This is chapter 4 of a multiple chapter learning exercise for those looking to buy a phone system for their business.
If you want to read chapter 3, you can find that here: What is Voice Over IP
What Role Does TDM Circuit Switch Have in a Business Phone System? Chapter 4
TDM / Circuit Switched
Remember the history lesson we discussed earlier? The big cord boards, and the old 1A2 phone systems. Those are circuit-switched systems.
A Circuit Switch phone system, sometimes knows as Time Division Multiplexed (TDM), essentially means that the communication from the handset to the phone system (control cabinet) is all done in a controlled manner. In a TDM world the voice packets are not sent along as data packets – all of the communication is controlled. That means no lost packets, delay, jitter, echo…that is because the phone system controls the mechanism between the handset to the switch.
By the way, having a TDM phone set at the desktop doesn’t necessarily mean that the phone system itself doesn’t support Voice over IP. Most major manufacturers have small and enterprising systems that support both TDM and IP-based sets – often on the same system.
Public Switched Telephone Network (PSTN)
The PSTN is the network of the world’s public switched telephone networks. Originally a network of fixed analog telephone systems, the PSTN is now almost entirely digital, and now includes mobile as well as fixed telephones.
The PSTN was the earliest example of voice traffic engineering. A. K. Erlang is credited with establishing the mathematical formula to determine, among other things, the amount of equipment, voice / line provisioning, and personnel to establish a specific level of service. In fact, the ERLANG formula is used to calculate the number of voice mail ports required to service a company. The formula uses number of calls into the system / hour, and average length of call, as point determinants.
Trunks, PRI, and analog lines are all examples of services in the PSTN network. A SIP trunk would be an example of an IP-based trunk service. Let’s take a closer look.
Trunks (business lines); PRI, analog lines (business line), & SIP
There are two types of phone lines that can be brought into a company.
Option #1: The traditional style phone line from the PSTN (TDM, circuit-switched) analog or digital (PRI) phone line.
Option #2: SIP-based phone line (which brings your phone lines in via a broadband connection). SIP-based trunks are very new to the market. Let’s take a closer look at these:
Analog Business Line, Trunk, 1FL
An analog line is the traditional type of PSTN access from the telco. The line coming in from the home would be considered an analog line. The analog line is being rapidly replaced by the PRI as the telco line of choice for most small to large customers.
PRI – Primary Rate Interface, Digital, Fiber:
PRI, sometimes also called Megalink, digital, or fiber, is a TDM, ISDN type (circuit-switched) type telco service which is rapidly replacing the older style analog line. A standard PRI service in North America has 23B channels and 1 D channel – if added together you have 24 channels. These 24 channels are comprised of 24 channels of 64 kbit / second, or 1.544 Mbit / second (24 X 64). The PRI is a similar service to the T1 – a T1 is made up of 24B channels, or 1.544 MB as well. The T1 has 24B channels, and the PRI has 23B + 1 D channel. The PRI is used for voice, and the T1 is used for data.
Essentially, the 23 channels are bound together and work collectively with all of the telephone numbers that come into the business. Let’s say, for example, that the company has a main line, a back-door line for their automated attendant / voice mail system, a customer service number, and a sales number. All of these numbers are bound together and share the resource of all 23 channels combined.
The PRI is a digital service, which means that the sound quality is near perfect. Unlike an analog line, which sometimes has a 6 to 8 dB loss, the PRI service has a 0 dB loss, making the sound quality that much better. I often hear from clients that do conference calls with analog lines that the two outside parties cannot hear each other that well. What happens with an analog line is that the person at the head office makes a call with a 7 dB loss. They put the first person on hold, and make a second call with a 7 dB loss. Although the person doing the conference call can hear both parties well, the two people on the outside cannot hear each other that well. The perception is a 14 dB loss (7 dB X 2). This problem doesn’t happen with digital (PRI) lines. A conference call with a digital line sounds near perfect. All three, four, five (or more) people can hear each other perfectly.
Unlike and analog line, call display comes in immediately on a PRI service (essentially before the first ring). Call display on an analog line comes in only after the second ring, so, in an analog world the first two rings need to be suppressed, providing the illusion to the outside caller that the company doesn’t answer their phones that quickly.
