I was speaking to an industry friend the other day and happened to mention that my workshop at VoiceCon San Francisco will be a review and analysis of the current flagship PBX systems from the leading enterprise communications suppliers. Why, he asked me, was I still focusing on the "switching system" when the hot topic today is Unified Communications (UC). I responded that the core IP Telephony System (IPTS), the current moniker for a PBX, is still a vital component of any customer’s enterprise communications network and that VoiceCon needs to offer attendees at least one workshop or conference session with a focus on its architecture and design. There are several key reasons why it is still important to educate customers on the current state of IP telephony systems, such as:
* An IPTS still provides dial tone and call connection handling for desktop and wireless voice terminals to initiate call connections among system endpoints (stations and trunk circuits)
* The core telephony system is necessary to provide a wide and diverse number of still-prominent station and system features that are integral to implementation of several UC processes involving call coverage, conferencing, and collaboration. This includes features such as transfer, forward, call distribution and hunt groups, shared line appearance, and, of course, conferencing;
* IPTS design and topology determines the degree of redundancy and resiliency—attributes upon which customers depend for support of traditional telephony services, including basic contact center operations;
Today's IPTS hardware elements bear little resemblance to the digital PBX of 10 years ago. Although the IPTS of today consists of fewer distinct hardware components, and the core switching operation has mostly been replaced by LAN switches, customers who are still using an old system design need to be brought up to speed on today’s product choices and understand differences between competitive models. There are many current-day IPTS technological issues and capabilities that did not exist during the heyday of the digital PBX that are important to understand as part of a customer's purchase decision process. For this reason my VoiceCon workshop will address the most salient current-generation IPTS technology attributes that customers must be aware of to make a wise product selection.
Understanding differences between IPTS models is also important to reconcile price differences between system proposals. Sometimes one system may be priced slightly higher than another, but offers customers discernibly greater performance potential. Most current IPTS models share a common basic architecture platform (centralized or distributed telephony server, media gateways, and communications devices), but no two are exactly alike. Capacities between models differ for several design elements, such as number of supported line stations, trunk interfaces, and media gateways. The degree to which competitive models support industry standards and/or are interoperable with third party solutions also varies. Whether a current generation IPTS offers a smooth migration for a customer with an older generation PBX from the same manufacturer is also a highly important consideration for decision-making purposes.
System Topology
To a layperson, all telephony systems may look similar, because the only part they usually see or interface with is a telephone instrument. Telecommunications managers know that it is the system design and topology in the equipment room that truly defines and differentiates one model from another. The necessary numbers of servers and media gateway cabinets/carriers that make up the configuration are important criteria for determining system cost, both upfront capital expenditures and ongoing operations/maintenance services. Other equally important evaluation criteria include hardware equipment design (type, size and footprint, capacities, power requirements, et al) and how all of the elements fit together, i.e. the design topology. Growth and expansion capabilities, redundancy and resiliency levels, and sometimes feature/function capabilities are also determined by the system design and topology.
There are currently a variety of design and topology platforms among the leading IPTS offerings. Most small and intermediate systems are still based on a centralized single server design, but larger enterprise offerings are likely to be based on a geographically distributed server topology. There are two basic types of distributed topology configurations, including redundant active/backup and multiple active load sharing servers. It's important to note that the number of installed active load sharing servers is a determinant of available port capacity, with highly variable server capacity parameters per model. In some cases, embedded processor blades are used in lieu of standalone servers, and/or peripheral servers may also be required to support control signaling, functional, and/or API requirements.
Another vital topology issue is how servers interface. Most system designs are based on LAN connectivity among server nodes, although some may require a direct connection, e.g., fiber optic cabling. Whether multiple servers can connect and communicate across customer facilities (campus or multiple sites) over a private WAN is an important consideration, especially for customers planning to consolidate an existing network of PBXs into a single system configuration. Distance between servers may also play a factor, because of control signaling latency restrictions. It quickly becomes evident when surveying the many currently marketed IPTS offerings that virtually no two competing systems are identical in design topology when factoring in all the variables. Individual customers have their own opinions as to which design topology is preferable for their requirements.
Standards
An ongoing IPTS design issue is increased support of manufacturer approved Commercial-off-the-Shelf (COTS) servers and media gateways. The days of proprietary common control and port interface hardware are past, and open standards are in--or as a friend of mine said, "A Linux appliance is a Linux appliance." IT departments that standardize on servers from IBM, HP, or Dell will be glad that the new IPTS conforms to their requirements. Media gateways for industry standard analog ports (2500-type stations and CO trunks) and digital T1-carrier, PRI service and SIP trunks are available from a large number of sources, usually at a reduced cost compared to proprietary equipment. COTS equipment reduces costs and provides customers with a greater choice of equipment options. The extent to which a particular IPTS model supports or does not support these important COTS equipment options can be a major decision factor in choosing between systems.
