...
Access to these domains need to be open regardless of the isolation zone (IZ0,IZ1) in which your account is built. These include some supporting services and libraries which allow Sharpen to run as designed.
Domain | Protocol/Port | Purpose |
---|---|---|
*.s3.amazonaws.com | TCP: 443 | Long-term audio and image file storage |
stun.l.google.com | UDP: 19302 | WebRTC STUN server |
stun1.l.google.com | UDP: 19302 | WebRTC STUN server |
stun2.l.google.com | UDP: 19302 | WebRTC STUN server |
stun3.l.google.com | UDP: 19302 | WebRTC STUN server |
stun4.l.google.com | UDP: 19302 | WebRTC STUN server |
*.yealink.com | TCP: 443 | Yealink auto-provisioning |
*.ckeditor.com | TCP: 443 | Visual editor/UI library |
*.loggly.com | TCP: 443 | Logging |
*.pendo.io | TCP: 443 | Analytics and logging |
*.ingest.io | TCP: 443 | Client logging |
*.gstatic.com | TCP: 443 | Font library |
*.googleapis.com | TCP: 443 | Font library |
*.fortawesome.com | TCP: 443 | Font library |
*.fontawesome.com | TCP: 443 | Font library |
...
Use Quality of Service to maintain prioritization
Many devices support Configuring Quality of Service (QoS) tags can help to maintain traffic priority across the network. It is beneficial to tag your voice traffic with the appropriate tags, so it can be prioritized anywhere in the network in the event of a saturated linksaturation. This will help to prevent any audio issues caused by voice and data competing for the same bandwidth over your internet connection.
...
Protocol | Port Range | Priority |
---|---|---|
UDP | 10000-20000 | DSCP 46 - EF |
UDP | 5060-5081 | DSCP 46 - EF |
Voice Protocols
SIP (Session Initiation Protocol)
...
<150ms latency to *.sipvbx.com (example: us1-webrtc-11.sipvbx.com for east coast US, or us2-webrtc-02.sipvbx.com for west coast US)
Average latency variation < 30ms
High variation represents interruption to your connection. This may be a result of competing network traffic, or general hardware/network instability
While high latency on its own simply means delay, latency variation typically comes coupled with packet loss, which will mean dropped calls and/or choppy audio
>10Mb/s internet connection
While voice itself is a fairly light operation, we recommend having enough bandwidth to handle all your operations. This value is more of a guideline, rather than a requirement. Most important is making sure your collection of tools, including Sharpen, have sufficient bandwidth.
The best way to determine bandwidth needs is to sample your tool set usage, and extrapolate from there.
Sharpen bandwidth utilization can vary widely based on how it is used some base-line examples of usage are as follows
Sharpen Q page load for single agent logged into 4 queues
~350 KB transferred
~7 MB page resources
1 minute outbound call from Sharpen Q
~125 KB transferred
~1 MB page resources
Reporting/Insights (10 reports) page load
~150 KB transferred
~ 8 MB page resources
< 1% packet loss
Voice requires basically no For a positive experience, voice requires minimal packet loss. If any packets are dropped, it will interrupt the audio stream. For this reason, packet loss directly influences the quality of a callYou may experience disruptive delay or choppy audio. Enough packet loss will cause dropped calls.
< 30ms jitter
Jitter is the variation in delay of packets. Having high jitter will also cause poor call quality.
Common interruptions
SIP ALG - Present in many is enabled
Perhaps contrary to its name, SIP ALG is not compatible with most enterprise VoIP solutions, such as Sharpen. Depending on the manufacturer, network device configurations
will show up as “SIP ALG”, “SIP”, “VoIP”, or something similar. Intermittent disconnection,
dropped calls, one-way audio, and the inability to register are common symptoms when SIP ALG is enabled.
Sharpen strongly discourages the use of “combination” network equipment such as all-in-one modem/router devices. ISPs typically provide or rent these out to customers. They are known for having issues with VOIP traffic as well as having limited access to critical settings. If you are using a combination device and experiencing issues, the first step is to acquire a stand-alone modem and router.
If you can not acquire a stand-alone modem and router, we recommend reaching out to your ISP to see if they can place your combination device in “bridge” mode, then purchasing and connecting a 3rd party router to handle the local networking.
Ensure UDP timeouts are greater than 240 seconds (phones register every 240 sec). If you’re seeing a sawtooth pattern in your latency graphs, or your phones are sometimes unable to be reached, the UDP timeout is probably incorrect.
Disable Stateful Packet Inspection (SPI) as it often flags VOIP traffic incorrectly.
Disable any VOIP specific functions that come pre-setup on your network equipment.
Sharpen traffic over a VPN is discouraged due to likely latency and quality of service concerns.
