re: No VOIP for PanamaThanks to wayland_smithy, I learned a lot of real experiences about VoIP.
More discussions follows:
"Deployments: ...For QoS, you also have to bear in mind that TDM exchanges are NOT engineered for n->n circuits across the network: blocking is introduced not only in the network, but also by the equipment..."
Here, I think QoS and blocking are two different things. When I make a call, it failed. Maybe it is because somewhere busy, blocked. This kind of blocking is acceptable. People will try other method to contact each other. What QoS mean? When I make a call, yes, connected. But there is bunch of noise, and not stable, even down suddenly. Then, try again, same thing happened, connected, quality is not good, down occasionally. This is a QoS issue. QoS issue is unacceptable, and will driving you mad.
"Benefits : It could work if one deployed a complete private network, but who will deploy this sort of network on an international basis - this is why VPNs were introduced So for the future of VoIP, the selling point will be features such as mobility and the ease of provisioning/managed services rather than pure transmission cost savings"
Do agree with you at this point. The other word, if VoIP only fit for VPN, that is to say, for Service Provider, there is really no need to implement VoIP. For enterprises, who have their own VPN, VoIP will be a pretty good solution.
But for mobility, it's not that simple as you said. If you roaming freely, without any extra pay, that means there must sb else pay for that. This must be SP. If so, why SP stupidly provide such service? Think about wireless phone, everybody will pay for roaming. The only reason why you don't need to pay for roaming in VoIP is, SP can't control or can't accouting on your roaming. This is another evidence to show that SP should not deploying VoIP solution.
Finally, if VoIP can't saving cost of transportation, then what it can take to SP? How can these SP who deloyed VoIP earn money? NoWay!! So, if you are a service provider, and want to make money, don't using VoIP solution!
re: No VOIP for PanamaYour comment: Packet Man, your faith in VoIP is moving... I really hope that it's supported by solid technical basis, and it's not son of the love for new stuff's love.
My comment: All I ever wanted to bring to the table is to set the record straight in regards to two of the most inaccurate statements that I keep hearing over and over. (1) IP cannot deliver QoS. That is piece of crap, IP QoS is becoming very mature and works very well if set up right, with the right gear. (2) You need ATM to deliver QoS. Also a piece of crap. You do not need ATM. If I were to be involved in a greenfiled carrier network, I would have IP/PPP/Sonet, not one shread of ATM. ATM is dead. Now if you are a major carrier and gots lots of existing ATM then sure there is sound business sense to put voice on ATM. I won't comment on MPLS cause I know almost nothing about it. If anyone tells you that you NEED ATM to make Voice over packet work they probably represent a ATM division of some vendor.
Let face it. IP voice is solid, if engineered right. If any of you say you did it and the quality sucked then you did it wrong. I was challenged in a previous article that I knew nothing about anything, didn't even know what a router was. So I had to post a sample config of one of my routers with IP QOS using DSCP to shut him/her up. There are dozen of very large companies who have put their entire business on IP, and that includes voice, and are reaping the benefits.
Now there are some 10-10-xxx VoIP Internet players out there. I'm willing to bet that if you made a call on some of them, during the right time of day, maybe to a certain place in the world, the voice quality will be less than perfect. Why? Cause the Internet as a whole does not have QoS and peering points can often congested. But do the customers care? BIG FAT NO. All they care about is the big CHEAPO calls they can make. So for those of you who say they made a call to some overseas land and the voice quality sucked, I say so what? You can call through the PSTN to certain parts of the world and get crap voice quailty too, and pay anywhere from $.10/minute to $.50/minute, instead of the $.03/minute with VoIP.
I am coming to one conclusion thought. A lot of the arguement put forth against VoIP has been from two fronts. (1) People with ATM interests. (2) People with traditional legacy Telco interests.
