I'm researching IPOffice for my company and it was unclear to me whether the voice traffic is transmitted using IP or not. Or is it basically a PBX with IP capabilities?
If its via any of the RJ45 ports (even the one marked WAN on the Small Office) then all the traffic is I assume Ethernet based and a two-way call is going to consume all the bandwidth. And if your using VPN whilst I don't know the extra headers required but its going to take the call well beyond 64K.
Also the IP Office isn't going to do any QoS prioritization for you. Even without any data and control packets you close to/probably over the edge. Switching to G.723 would give a bit more elbow room but still not enough.
For such a thin pipe (I didn't know you could get such low bandwidth DSL) you really need to be using the rear WAN port on the control unit or a WAN3 module. In that case a 64K pipe should give you bandwidth for 4 calls G729/6 calls G723.
IP Office connected to our network via a 3com Layer 3 QoS switch.
IP Office is connect to DSL via a Watchguard Vclass router with QoS. This is what was recommended.
I have confirmed with our telco that all that is available for DSL in our area is 640kbps upstream, although we can have our downstream increased(fat lot of good that does us). We were assured by the company putting our VOIP in that our DSL would be fine but I'm starting to wonder if anybody did the math on this.
Our only other options are getting something other than DSL and management is cringing at the costs for this(T1 and Fibre around here are around $1200 monthly).
The setup you describe should certainly be workable, though you need to get your telco to confirm that there network supports the same QoS settings at all points through which the call might be routed.
Otherwise the solution being adopted be most is to use VPN for the VoIP traffic with a guaranteed VPN bandwidth agreed with the provider.
My concern is whether it is workable in our situation.
Our telco stated that they cannot guarantee bandwidth or quality with our current connection(ADSL), which, although it is considered business service, I don't think is any better than basic residential service. The only other services they offer with guaranteed bandwidth are Frame, T1 or Fibre
We have 4 branch offices(each with an IP Office device). 2 offices have 6-9 staff each and 2 have 2-3 staff each. coming into our main office(with it's own IP Office device) and we are told we can setup least-cost routing, etc. We also want to route all calls going into outside offices through the main office and transfer back out(to have one receptionist). Of course, this would in effect double the traffic for a call(at least I think so).
No provider (that I've ever seen) has a Service Level Agreement for DSL whether it be residential or business, so there will never be a guarantee on bandwidth or quality. Do they not offer capped T1s in your area? Still won't be as cheap as any DSL, but reliable.
Will the IPO dynamically change the codec according to available bandwidth? In other words will it compress at a higher rate if it detects more calls are taking place? Kind of like a VCR that starts recording at SLP when it detects there isn't enough tape to continue recording SP.
I have seen some guranteed SLA for DSL but with the proviso that the start and end points where on the same ISP's network (it was in Holland). Telstra in Australia also seem pretty much on the ball from what I've seen of their documentation.
If you have the option of Frame Relay give it some firm consideration. I don't know how costs compare but with all other things the same, for VoIP the improved header handling (especially with header compression) makes frame relay 3 to 4 times more bandwidth efficient. You also get QoS priorization and large data packet fragmentation support from the IP Office.
mmount, it would not double the traffic to use a centralized receptionist taking calls then transfering them back to extensions on the switch the call came from. Once 'she' (excuse my total lack of PC) transfers the call to an extension on the switch terminating the PSTN call it ceases to be a voip call (unless your users have hardphones of course) and is routed directly through the TDM bus from the trunk to the digital set.
I have done ADSL to ADSL connections where both ends were provided by the same ISP (Toronto to Montreal) and it worked out great. Helped that it was a smaller company that was willing to ensure my customer was happy, knowing it would bring them more business.
Perhaps I am wrong but here is the situation as I understand it:
1.Call from PSTN comes into branch office.
2.Call is automatically forwarded, inside our VPN, using IPO, to our main office receptionist phone. This is all done within our network.
3.Receptionist in main office answers call. Turns out call is for a staff member in the branch office(as they dialed the number for the branch office).
4.Receptionist in main office(who uses softconsole) transfers call to staff member in branch office.
5.Call is sent back over the VPN to the staff member's extension in the branch office.
In the branch offices, the extentions are not directly connected to the IPO. We have too many users for that - they are connected to a 3com switch. The only digital set in the branch office is the receptionist. All other users have 4602SW IP phones. In the main office, all receptionists have digital phones. All other users have IP phones.
So are you saying that if the call was strictly limited to digital phones, there would be no issue?
I can confirm that voice quality on the IP phones dramatially improves if we have the branch office take it's own phone calls instead of having them go to the main office first.