DID’s – Direct Inward Dial (DID) numbers
This is a feature offered by the telephone company which allows a range of numbers to be connected into the phone system and route directly to an individual’s phone, bypassing reception and the automated attendant entirely. By way of example, each extension off the phone system may be assigned a seven-digit external telephone number. Someone who knows the DID of the person they wish to reach can connect directly to that person. DID’s are only available on PRI service. When people give their work number and say it is a “direct line”, often what they mean is that it is a DID.
Fractional PRI – A quick note – A fractional PRI would be the equivalent of a full PRI but with fewer B channels. Because some of the carriers will sell a fractional PRI at less cost than a full PRI, this smaller “pipe” is attractive from a price perspective for a company not requiring a full PRI.
SIP Trunking (Session Initiation Protocol)
Let’s simplify this otherwise somewhat complicated concept. SIP Trunking is the mechanism used to connect phone lines into an organization’s business telephone system, and is emerging as a viable alternative to legacy (TDM PRI and analog lines). A SIP trunk is essentially a telephone line, or series of telephone lines and DID’s brought into the office over a broadband (Internet) connection. The SIP trunk is a viable option, although at the time of this writing (August 2009) the technology is relatively new and still not that well understood by the carriers. The problem with a SIP trunk, which must be addressed, is that since the telephone line is coming in via a broadband / Internet connection, the quality of the data line from the organization’s offices into the SIP providers premises is absolutely critical. Again, remember my VoIP Engineering rule:
Terrible Internet / data communication = Terrible voice
Amazing Internet / data communication = Amazing voice
(Don’t forget my VoIP Engineering rule. This is one of the most important elements of a properly designed VoIP network).
If running SIP trunking (or a hosted solution) you should bring in a dedicated circuit from the provider directly to your facility. DO NOT GO OVER THE PUBLIC INTERNET.
A SIP trunk should not be confused with a SIP client. What is a SIP client?
SIP Client (or SIP endpoint):
A SIP client uses TCP or UDP, and typically uses port 5060 to connect to SIP servers and other SIP endpoints. The most common application of SIP is in voice and video communication. As an IP-based communication protocol, it mirrors the traditional circuit-switched telephone network architecture with similar features and signaling protocols. This includes things like dialing a number, causing a phone to ring, hearing ring back tones, or hearing a busy signal.
SIP is a peer-to-peer protocol, and as such, it requires a very simple core network with intelligence distributed to the network edge, intelligence embedded into the endpoints (terminating devices built in either the hardware or software).
SIP is similar to HTTP in that it is request-response structured, and it shares many of the same HTTP status codes (making it more human readable). As such, SIP is a stateless protocol, making it possible to implement ‘failover’ and other features that are more difficult to implement in a stateful protocol (such as H.323). SIP is not limited to voice communication; it can mediate any kind of communication session from voice to video, video to voice, or to some other future unrealized application.
That was the technical explanation of a SIP Client. Here is the layman explanation of a SIP Client: You have a phone on your desk at work. The phone is an IP-based handset. BUT, instead of the phone talking directly to a phone system sitting in the network room, the phone (SIP endpoint) talks to a switch which sits outside of your offices. All control, talk paths, dial tone, provisioning…everything is done by the outside provider. Hosted PBX is an example of the use of a SIP Client. Vonage is another example of a SIP endpoint.
You could also have multiple end points like a soft phone, PDA, cell phone, desk set, and since SIP is an interoperable protocol which could work in a multi-vendor environment, it provides mobility and system flexibility.
Again, if running SIP trunking (or a hosted solution) you should bring in a dedicated circuit from the provider directly to your facility. DO NOT GO OVER THE PUBLIC INTERNET.
Since bringing in a dedicated circuit / DSL point to point for a standard SIP endpoint at an end users home is not always cost justifiable, and if you need to run your VoIP end point over the internet, just keep in mind that the quality might not always be perfect.
My personal preference
At the time of writing I still have a fairly strong preference for anything circuit-switch related. Not that I have a problem with Voice over IP – I absolutely do not. I think VoIP is an amazing technology if deployed properly. And even then, a VoIP connection / SIP trunk, although perhaps having some benefits, the cons and risks far outweigh the good points. SIP trunking is still a bleeding-edge technology, not even a leading-edge technology. (I am not referring to VoIP end points here).