I suspect alot of the big telcos fear VoIP simply because they will lose major major revenue in the near few years to come. I foresee serious problems in the voice-switching side of the buiness. Transport will always be reasonalbly good buiness because all the Internet circuits have to get carried. With the new slew of devices/software coming out it will be getting very easy for organizations (business, families, etc) to create virtual "dial plans" to link there contacts. They will need little services from a Telco/ISP to help them set up their own calls on the Internet. If govt/industry steps in to block IP TCP/UDP ports then people simply use diffent ports. Much like the music industry. And if they start blocking all the ports then we might as well turn of the Internet, and all go back into caves.
I encourage people to put forth opinions that VoIP does not work. Its all about healthy discussion. I challenge the Nay sayers to give us examples of where a carrier or large enterprise recently tried pure VoIP and failed. VoIP five years ago does not count, IP QoS was pretty much not possible then. Bet ya won't find one case study where it could not work. Now go out and find examples of where pure VoIP has been implemented with success. You will find more than you can shake a stick at.
Some good points here, and I agree with some of them. Let me come back on a couple :
> Here, I think QoS and blocking are two different things:
Yes, agree completely. I guess I was making a point that, on TDM circuits, let's not assume that every call originated gets through at 100% quality. Some calls are blocked, other voice calls can sound like your head is buried in a tinbucket if, for example, compression/cancellors are used. Also, bandwidth is "nailed up" for TDM for the life of that call - inefficient but dedicated. With IP, QOS is a different proposition. What it effectivley allows you to do is to regulate the quality of the service versus bandwidth across a network: a completely different way of addressing the problem. You could, therefore, go for a high QOS where you assign higher priority (~ bandwidth) to voice at the expense of other traffic types. QoS could essentially be used to trade off number of calls/bandwidth usage versus quality (that comment will get some posts !!)so you could argue you can engineer a network to use bandwidth more effectively at the expense of "perceived voice quality". IP makes the deployment a little more flexible in its nature.
>>>But for mobility, it's not that simple as you said. If you roaming freely, without any extra pay, that means there must sb else pay for that. This must be SP. If so, why SP stupidly provide such service ?
Again, a fair point. With IP Telephony, however, the assumption is that many Enterprises already have their own IP Data Network, and are carrying the call to the endpoint as far as possible over this IP network. In this case, the SP will never see the traffic. If it breaks out the voice to the PSTN, the SP will see a call originated on thier network as close as possible to the destination phone: this means reduced revenues anyway. This is not dissimilar to doing this on a private voice TDM network using Least Cost routing - keeping the voice call on your own network as far as possible before breaking out over the PSTN to minmise call payments to the SP.
However, with IPT the argued advantage is that you can carry telephony over the data network, as opposed to having either 2 separate networks (the voice one of which is inefficiently used since the bandwidth is only utilised for the duration of the call) or having a voice/SDH/SONET network over which 90% of the traffic is typically data & therefore also inefficiently handled. There is also the cost of managing the network - when I worked at Nortel, one of the biggest problems we saw for carriers offering managed services (Centrex) was the cost & time of Add/move/changes on the networks requested by the end users: if this "mobility" exists, costs of managing the network decreases a lot since you won't be moving phones or phone numbers around manually.
For the SP revenue side, you are correct: SPs will now have to start moving away from offering raw bandwidth circuits & voice minutes to moving up the service revenue ladder. So, to do that, they will need to minimise infrastructure management costs & start looking at service revenue generation over that infrastructure, whilst reducing the cost of managing the network.
This has happened to an extent in the public voiceinfrastructure - whereas 10 years ago, your 56K/64K phone line to your house delivered only dialtone (and in the US your SP didn't even get revenue for local calls),but now, on that same piece of copper, SPs can offer voicemail, call forward, do not disturb, etc. and such features generate additional revenue along that same piece of cable - this moves them up the value chain and increases revenue for the same chunk of bandwidth. If the same argument now applies, for example, to delivering an Ethernet pipe to the home/business, IP Telephony, VideoIP and Data/Internet services all could generate additional revenue for that same piece of copper/fibre (this isthe trend you are seeing in Europe now with newer SPs or CLECs). It also allows service differentiation to other competitors, whereas currently differentiation tends to be on a price per circuit basis for a limited number of services., and this leads to commoditisation.
