SIP Provider recommendation

FreePBX Providers
21

We’ve talked about “warm spare” systems a lot in the past - this is a good start.

There are economies that come into play with this design discussion. For example, the likelihood that your phones are going to fail (hardware wise) is pretty remote. A phone, once in a while? Sure. All of your phones? Probably not.

Primary and backup switches are a reasonable plan. In fact, setting up two switches with half your phones on one and half on the other isn’t a bad plan. If you suck a switch, half your phones die.

Setting up a warm space server that you can cut over to isn’t a bad plan. You could also do something with a Virtual Machine where you lose one in the cloud and another one that’s configured exactly the same way magically stands up (this, by the way, is not a path I’d use because I still don’t trust other people managing my critical infrastructure).

Even if you end up with two live PBX servers (one primary and one set up for failover for incoming calls), you should be able to connect two buttons on your phones with extensions on the servers. I’ve set this structure up a couple of times, and the second server spends almost all of it’s time collecting dust.

The thing is that you need to make sure that your network (the totality of your infrastructure) is redundant in smart ways.

22

@sorvani, thanks for teh Skyetel reco. I’m checkin that out, but I have to say their “Jessie” chat bot started off on the wrong foot by disconnecting and losing our chat just after I told “her”/it I was looking for an HA SIP provider as good ro better than SIPStation. D’oh!

23

@billsimon, thank you very much.

The system I inherited was running on a very old Nortel PBX. When we were cutting over to VoIP, we were definitely drinking from the firehose, so I had not reviewed the SIPStation SLA. Also, my partner had used them for a long time at his previous job and had not suffered a service interruption.

The SIP trunk is particulalrly important because it’s a single point of failure (even if it offers geo diverse SIP servers) because there is no way that I know of to share inbound DID’s across vendors.

As we have just seen, any single vendor may have internal single points of failure that are difficult or impossible to identify from the outside (e.g. the Sangoma database).

Honestly, I had simply presumed that a SIP provider with as many customers as SIPStation has would have a telco equivalent SLA. (My bad.)

We do realize that 5 nines availability of SIP connectivity does not provide 5 nines VoIP availability. We have our FreePBX hosted across 3 geo diverse locations (one is cloud based at freepbxhosting.com) and with diverse WAN connectivity. And, yes, I did read the ISP SLA’s.

This SIPStation outage is the only service interruption our dispatchers have experienced since we cut over a little less than 6 months ago.

If I can find a way to diversify our SIP vendor, I’m hoping we can go for a long while without another outage.

If I’m mistaken about sharing inbound DID’s across multiple vendors, I would very much like to hear about it.

24

We do have analog (POTS) as a backstop in the event of total failure. That got us through the SIPStation failure. But, I don’t consider that acceptable and will continue to work to get our trunking as reliable as the rest of our VoIP system. five-nines is not unreasonable and has been offered over POTS since the dark ages.

We also have proceedures to cut over to cell phones as a backstop to our POTS backstop, fwiw. Didn’t need to cut over to that. :wink:

25

The guys that gave you the expectation of five-nines for phone service provide sip trunking , even prefer it, they also have a customer facing website to forward any particular PSTN to any other PSTN for whatever reason including unvailabilty of your system or the failure of their network.,

Given their very heavy investment in “the backbone” of the internet, they would have to figure at the top of any reliability list. Apart from cost the only niggle I ever had with them is to insist on G711 codec on all calls otherwise you get g729 for all calls, faxes switch to g711 but any other “modem like” call will fail.

26

@dicko, I respect how active you are on this forum, but I find your responses unhelpful and frankly, offputting. If your intention is to help answer questions and help expand the FreePBX community, I think you might be acting against your own objective. If your intention is to demonstrate your superior knowledge of the subject, I don’t think this is the best forum, and I don’t think it’s working terribly well.

27

I can also recommend Telnyx for wholesale/enterprise. They are an actual CLEC carrier. Not an aggregator like Twilio and most other SIP providers. There are no minimums or commitments of any kind.

As stated, they have their own private network. I believe they use AnyCast for SIP failover/redundancy internally so you only have to deal with one IP address. Their prices are very aggressive as well.

They recently expanded and have PoPs in Canada, Europe, Singapore, and Australia now. Brazil and Hong Kong coming soon. You can kind of think of them like Level 3/CenturyLink but global, specialized on only VoIP, without all the traditional Telco bureaucracy, and very inexpensive.

My one gripe about them was that they didn’t have a media direct option to lower latency but now they do.

28

With a toll free number, you can (and should). For example, Voip Innovations toll free service has carrier redundancy by default. Also, the resporg can update routing in real time when needed; you can think of this as being able to port your number in one minute. That’s why most large organizations still use toll free numbers, even though the vast majority of callers now have nationwide calling. Of course, you’ll pay per minute for incoming calls, typically $0.01 to $0.025, and you would have to get a new number.

With a geographic DID, you can choose between getting service from a carrier (somewhat more expensive but eliminates a middleman), or from a VoIP provider who buys from the carrier. However, depending on your infrastructure, connecting via the provider can be more robust. For example, the carrier failover options may be limited to sending calls to PBX B when it can’t reach PBX A. This won’t do you much good if you don’t have a PBX B, or if both systems are behind the same failed internet connection. OTOH, the provider likely has options to fail over to POTS or mobile, or to at least take a voicemail with notification by SMS.