Hmmmmmmm. I have verified with centralized AA that the call hairpins and no longer takes up bandwidth on the WAN after being transfered back on the orinating site, assumed it would be the same with a centralized receptionist. It should not matter that you are using IP hardphones in terms of WAN bandwidth, my comment on the phones was only to clarify my statement the call would no longer be voip. In your case it obviously will still be voip but SHOULDN'T consume any bandwidth on the WAN after the transfer is complete, let alone double.
Are the IP hardphones registered against the IPO at the branch? It is possible to tell them to register with the IPO at HQ but you don't want to do that...
Calls transfered back accros a link to the originatinng site DO NOT consume exctra bandwith or speach channels, the original channel will be disconnected as it is no longer required.
this is called Anti-tromboning (at least it is in the UK).
Yeah I've heard it called that also, more commonly referred to here as hairpinning. I am of the same opinion, it won't take up any chanells or bandwidth after transfer.
Just trying to figure out why mmount reports better quality when calls answered directly at branch than when transfered by central receptionist...??
What about codecs.. are you using the same codecs on the WAN as for the IP Hardphones? That would explain it.
I believe the codec is 729a between offices. I am not the person who installed the IPOs - a company did this for us, but I am the one who is left trying to figure out why our voice quality is so poor.
General user concensus is that it is better than it was when branches had calls transfered but many say it is still poor between offices.
The total number of all branch office users is 22-24(sometimes varies) in 4 seperate branch offices. The way our company works there can be alot of calls between offices.
If you are using g.729 between offices, trying setting the phones themselves to g.729 as well (you're likely using g.711 for the phones currently). Or better still, if you have enough WAN bandwidth set the inter-branch calls to be g.711 also. There is a voip bandwidth calculator included in the 2.1 docs CD - you can download it from support.avaya.com if you don't already have it.
How many VCM channels are provisioned in each of the branches? That's your upper limit for calls, and a good way to pick which codec you can use.
Codec actual bandwidth * VMC channels must be < upstream bandwidth.
As far as I can tell, there are 10 channels to each branch office(4 total branch offices).
Would this mean a total of 40 channels?
[There are also single channels listed for each PTSN line in the main office but I don't think this is pertinent to the situation]
Two branch connections are set to 729, the other two are set to automatic.
if 729 takes up 24k(or even 30k) with overhead(this is based on the post way up above by sizbut, and maybe I am misunderstanding how it works), then would it not take up 960kbps or even 1200kbps? We have a DSL link in the main office with a max upstream of 640kbps(I have verified this with our Telco).
Let me know if you think my math is faulty - grin.
I will check out the bandwidth calculator on Avaya's site.
Mike, math is right precept is wrong. 10 channels to each site does add to 40 if 4 sites, but that is irrelevant. Look at each branch individually. If I have a VCM 10 in one of my branches, then I can squeeze max of 10 voip calls down the pipe between that branch office and <wherever>. Therefore you are looking at max of 300k for voice calls at any given time, so you are fine with a max upstream of 640k but could not go to g.711. Do that analysis for each of the branches seperately.
2 suggestions - first, CHANGE the 2 branches that are automatic selection and force them g.729
Second, try setting one test phone at each of 2 branches to g.729 and make a test call, see if quality improves. The goal is to keep the call g.729 the whole way without translating from g.711 to g.729 then back to g.711 again.
One other tip - launch monitor, when it first connects you'll get a bunch of info including "vcm=xx" the xx is the number of vcm channels you have on a particular system.
You'll probably have more at the main site than the branches, but that doesn't change what I've said above.
In the main office, why wouldn't max be greater than 300k? Lots of calls would be going out to all branch offices.
Sorry if I'm not following you here. I'm just trying to get my head around this.
I can find nothing with "vcm" at all in the log file(I did a find within the monitor program and also saved the log and did a find in Notepad). We are on 2.1 now, I'm not sure if this makes any difference.
Simplest way is through the IP tab of the Extension settings in the IP Office configuration. Change the default from Auto to G.729a.
[Whilst we're there, the Auto setting is for the IP Office to try to auto-negotiate G.729a/723/711 in that order of preference with the IP phone. ?Does anyone know what the codec preferences of 4600s are, I know they supposed support G.729a/711 but not their order of preference].
Mike, it all comes down to how many VCM channels at your head office - if only 10 that's the max that can come in/out of THAT LOCATION. Even if other calls are passing back and forth between branches, no more than 10 will be coming down the pipe between your WAN cloud and head office's physical location. Branch to branch traffic will be passing through your ISP's backbone and not affecting head office's bandwidth.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.