COTS support is an outgrowth of IPTS conformance to industry standards. Digital PBXs were always criticized for being designed using proprietary hardware, from the cabinets to the port circuit cards to the telephone instruments. Even the operating system used by the common control processor was usually proprietary. Go back even further and you will discover that the processing elements were proprietary. It was considered a major breakthrough when leading suppliers like AT&T and Nortel began using processor chips from the likes of Intel and Motorola. Today's increased number of systems using appliance servers and the Linux operating system is the just the latest step in the longtime evolution of PBX design.
The most important standard that has been incorporated into IPTS design is Session Initiation Protocol (SIP). SIP is a protocol widely used for call set-up and control signaling of multimedia (voice, data, video, text, and message) communications sessions and also for supporting the transmission of Instant Messaging (IM) and presence information. It is quickly evolving into the protocol of choice for IPTS system designers and is important from both a station and trunk endpoint perspective. An IPTS that is based on SIP standards can potentially support a variety of SIP station equipment (desktop instruments, mobile devices and soft phone clients), offering a customers a greater choice of voice terminals from a wide selection of suppliers. Choice, in this instance, can be important for performance and/or price considerations. An IPTS that supports SIP trunk services allows for Voice over IP (VoIP) network connections between models from competing suppliers; it does so more simply and at a lesser cost compared to more traditional PSTN trunk network solutions.
Most current IPTS models were originally designed before SIP was a design issue, and support was added through software upgrades. But a few of today's offerings were originally designed using SIP as the native protocol. Virtually all current IPTS models claim to be SIP-compliant, but the means and degree of compliance varies from system to system. For example, there are systems that currently require a peripheral server or proxy gateway to support the signaling protocol, or feature provisioning may be limited to station users with SIP telephone instruments. SIP telephones from any supplier may also not work with a particular IPTS model, nor may a SIP trunk service from a particular provider be supported unless compliance testing is done. The SIP situation today closely parallels what happened 20 years ago when digital PBXs first started incorporating ISDN service support into their design. It is still caveat emptor in the evolving world of IPTS and SIP.
Capacities
There are several capacity parameters a customer must consider before installing their next IPTS. The most immediate that comes to mind is port capacity. Some systems are rated based on total ports, some on a combination of stations and trunks. It is also important to understand port capacity parameters based on station type, e.g., analog, digital, or IP, and trunk, e.g. analog, digital, IP, or SIP. It is a common marketing tactic to use the largest number possible in brochures and presentations without fully describing the configuration specifications. Sometimes the quoted number is based on a network of systems, not a single system. Actual station capacities may also be based on configurable telephone models, the number of instruments with displays, or programmable button assignments.
Media gateway cabinet/carrier capacities can also factor into system design considerations for customers with very large port capacity or distributed network requirements. Though many IPTS models can support large numbers of media gateway equipment, some smaller models may be quite limited. The number of remote media gateways may also come into play. Media gateway capacity is important for systems equipped with a significant number of non-IP ports, station and PSTN trunk, because it determines the number of concurrent connections to IP endpoints. The number of available media channels is equivalent to digital PBX TDM bus talk slots for purposes of analyzing busy hour traffic handling capacity. Non-IP ports may be on the decline, but they still represent a sizable percentage of the installed base, even for IPTS softswitch models.
Call processing capacity, as measured in busy hour calls, is no longer the important limiting factor it once was, because today’s servers are far more powerful than the old digital PBX common controllers. Server limitations, however, remain an issue based on the many different software program loads that may be configurable to satisfy a customer’s basic and advanced communications requirements. Dedicated servers are likely to be required for a variety of processing-intensive application system functions, such as:
* select contact center operations, e.g. enhanced routing, self-service, or reporting;
* mobile communications;
* API and/or SOA interface requirements.
Converged telephony/UC systems also have subscriber limits per primary control server based on programmed function support requirements. Customers should be aware which IPTS models require one or more customer-provided servers to satisfy the operational requirements of the system. More often than not the answer may be found, too late, in the fine print.