I should probably leave my own bias out of this book and just present the facts; however, I live and then explain through experience. If you find, after doing your own research, that you are comfortable with the SIP line provider, then give it a shot (just make sure to do it right). Cover your bases. Just to recap, how do you cover you bases?
– NEVER, NEVER get a SIP trunk over a broadband Internet connection. You should always have a point-to-point connection between your office and the carrier’s own office (this applies of course only if you care what the voice quality sounds like).
– In Addition, always make sure that the contract you sign with your SIP line provider allows you to get out with a thirty-day cancellation clause.
– When purchasing point to point for the data circuit always purchase committed bandwidth from your office to the IP trunk providers offices. 10MB committed means you have 10MB of your own data – point to point. 10MB uncommitted means you and everyone else on the same pipe (2,3,5, 15 companies) share the same uncommitted bandwidth.
I am sure that there are some carriers in North America that know how to sell, install, and properly service a SIP trunk. I just haven’t found them! It is still a market of cowboys selling technology for technology’s sake. And until the folks selling SIP and IP-based technology understand that voice quality should NEVER go down, that 99% reliability is not acceptable in the voice world, that packet loss and echo are NOT acceptable on a business grade line, until they understand all of the above, I will NOT recommend SIP trunking. I will present the facts to a client as objectively as possible and let them make that decision. I will even install it for a client, if they so choose, but only after explaining the facts.
By the way, in case you are wondering, 99%, although perhaps a great mark on your child’s report card, is a very poor quality level. Ninety-nine percent up time means 1% downtime. One percent down time means over 85 hours down in a year – that is more than three days!
99.999 (or we call it five-nines reliability) is what you can expect in the circuit-switch market (PRI or analog lines), and until SIP providers can guarantee that, STAY AWAY!
Cabling & Switches – The GLUE!
Often overlooked, but probably one of the most important components of a properly designed voice and data network, is the cable infrastructure sitting at the backbone. Cable is the glue that holds everything together. You can install the best voice infrastructure in the market. Mess up the cable, however, and the system won’t work properly. I suppose the same can be said for the entire design component. This includes the endpoints, switches, cable, and routers. So, let’s investigate the GLUE !
Cabling
Since most VoIP phones have built-in switches, it is possible to have only one set of data cabling terminated to each desk. In this case the cable would run from the data network switch directly to the phone, and then from the phone to the computer. In my opinion, this is highly unadvisable. This makes the network highly dependent on one switch – all of the phones and computers would terminate directly to one data switch. Essentially, there is one point of failure on the network / office. If that one switch goes down, or has a problem, your office will not only lose all phones, but also all computers. It also makes troubleshooting the network and phone related issues much more complex – trying to determine where the problem lies can become quite problematic.
Of course, the prospect of only having to run one set of cables for all phones and computers seems attractive when looking at cost. This potential saving is quite deceiving though since a company looking to have only one set of cables running to each desk will require a more expensive data switch than otherwise would be required if the same company were running a separate voice and data drop to each desk. This type of high-end switch could cost $600 more, so the saving might not be as high as first perceived.
Having been exposed for so many years to the myriad of problems that clients face from network, voice, power…and every company’s sensitivity to their phone systems being operational, I almost always take the cautious approach. I would say, given the importance of voice quality, for the sake of $2,500 amortized over a five to eight-year period, the cost is quite negligible.
My training has come from the voice side – voice needs to be up 100% of the time. Well, 100% is not completely realistic, but as close to 100% as possible. A difficult and elusive goal to achieve, but something I try and get to on every installation. Moreover, considering the client base that we have, I am quite sensitive to the types of problems that we will encounter after the installation. If we sell even one concept which has pitfalls, and sell that concept successfully, we might win a deal today on the basis of cost, but could potentially cause our company many headaches in the future. My clients have made me very sensitive to hardware related problems, and I have become less tolerant of manufacturer related issues (which I could write a separate book about). Therefore, needless to say, I am cautious about installing a phone system with one set of cables for both the phone and data network. I want to make sure the back end infrastructure can support it.
