• V20: 3CX Re-engineered. Get V20 for increased security, better call management, a new admin console and Windows softphone. Learn More.

Intermittent Call Quality Issues

Status
Not open for further replies.

twisted1

Gold Partner
Joined
Feb 18, 2011
Messages
34
Reaction score
7
All,

I have a client we are hosting 3CX v15.5 from our Data Center. They have multiple locations with only a few phones at each (Restaurants). We have them set with a VPN connection back to our DC so all phones are provisioned over the local lan. For the most part all is good but they have periodic episodes of horrid calls. Mostly choppiness issues which I understand to be more jitter related than latency? When it happens, it is only the restaurant that can't hear the caller. The caller hears just fine. I have listened into these calls from MY remote connection and I hear both sides crystal clear.

We have struggled to capture both sides of a call with WireShark when this happens as it is usually after hours and by the time we get on there to start the capture, the issue self resolves.

Each site is on a Fiber connection through Comcast with PLENTY of speed. SHOULDN'T be an issue there. We have seen the issue at multiple sites at the same time and also only one at a time so it is not apparently the server or the DC network which has a Gigibit connection.

I had my team add PRTG probs on that network. The issue happened again last night and was gone before I could log in but I pulled the PRTG report and saw latency cpikes up to 200ms when it is normally 20ms. I saw similar spikes in that time frame on other restaurants in that chain but NOT with other clients on different providers where we have the same probs and hosted 3CX.

I am wondering if this Latency is my smoking gun but since Latency doesn't normally result in choppy or intermittent lost one way audio, I am unsure.

Ideas/suggestions?
 

Attachments

  • Latency Issue.JPG
    Latency Issue.JPG
    44.9 KB · Views: 21
So you say that hey have a fiber connection with plenty of speed, but that is subjective. If it really is fiber then I doubt they have plenty of speed unless money is no object :). Comcast fiber typically starts around $500 for 10mb symmetric and 10mbit is not a lot, especially if it is being shared. I would monitor your bandwidth usage at both the restaurant side and your DC. If the restaurants have WiFi then you could be looking at a saturated connection from WiFi usage (either guests or employees). And since you said it is your DC you could be looking at saturation at that end for a number of reasons. Hosted backups out of your DC comes to mind or congestion at the peering point between your provider and Comcast. PRTG between your customer and DC as well as to a non Comcast location and your DC would shed more light.
 
We are doing what we can to monitor usage so have that covered. The fiber tests out at 22mb up and down. Given the systems involved, that should be more than sufficient.

Any take on the latency graph? Could that indicate an issue that would cause one side of the conversation to have choppy audio?
 
In the purest sense, it is all latency. It is simply a matter of over what time and the packets involved. Latency as a measure is the amount of time it takes to get data from one point to another. Jitter is the delay variability between the packet delivery/arrival. If the packets were to be delivered at the same rate as when sent and the delay were 500ms, there would be no jitter, but you might have an issue of talking over one another. The latency could get to be excessive whereby packets are dropped.

There are a few things -
1. The graphed latency in and of itself is not granular enough to really grasp any likely impact. The time period is so broad, that the graph may be smoothed to be able to display the result over time. 140ms is still considered as an acceptable latency by most for VoIP. However, if the graph resolution is masking the swings that might be occurring within the time periods, then this may account for some issues.
2. The graph does indicate similar results over the approximate same time period which may be an indication of some other time oriented event such as that alluded to by cobalit.
3. Presumably the remote phones may have a jitter buffer function. You might be able to take a single site and set the jitter buffer up for a 100ms fixed setting and then see if it makes a difference. While the phone may have a variable setting, I have had some issues with using it on some and as a result only used fixed.
4. Hopefully you have a decent router and perhaps you can look at flow to determine bandwidth utilization at the remotes to confirm speeds are not at issue.
5. Finally, you might take a look at voipspear.com. It offers a monitoring solution that provides some easy to understand performance values that allow you to gauge the quality and reliability of the connection. I have personally used it and it has helped me immensely in driving home facts to reluctant ISPs that would otherwise insist the issue is mine. I have no ties to them other than subscribing to the service, but it has proved more than worth the cost to me.
 
  • Like
Reactions: YiannisH_3CX
We have been tracking this issue for weeks now. It seems that the root is pointing back to the Comcast network. Running Pathping every night and we are finding that when the traffic transitions from a spot a few hops past our DC and into the Comcast network it incurs a 130ms or greater delay and begins to lose up to 40% of its packets. Certainly enough to kill a call... We are pushing our DC to open a case with Comcast.
 
Status
Not open for further replies.

Getting Started - Admin

Latest Posts

Forum statistics

Threads
141,603
Messages
748,762
Members
144,715
Latest member
iTVerse
Get 3CX - Absolutely Free!

Link up your team and customers Phone System Live Chat Video Conferencing

Hosted or Self-managed. Up to 10 users free forever. No credit card. Try risk free.

3CX
A 3CX Account with that email already exists. You will be redirected to the Customer Portal to sign in or reset your password if you've forgotten it.