Redundancy/Resiliency
I have written many articles about the superior IPTS redundancy and resiliency attributes available today compared to digital PBXs of only 10 years ago. The cost to the customer for a high availability enterprise communications system is lower today for several design components, sometimes by a significant factor. Redundant and/or duplicated servers are priced far lower than the old proprietary digital PBX option, and the level of redundancy may also be superior; fully distributed design topologies may allow for more than one backup server per station user domain. Declining LAN switch prices, with costs allocated for both voice and data communications, make redundancy of the switching function more affordable. Pooling media gateways with logical connections to endpoints, not directly tied-up physical links, is a major design improvement compared to more traditional port circuit card/endpoint pairings which have more than a few points of failure to worry about.
An IPTS offers customers important geographic redundancy and resiliency options. Control servers need not be co-located, but can be distributed over a LAN/WAN or via fiber optic cabling as described in the above topology section. Customers with remote branch requirements can implement a local survivability option. Two basic design options are currently in use: a dedicated local server or a survivable media gateway carrier. The degree of survivability varies, though, from system to system. Several systems will support full feature/function survivability, while some will support a limited feature set in survivable mode. There are also port capacity limitations to consider. Seamless survivability without temporary loss of telephony services or call disconnects may be another distinguishing attribute between two systems. For customers planning to consolidate large network configurations, the number of distributed survivable servers and/or media gateways must also be taken into consideration.
Network failover resiliency among multiple systems is something that was never available with a digital PBX. It is the capability for station equipment to failover to a pre-programmed backup system if the primary system control fails; the backup system may or not be an active system prior to failover. Customers should fully understand how network failure is implemented, the capacity parameters and limitations for each competing IPTS model, because there is variance between the solutions. Knowing which systems even have this desirable capability is, of course, the most important evaluation factor.
System Migration
An upgrade migration from an installed digital PBX system to the current generation IPTS model from the same system supplier is usually a smart customer move--but not always, once all factors are considered. For example, a prime motivator for an upgrade migration, instead of a forklift, is financial, because customers may be able to protect a significant percent of their earlier system investment in hardware and software. There are also training costs and factors that play into the decision. Replacing common equipment and especially telephone instruments is usually something to avoid.
There are situations, however, where a system supplier's current IPTS offering may require a significant swap of hardware equipment. The older the installed system, the more likely this will be a factor, and if a customer decides to replace virtually all of their digital telephones with IP voice terminals, desktop or soft phone, then the financial savings will significantly shrink. Knowing what the upgrade migration process is for a select system pairing (installed system and current generation system models) is vitally important for the decision making process.
If the upgrade migration makes financial sense--an especially important consideration during tough economic times--it is possible that the upgrade will not result in a solution optimal for the needs of a customer today. The dynamics of business and organization influence the changing enterprise communications requirements over time, and it is necessary to determine if the newly upgraded system is best for today and tomorrow's needs. The distributed and virtual nature of today's business organization also requires an IPTS that is distributed in topology, but at the same time remotely survivable, if a data center solution for network consolidation is warranted.
As cited above, there are varying degrees of redundancy and resiliency between competing IPTS models that support multiple remote branches and satellite facilities. It may be necessary to replace one system supplier platform for another. The same holds true for other system evaluation factors, such as generic software features and peripheral applications, particularly contact centers and UC-like capabilities. The best core system does not necessarily equate to the best contact center solutions or UC application. Something as seemingly mundane as cellular handset support may be a deciding factor selecting between systems.
Summing Up
Anyone who has not been in the market for a new enterprise communications system during the past few years will be amazed how everything has changed from when they purchased their last system. Even customers with an early IP telephony system platform will find most new system design and architecture choices foreign to their knowledge base. Voice calls may be on the decline for casual communications, supplanted by email and IM, but it remains the preferred method for important interpersonal contact.
After 30 years of performing PBX system competitive research and analysis, I hope that some of my knowledge and insights will benefit the VoiceCon workshop audience. I was privileged to present a PBX review workshop to kick off the first conference in 1991 when system suppliers such as AT&T and Fujitsu were still in the business and Nortel was going strong. At the time no one speculated that Cisco Systems or Microsoft would join the competition. Many things may have changed in the interim, but I know that customers still need a solid understanding of the core telephony system to maintain a functional and reliable enterprise communications network. I hope to see you in San Francisco soon.
Allan Sulkin is the founder and president of TEQConsult Group and the industry’s leading consultant for analyzing and dissecting enterprise communications systems. His consulting services have been engaged by virtually all of the recognizable system suppliers and many leading edge customers. He authored PBX Systems for IP Telephony (McGraw Hill Professional), has been a very frequent contributor to No Jitter and its predecessor, Business Communications Review, for almost a quarter century and played a large role in the growth and success of VoiceCon since its inception. You can email him at [email protected].