So, what does this mean ? VoIP should indeed be cheaper to deploy over time for Enterprises initially, and then SPs (one IP/MPLS/ATM network for voice & data, reducing management & maintenance staff costs & getting rid of inefficient voice networks carrying 90% data) with additional services enabled on the same infrastructure to deploy revenue-generating services....well, that is the theory anyway. Over time, if it does take off, Enterprises will less likely be asking for an SDH/SONET tail circuit to the premise (since they won't have a PBX & it doesn't makesense to carry IP over SONET: and the price of a SONET interface for their router is 2-5 the cost of an Ethernet interface) and more likely to ask for a "big fat IP/Ethernet pipe with VPN/guaranteed QOS/VPLAN" type functions. They are also possibly more likely to take provisioned/managed services from the SP since "mobility" makes it more cost effective for both parties to offer/taker these services.
Where I;m not yet convinced, however,is the argument where you are simply replacing a TDM 2MBcircuit interface on a PBX for a packet interface on a PBX as a "protocol converter" - the key to VoIP telephony is in the services & mobility across a very wide area network and not a PBX node/cluster.
re: No VOIP for Panama"In TDM world, QoS is guaranteed because you only have so many time slots to assign and when you fill up and run out, the n+1 caller gets fast busy - no trunks available.
In ATM or IP or MPLS, we essentially have to have an equivalent or replacement for that fast busy to deny originations. Is this not accomplished by ensuring that at the TDM to packet gateway we can deny the n+1 caller from ingressing into the Traffic engineered ATM/IP/MPLS pipe. Doesn't this provide the true QoS that you speak of regarless of the transport layer?"
My Answer:
I think you are talking about Admission Control. Yes, this is one of many key points to support QoS. If we can not deny as your said the n+1 call, every other call will be affected, because lack of BW after the n+1 join in.
To accomplish Admission Control, the AC's target must be a connection. For IP, you can't do admission control, because there is no connection. The only thing you can do in IP is per packet, through ACL(Access Control List, which is the best feather to a pre-sale engineer and nightmare to post-sale engineer. Sale it, nerver using it.)
Then, how can you know when you should deny the connection? In TDM, it's easy to achieve, there is no time slot. In packet network, there must be someone auditing on the usage of network resource. That is Routing protocol. For ATM, it's PNNI, which is both routing protocol and signalling; for MPLS, it's IGP-te,BGP-TE and rsvp-TE etc.
re: No VOIP for PanamaPacket Man, your faith in VoIP is moving... I really hope that it's supported by solid technical basis, and it's not son of the love for new stuff's love.
Let's assume that you speak after due consideration, OK, but please don't bring to the discussion arguments like "My comment: Do you honestly think that these major carriers all over the world would deploy this is it did not work?"
Have you ever noticed the amount of rubbish that even large carriers have deployed through the years for the most bizarre reasons?
So, if you tell me that VoIP can work in complex and 'crowded' networks, I can even consider this as a valid standpoint, if not right now, say, in a few years from now, but please don't use flimsy conjectures selling them for convincing evidences.
Deployments: Telecom Italia announced last week the replacement of the Class4 & some class 5 trunking network over the next 2 years by VoIP technology, replacing the legacy TDM exchanges countrywide. I think this is the biggest deployment from a sizeable carrier, although in Europe a number of other carriers use VoIP mainly for transit traffic (eg Tele2). I'd suggest that if TI are confident enough to deploy this, then there should be no reason for other major carriers to do so...in theory. For QoS, you also have to bear in mind that TDM exchanges are NOT engineered for n->n circuits across the network: blocking is introduced not only in the network, but also by the equipment vendors in the design of the TDM exchanges (remember Erlang ?). So, as an example, a typical Class5 switch in N. America will deploy resential lines to a ratio of trunking backhaul circuits at a ratio of around 7-10:1. In fact, for those of you conversant with Nortel's DMS100, you'll know a line/trunking peripheral can support 480 backhaul circuits (Intl), but will be provisioned with up to 8x640 (5120) residential lines on the Line cluster modules, hence blocking can be easily introduced at a factor above 10:1.