- Sharpen traffic over an MPLS is discouraged due to potential inefficient route paths to voice resources
Depending on your device manufacturer or ISP, it may be difficult to get a straight answer on confirming this setting is off. Commonly, ISPs will have this setting enabled, because it supports their own options for VoIP solutions. It is not uncommon to have to work through a couple layers of support or technical team members to validate the proper setting is turned off.
Stateful Packet Inspection (SPI) is enabled
Similar to SIP ALG, SPI is a setting which has its benefits, but can conflict with the proper operation of voice traffic. Stateful Packet Inspection is a dynamic packet filtering technology for firewalls which inspects the state of packets to determine whether packets should be blocked or allowed. WebRTC traffic is stateless, and thus you may experience issues with the inability to establish a connection or 2 way audio if it is enabled. This happens because SPI may not recognize webRTC and thus treat it as unapproved UDP traffic.
NAT misconfigurations
NAT (Network Address Translation) exists to provide security and preserve local IPv4 addresses. NATs associate or bind private IP:port addresses to public IP:port addresses. Hosts outside the local network will only see and know of the bound outside IP:Port. Hosts directing traffic to this bound IP:port will first arrive at the NAT before being translated and sent to the actual destination IP:Port.
With VoIP and webRTC, NAT can cause challenges if not configured optimally. Though the STUN protocol exists to overcome this challenge through its participation with ICE candidate negotiation. Some NAT configurations can still get in the way. For instance, utilizing a Sonicwall firewall without the use of its “Consistent NAT” feature, will result in 1-way audio. Consistent NAT uses an MD5 hashing method to consistently assign the same mapped public IP address and UDP Port pair to each internal private IP address and port pair.
UDP timeout is set less than 240 seconds
Sharpen’s expected SIP registration interval is 4 minutes (240 seconds). If your network is set to “timeout” UDP connections at less than that, it will disconnect an active registration. Depending on when this happens, you’ll see the following symptoms.
Dropped calls
Inability to be reached on the phone which has lost its connection.
You’ll be able to dial outbound without issue, since registration is established on an attempted outbound call, if it does not already exist.
You’re probably seeing a sawtooth pattern in your latency graphs.
ISP provided “combination” network equipment
Especially if you’re working from home on your residential internet connection, be weary of the Internet Service Provider (ISP) controlled settings which may exist on these managed devices. “Combination” devices are typically those which integrate modem, router, and wifi into one device. While in principle, the integration of these functions problematic, they can sometimes come with configuration hurdles which are difficult to overcome since you, as the borrower of the device, do not have administration access to the devices. It is not uncommon for settings like SIP ALG to be enabled but invisible to you as the user. These situations require that you work with your ISP’s support team to change a setting.
As a result, Sharpen discourages the use of combination network equipment. Instead we recommend purchasing a stand alone modem which is compatible with your ISP, and connecting a router of choice to it. This allows you full control of any potentially conflicting setting. Most self-managed devices have these problematic settings disabled by default.
If you can not acquire a stand-alone modem and router, we recommend reaching out to your ISP to see if they can place your combination device in “bridge” mode, then purchasing and connecting a 3rd party router to handle the local networking.
VPN
VPNs have many valid use cases, especially for work from home users. However, VPNs are not always configured to be optimal for voice traffic. Adding the additional virtual layer to the network, in most situations, will cause a recognizable degradation in network performance. Voice requires low latency, with minimal packet loss. As a result, if the VPN introduces too much interruption, your quality of service will be impacted. If VPN is necessary, it is recommended to configure it so Sharpen traffic can be omitted through the us of split tunneling.
MPLS
Similar to VPN, an MPLS has its place in the enterprise network configuration realm. However, the use of an MPLS with Sharpen is strongly discouraged. Most commonly, customers will have had MPLS configurations remaining from previous on-premise or direct to data center VoIP solutions. This makes sense because you have narrowly identified VoIP endpoints which you can create a specific path to through the MPLS. Your users then connect to your corporate network and then their traffic tunnels through that optimal connection. However, this architecture actually causes problems for a more universally available cloud-based solution such as Sharpen.
Consider this example. You’re a user based in San Jose working for an organization headquartered in Chicago. You have an MPLS configured to connect from Chicago HQ to your datacenter in Indianapolis, where your ISP has its terminating hardware. All internet traffic from Chicago traverses the MPLS to get to its final destination. I, the San Jose user, connect to the VPN so I can access my corporate network assets. I login to Sharpen. My connection, which could go directly, over the internet, to our west coast Oregon location, instead is connecting over 1 virtual layer, then a physical MPLS to Indianapolis to get to the internet. At this point, the quickest available path to Sharpen becomes our east coast Virginia location. You know how a user connecting across the United States for a voice connection which most optimally works on a low latency, low packet loss environment.
The best solution is a solid internet connection with standalone network equipment. This allows for the lowest latency path to our resources, and the best experience.