All of our calls are serviced by four independent, siloed and high availability data centers. Our Data Centers are located in Oregon, California, Virginia and Quebec:
The Penguin Network
Penguin Network Call Initiation
When calls come into our network, the PSTN will round robin the calls so that they are offered sequentially to any available Penguin data center. If a data center is not responding because of an outage, the PSTN will simply skip it and move onto the next data center.
This has a few unique advantages:
- No call ever fails to come into our network - even if a data center is completely down, the call is simply routed through another region.
- There are no "Primary" and "Backup" data centers - all of our data centers are fully in production and use the exact same networking backbone.
- Its FAST - because we route all calls on a call-by-call basis across four regions, we can accept an enormous amount of traffic.
- Our network can tolerate an incredible outage. If three of our four regions have a complete failure, calls will still complete normally.
- Our data centers are not aware of the other's existence - so there is no risk of a cascading outage.(In fact, the firewalls in our Data Centers explicitly deny access to our network from other data centers.)
- Its scalable - we can simply add more regions as needed to provide more capacity.
Conventional SIP Trunk Provider
Conventional SIP Trunk Provider Call Initiation
As you can see in this example - a conventional SIP Trunk provider generally has two data centers with one operating in production and the other as a standby.
The standby data center uses a monitoring system to determine if the Primary data center is online. When it detects a failure, it updates the routes with the PSTN and starts routing the calls accordingly.
There are a few problems with this conventional design:
- It only is effective after a failure. Calls that failed are not recovered and the callers likely heard a "This number is not in service" message.
- There is often a delay between the time the failure is detected and the failover is completed; this can be up to 5 mins depending on the network.
- Configuring a PSTN Backup route with a Heartbeat service is a complicated design; so if there is a configuration failure it can take hours to figure out what the problem is and resolve it.
- There is only one backup data center. In the case of a large event (like an earthquake, tsunami, power outage etc, its possible that both data centers will be impacted and will be unavailable
- Many SIP Trunk Providers buy cheaper bandwidth and lower quality equipment for their Backup Locations. This is because the conventional expectation is that they are used infrequently.
- This design is slow - you are limited in capacity to what one data center can provide
- The two data centers are linked - so its possible to have a cascading outage (where one data center's issues impact another)
Conventional SIP Trunk Provider Call Audio Path
This is a pretty conventional way most SIP Trunk Providers handle audio.
The fundamental reason why this is the industries standard practice is because this design hides who the SIP Trunk Provider's underlying carriers are and so the customer is always left to assume everything is being handled by the SIP Trunk Provider themselves.
This is problematic for many reasons
- If the SIP Trunk Provider has network issues in their Data Center (which happen a lot) the calls audio will be negatively impacted
- If the data center your call is going through has an outage, your call will drop
- This is a security risk - unscrupulous providers can listen in on your calls or record them.
- There is a potential for a delay in your audio
- Your never going to get the highest possible call quality because you will always be limited by the options offered by the provider.
- This is an expensive design - bandwidth is extremely expensive; conventional providers have to charge customers more to cover the costs of the bandwidth used to service the calls