Benefits : For VoIP, there are two strategies to attack the market: firstly, cost savings on the trunking network (either by utilising existing/combined IP/ATM infrastructures OR by using toll bypass internationally) but, secondly,the real value is to the Carrier & Enterprise user in the areas of feature deployment and mobility. For example, I'm currently sitting in an international office and receiving calls on my "softphone" via my PC, whose PSTN number is assigned to me in a different country. If I don't like the PC Softphone, I can just log into one of the IP phones on the desk and calls to my home number are routed automatically to this phone across international boundaries (and, yes, the carrier will lose money on this !) This mobility allows customers to reach me anywhere I am connected to a LAN on one number. This would not be possibly on TDM infrastructures (god knows, we tried to sell PSTN Intelligent Networking for years at Nortel !) but the main reason is not necessarily technical, but that different countries do not allow interworking to take place of features or telephony/E.164 numbering plans - this is particularly difficult outside of the US also, when signalling systems for each countrty are non-standardised across international boundaries. It could work if one deployed a complete private network, but who will deploy this sort of network on an international basis - this is why VPNs were introduced. So for the future of VoIP, the selling point will be features such as mobility and the ease of provisioning/managed services rather than pure transmission cost savings.
Banning VoIP : this is more prevolent thsn you think. A couple of years ago, I did some extensive research into deploying VoIP systems in Europe. What exists in many countries in Europe are restrictions that "indirectly" also lead to a ban on VoIP commericalisation for carriers. For example, countries like Denmark & Germany insist on network "break-in/break-out" round trip delays of <35ms for transit - difficult on TDM networks, almost impossible on VoIP networks (the standard for GSM is (or used to be) 125ms). Other restrictions such as, in Germany, every voice call has to be capable of being monitored also causes implementation issues & issues of "political" acceptance....so the uptake of VoIP commercially will depend on a number of factors beyond QoS and voice quality (which, I believe on a personal note, is not an issue on a well engineered VoIP network: I have worked on TDM systems for 20+ years and there are as many "ropey" TDM networks as there are VoiP). Having recently been placed on an IP PBX system, however, I can say that the future of voice will be in this technology (particularly in the Enterprise) - I have had more useful mobility features enabled in the last 3 months on the IP system than we could ever have designed on traditional PBXs, and with no perceivable drop in voice quality. For the TDM heads out there, my suggestion would be to try it before you knock it - take this from a converted TDM skeptic !
re: No VOIP for PanamaIn TDM world, QoS is guaranteed because you only have so many time slots to assign and when you fill up and run out, the n+1 caller gets fast busy - no trunks available.
In ATM or IP or MPLS, we essentially have to have an equivalent or replacement for that fast busy to deny originations. Is this not accomplished by ensuring that at the TDM to packet gateway we can deny the n+1 caller from ingressing into the Traffic engineered ATM/IP/MPLS pipe. Doesn't this provide the true QoS that you speak of regarless of the transport layer?
The biggest VoIP service I have heard is in China. The former China Telecom provide VoIP service in nation wide. But this service was proved to be not true VoIP later, except the price is same as VoIP's. They just using their legacy PSTN pretend to be VoIP. And, this is the best VoIP service I have ever tried, only because it's not true VoIP, it is just PSTN. What other SP's VoIP, say China Unicom and JiTong and former CNC? Both Jitong and former CNC bandrupted and let China Telecom pay the bill. China Unicom,eh...focusing on wireless only.
Before there is a way to make it real work, never waste money on VoIP.
re: No VOIP for PanamaSo does anyone know who has actually implemented VoIP in any big way? I gather from the trade press that there is quite a bit of it in europe and in some large enterprises. I have also heard that there is also a number of the toll bypass players who have also been doing it for a while? Most of the stuff I have seen sounds strictly like a trunking application as opposed to a line side access.