Cautious does not mean that we will not proceed if a client so chooses. It means that all things being equal you are better off with one set of cable for the voice network, and a separate set of cables for the data network. If you are moving into a new building that you intend to occupy for a few years (5 years or more), and you will be re-cabling anyway, the weigh those costs.
Types of Cable
As a quick note as well, some of the cabling standards which should be considered when cabling a new office are:
CAT3: Older style typically voice cable. Generally $80 / drop. Suitable for TDM / circuit switch style sets.
CAT5: This cable is the predecessor of the CAT5e cable, and is capable of running speeds of 100 Mbps. I mention this cable because it is still in use in many offices, although essentially obsolete.
CAT5e: Most common data cable. Generally $90 / drop. Can run at 1 GB speeds
CAT6: More expensive data cable, generally $120 / drop. Like CAT5e can run 1 GB speeds, although with more room for error (and so fewer lost packets).
CAT6A: Newer (and still as yet not ratified standard) capable of running at data speeds of 10 GB. Approx. cost is $160 / drop.
Phones in the phone system (digital or IP)
Of course the decision to implement digital vs. IP-based handsets will be dictated by the type of phone system chosen. Some phone systems only support digital phones, some support a combination of digital and IP, and some support IP based sets only.
Now, if you choose to run IP-based sets in the network, you will need to plug the phones into the wall (they need electricity to operate). IP-based sets actually need power (electricity) in order to operate. You have two choices in terms of how you power the IP-based handset: plug it into the wall directly, or run the power back to a central source and run the electricity over the CAT5e (CAT6…) cable. This is called ‘Power over Ethernet’, otherwise knows as PoE.
From a pure handset perspective there is not much of a difference between a digital versus an IP based set. An IP set has more flexibility though when designing a redundant voice network (we will cover more on this later). An IP set can authenticate to a specific switch, and automatically overflow to an alternate switch for authentication in the event that the primary switch fails. A digital set can NOT do this !
Power over Ethernet
PoE describes a system to transmit electrical power over your traditional twisted pair cable in an Ethernet network. PoE is useful for powering IP-based telephones, but can also be used to provide power to other IP-based endpoints that require power. This includes things like Wireless LAN access points, webcams, and embedded computers, in addition to other applications where it might be infeasible to provide power to an IP-based endpoint requiring power.
The PoE switch or patch panel is not mandatory. All IP phones need electrical power. You have two choices. Plug the phone directly into the wall, or run the power through the cable and centralize the power back at the switch end. That is where PoE comes into play. PoE will allow you to plug the phone in at the remote end (instead of plugging it into the wall). That way you can centralize the power at a common source.
Quality of Service (QoS) Switches
In the world of packet switched and computer networking, the traffic engineering term ‘Quality of Service’ refers to the control mechanism that can provide a different priority to different users, or data flows, or guarantee a certain level of performance to a specific program or application. Quality guarantees are important if the network capacity is limited, and especially in a situation where packet delays are extremely sensitive. That is certainly the case with voice and data where a delayed or lost packet will result in a poorer quality voice.
A QoS Multilayer switch (MLS) can prioritize packets by the 6 bits in IP DSCP (Differentiated Services Code Point). Alternatively, what this really means is that a QoS enabled switch will allow for voice / video packets to get priority over other packets (web surfing for example).
In addition, you can get a QoS PoE switch instead of getting two separate devices (PoE patch panel and QoS switch). You can now get one device to do both (PoE QoS enabled switch).
Keep this in mind:
You can run digital sets inside the office, and IP-based sets to remote works outside of the office. Digital sets inside the office do not need PoE and QoS enabled switches. There are no voice-quality related issues with a digital set. If you are considering a phone system with IP-based sets, whether the solution is hosted or IP-based, you will still need PoE and QoS switches.
Here is my recommendation on this issue:
If disaster recovery is not an issue in your system purchase, then purchase a phone system with digital sets internally (this applies to smaller office systems – fewer than 300 sets). Remote workers will be IP-based. Connection of multiple systems between offices will also be IP-based. Run a network drop for voice, and a separate data network drop for your data network (no matter what – whether you have a digital or IP-based set).
This was chapter 4 of the series: How to Buy a Phone System for Your Business.
Let’s now move on Chapter 5 – Features of a Business Phone System