Some questions for your provider:

What geographic redundancy do they have?

How do they determine where to send your calls? If possible, avoid schemes that require registration.

If (because of a routing problem on the internet) they can’t reach your PBX from their server A, will they automatically retry from B, even though A is fully functional?

Does their architecture have carrier-facing servers separate from customer-facing ones? That’s important so in case of e.g. a DDoS attack, they can still forward calls to your mobiles, take voicemails, etc.

How do their fraud controls work? You don’t want a situation where your account is drained by a fraudster (even if it’s your fault) and they hold your incoming calls hostage until you pay up.

Who is the underlying carrier? In small and medium size cities, the CLECs such as Bandwidth, Level 3, etc. have only one point of interconnect with the local PSTN, so one fiber cut will take out your service. I estimate that this occurs on average once every ten years and takes two hours to fix. That alone is more than twice the downtime implied by five nines. IMO, a good solution for a very small organization that needs no more than three concurrent incoming calls is to get one line from the local cable company, with call forwarding on busy to a VoIP provider. While the overall cable service may not be very reliable, the head ends of Comcast, Spectrum, etc. do achieve five nines reliability and they will fail over to your POTS or mobile when the cable is down.

29

There was no intention to be unhelpful, Calls to a geographic Phone number are only sent to one destination, be that a CLEC a ILEC or RBOC , The better the network that that organization has access to the more likely the call will be well handled, most all “VSP” have fail-over abilities, but if the “VSP” itself has network problems the call will fail as will the fail-overs.

The bigger and more experienced the Carrier, the more redundant their network will be, AT&T is both big and experienced, and can deliver SIP calls to you, generally over fiber to your premises.

They further provide a private network and a managed router , often a Cisco, and they ask that the Cisco has a dedicated analog backup line attached so their engineers have access when the main connection fails, The router has the ability to be “multi-homed” so an alternate route via another fiber or Wireless or even Satellite or GSM can be arranged.

The Phone Numbers can be provisioned to follow whatever route is still available. So if, as you have been asking for, a provider that can send ALL calls reliably to your system , which of course needs 99.999+ availability, then they deserve your consideration.

I suggested you look at the only “big boy” left that can relatively seamlessly, albeit at a premium price, do all you need.

(I am bemused if you find any of this thread off-putting or unhelpful, many of us have actually done this , largely succesfully, you as yet “not so much” (to your ongoing angst) )

30

In any system, there is a risk of failure. Even AT&T and Verizon have had failures. A few years ago, Callcentric (considered a very reliable voip provider) had an outage that lasted for three days.

Redundancy is a concept that is well known in aviation. It’s why commercial airplanes have two engines, two pilots, two radios, two methods of navigation, etc.

For inbound calls, redundancy that means at least two separate DIDs, hosted by different providers, which can be configured to use completely independent call paths to reach you. For example, the first DID might be your main number. The second DID might be a cellular phone. If all of your DIDs are hosted at one provider, you lack redundancy at a critical point. If the only way for you to receive a call is via your internet connection, you lack redundancy at another critical point.

For mission critical services, like an air ambulance service, both DIDs should be listed everywhere you provide a phone number to the public:

i.e.,
John’s Air Ambulance
(212) 555-1212 | (212) 999-1212

During normal operations, the secondary DID could be pointed to the first one. In the event the first one fails, you can then point the second one to its dedicated path.

I cannot think of any other way that secures full redundancy on at every point of failure for inbound calls.

If you’re looking for another provider, then you’re not solving the real problem, which is your failure to design your system with complete redundancy.

31

For our company sites, I have been using SIP.US. Found them to be very reliable. The only downtime was twice in a year for a total time for 3 minutes. During scheduled maintenance. This comes out to 99.9994% or something. Their support is extremely responsive when I’ve needed them. Not sure how competitive their pricing is, but there is an effective web control panel. So I can call-forward any DID’s to an alternative destination in the event I have a PBX-side issue.

1 Like
32

I’m going to second J Finstrom’s suggestion for SIP Trunking. You’ve essentially got the bulk team that took ownership, improved, and maintained FreePBX for the last 6+ years now working there, as well as an established trunking company as an acquisition. The service so far has been cost effective and enterprise class with the customers I have moved over and the 2 acquisitions that Clearly has done were with established providers: Cyclix is the former/re-branded BandTel guys who were one of the first to offer SIP trunking 12 years ago.

However, I think this is a larger issue than just SIP trunking. I can tell you that our most recent rollout of PBXAct, which is supposed to be a “well tested, controlled” environment is anything but that. I logged 20+ hours of “this should be working out of the box” tickets into this last rollout. Non-PBXAct updates needed to be shoehorned in. The Vega module is flat out broken. Sangoma phones weren’t working properly. It was a disaster, and had it been an early install, I would have jumped ship. Honestly, my old systems that I thankfully haven’t updated still are rock solid.

Even trying to get a product manager on the phone was an ordeal. None of the Customer Service or Sales reps that are with Digium / Switchvox have a clue about PBXAct.

Feels very much like Fonality / Trixbox CE. Anyone remember that?

Now you’ve got the team that built the rock solid platform all in one place, just wearing a different badge. It’s seems to be an easy choice.

33

That is quite a humorous thing to read. It made me have a giggle fit.

1 Like
34

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.