I suspect that the people who have done it believe that they are enjoying a cost advantage and dont't want to advertise it too heavily because the service should be transparent to the end user and what's the point of alerting your competition to your cost advantage.
I speculate that this will likely change as more VoIP gets moved in to the line side and naturally the focus becomes not on lower cost networks but new revenue generating features that become enabled by VoIP. That's the marketing speil I hear anyway.
Any opinions on how important VoIP is to the Cisco's and Junipers? Being relatively small traffic generators, I wonder how much of an opportunity it can be for systems sold on the basis of cost per bit capability. It's hard to figure how VoIP is going to be a killer app for them unless the shift goes from capacity to features.
re: No VOIP for PanamaWhen we talk about VoIP's quality, actually we are not talking about IP. IP itself can't provide any quality guarantee. If we want VoIP have something quality, we must find some right carrier of IP, and rely on this carrier to provide QoS guarantee.
I believe, the guys who support that VoIP have toll quality, will give out a lot of evidences. If we take look insight these evidences, you will find that they are not of IP's, they are all of a lower carrier's, say, MPLS, which is layer2, ATM, which is also layer2, FR,layer2; PPP, etc.(Upper layer, TCP's flow control not fit for real time applications, for on-line tractoin applications, it's ok; then UDP is the only choice, but UDP even have no flow contral. And, you will never expect the immediate node are TCP/UDP awared. Then I say lower layer is the only direction where you find QoS support from)
Then, the question changed to, who is QoS ready for VoIP NOW, MPLS or ATM? who have the capability to provide an end-to-end QoS guarantee, again, NOW, MPLS or ATM? This time, I believe the answer is quite easy, it is ATM. Maybe MPLS will, in the who knows future, provide QoS, but absolutely not now.
Further more, I am doubting about MPLS's QoS. In my opinion, MPLS can only support CoS, not QoS. Remember FEC, what the C stands for? Class!!! That means all packets in same class should belong to same FEC and share one LSP. If somebody implement MPLS QoS, he will break this law. He must assign every single call a seperate end-to-end LSP. Every call in different FEC, this is not MPLS want and can do. And if so, ATM's PVC/SVC is really better than MPLS now and future.
Now, it's easy to understand that practical VoIP solutions are all based on ATM backbone; also it's right, that so many people doubting about VoIP's quality, when VoIP running on IP/MPLS network.
re: No VOIP for PanamaIt makes a lot of sense to ban VOIP. It protects the franchise of companies that pay *hard cash* for a voice concession. __________________
And exactly why should a voice concessionaire be protected?
re: No VOIP for PanamaGÇ£IGÇÖm not sure how theyGÇÖre technically accomplishing this,GÇ¥ Frost & Sullivan's Popova says. GÇ£I think that carriers and ISPs in general have a variety of ways of bypassing regulations.GÇ¥
I disagree. It's pretty easy to rip a gateway out of a colo space. The remedies she is talking about are for high school hackers, not companies with profitability targets.
re: No VOIP for PanamaBooby... If I thought you were not very nearly a negativity Robot, I'd almost think you were imitating me imitating YOU! With this post you have sunk to new lows, and you sound like a parody of yourself.
Of course VoIP can provide toll quality service. It's actually used intra-company by many big compaines. It's highly likely you've used it many times and don't know it.
By the way, we never heard your response to your article..YOU'RE FAMOUS! And, I see, still "number 1" on most people's hit parade list!...(oh yeah, before I forget let me give you a 1 too, not that it matters anymore.)
re: No VOIP for Panamaa) It makes a lot of sense to ban VOIP. It protects the franchise of companies that pay *hard cash* for a voice concession.
b) you don't need ATM for toll grade voice service over VOIP. You can run toll grade VOIP over Cisco routers, but you need to provision enough bandwidth and you need to own the network equipment end-to-end. VOIP falls apart when you hand off to third party IP networks without their own QOS.
re: No VOIP for PanamaVoIP can provide toll-quality voice; it's just very difficult to engineer the IP backbone without using ATM to provide QoS and connection restoration...
I don't know what Telus is actually using but I do know the following:
1. Passport 15K is multiservice. ATM and MPLS NNI. 2. The Packet Voice Gateway supports VoATM as well as VoIPoATM, with and VoIP (over GigE)promised "real-soon, really!"
My guess -- Telus is hedging their bets, marketing VoIP (because Cisco cannot be wrong, can they???) and using ATM as the Layer 2 so they can actually build and operate a quality service.
What does have to do with a small poor country that has outrageously high international voice tariffs and an antedeluvian attitude to regulation...
re: No VOIP for Panama LR editors wrote an article about your posts? We can debate technical ignorance if we can decipher the point of your post. The cracked grammar and mis-spellings combined with the sweeping generalizations leave whatever point you are attempting to make unstated.
re: No VOIP for PanamaI'm not sure if there is a contradiction.
It seems very typical that a Canadian carrier would - true to form - buy from a Canadian equipment provider (let's keep those loonies flying in tight formation, eh?). Also not unusual to have a legacy ATM core (hence the Nortel Passport). Finally, I can't see them ignoring the eventuality of VOIP. Seems like the usual progression. Am I off-base?
As far as Panama goes, I believe that - for obvious reasons in that neck of the world - they are understandably suspicious of "packets".
re: No VOIP for PanamaHi Packetman, Nortel Passport only do Voice over ATM now... Nortel promised that Passport will do VoIP some time in the future...
Here is a sample of the article: RESEARCH TRIANGLE PARK, NORTH.CAROLINA--Nortel Networks (NYSE:NT)(TSX:NT) announced today that TELUS, one of Canada's largest telecommunications companies, is migrating its national circuit-based long distance network traffic to a packet-based network with Nortel Networks voice over IP (VoIP) equipment.
At the heart of TELUS' new packet-based national voice solution are Nortel Networks Succession Communication Server 2000 softswitches and Nortel Networks Passport Packet Voice Gateways (PVG). TELUS' national VoIP network includes voice gateways deployed at 11 major locations across Canada, enabling voice traffic to be carried on TELUS' national high-speed IP backbone with carrier grade voice quality.
My comment: Do you honestly think that these major carriers all over the world would deploy this is it did not work?
re: No VOIP for Panama"At the heart of TELUS' new packet-based national voice solution are Nortel Networks Succession Communication Server 2000 softswitches and Nortel Networks Passport Packet Voice Gateways(PVG)."
Packet Man wrote: "My comment: Do you honestly think that these major carriers all over the world would deploy this is it did not work?"
------------------------------
My comment: How many of you people are aware that the Passport is an ATM switch? What does that say about Telus' confidence in VoIP?
re: No VOIP for PanamaIt is well known that VoIP cannot provide tolll quality service. Some companies like Cisco, with no experience, in voice telephony service. The UIS VoIP equipment vendors have been targeting third world countries such as India, China, have been to sell their VoIP equipment.
The reson that the US regulators have not banned the VoIP service in the US because it hurt business abroad. They want the US companies to go outside and sell as much as they can. I am positive will fireback soner or later.
Lyse Tele, a power utility company that serves 110,000 customers in southwest Norway, is deploying direct fiber connections to residential customers in the Stavanger region of the country using a Cisco Metro Ethernet network. Lyse has connected 500 subscribers in the initial phase of the network rollout and is offering flat rate VoIP packages, video on demand, access to 40 channels of broadcast TV over a single IP-based Ethernet network. The deployment includes Cisco Catalyst 6509 switches in the core, Catalyst 4006 switches in the Ethernet distribution network, the Cisco AS5350 Universal Gateway, and Cisco 3640 Internet routers. Financial terms were not disclosed.