Virtually all currently marketed IP telephony systems share a common design platform of server-based call control and distributed media gateways, but there still remain noticeable differences that customers should be aware of when evaluating choices. The annual IP Telephony System RFP workshop at the recent VoiceCon Orlando confirmed that the offers from the participating suppliers are still unique from one other in some important ways, such as: centralized versus distributed call control topology; single-purpose versus multi-function media gateway equipment; standalone server versus integrated server/media gateway for remote survivability; or use of proprietary versus third party industry standard hardware equipment.
The basic configuration requirement for this year's RFP workshop was for a large system design of 1,600 line stations (50% growth capacity) housed in a headquarters central facility with control and port interface hardware equipment distributed across two equipment rooms for geo-redundancy. Note that high speed direct fiber optic cable and 10 Gbps LAN connectivity was available between the equipment rooms to connect the equipment (see diagram below).
There were also two remote branch options behind the central facility: a small line-size facility (50 stations) and a medium/large 200-station facility (see diagrams 2 and 3).
It was requested that system solutions be proposed for each of three configurations:
* the central facility, only;
* central facility with small remote facility;
* central facility with medium-large remote facility.
Each of the remote branch options was required to support local survivability with seamless call preservation in case of WAN failure or loss of control signaling from the central facility. In survivable mode, basic feature support was sufficient for the 50-line site, but full feature support was desired for the larger remote site. Although a single system solution was preferred to satisfy the distributed central equipment rooms and optional branch requirements of the RFP, multiple networked systems were acceptable if a single system image was fully supported, i.e.: 4/5-digit dialing plan across the facilities; transparency of feature, function, and application operations; automatic alternate routing; and unified systems management.
Ten system suppliers responded to the RFP and participated on the panel:
* Aastra
* Alcatel-Lucent
* Avaya
* Cisco Systems
* Digium
* Microsoft
* Mitel Networks
* NEC
* ShoreTel
* Siemens Enterprise Communications
The following are summaries of the proposed system designs for each supplier for the central facility and remote branch configurations. There are many similarities among the proposed system designs, but each has at least one or more distinguishing elements from the others. Though most suppliers proposed a single system design, some were required to propose a networked solution to satisfy the RFP redundancy/resiliency requirements. Also, note the increasing deployment of third party servers and/or media gateways compared to system designs of a few years ago, a result of the increased focus on a software-centric approach to IP telephony.
The RFP document and detailed responses from each of the system suppliers are available for download to registered conference attendees.
Aastra
Aastra proposed a single fully distributed system design, a MX-ONE based on multiple off-the-shelf industry standard (HP DL360) telephony servers with associated media gateways. Redundant (N+1) telephony servers were proposed for each of the central facility equipment rooms and a non-redundant full-featured telephony server was housed at the optional remote facility for survivability requirements. Note that a fully configured single system can support up to 124 distributed servers. The MX-ONE telephony servers were programmed to operate in load sharing mode, though each can operate independently of the others if signaling connectivity is lost and could have fully supported the full capacity communications service requirements of the RFP.
The current system release allows a single MX-ONE telephony server to support up to 15 associated media gateways and 15,000 station users. Fully distributed MX-ONE Lite Media Gateways (which includes an embedded Media Gateway Unit with digital trunk interface, codec gateway DSP resources and TDM time switch) were proposed to work behind the telephony servers in each of the central facility equipment rooms and at the remote location for local analog station and PSTN trunk requirements.
Alcatel-Lucent
Alcatel-Lucent proposed a single system OmniPCX Enterprise solution with redundant IP Appliance Servers (IP AS) based on industry-standard IBM hardware. The redundant server pair was distributed between the two central facility equipment rooms in active and standby modes: during standard operations the active server supported all communications services at the central and optional remote facilities. A Passive Communications Server (PCS) was proposed for remote site survivability requirements in case of WAN failure or signaling issues. PCS provides full feature support (except attendant service) in survivable mode, but requires a warm operations restart which terminates ongoing communications and affects all calls, though new calls can be established within about a minute. Alcatel indicated that an alternative option is to install a SIP proxy server to preserve communications during a WAN failure, but without the full feature support of a PCS.
Alcatel proposed M2 Crystal Shelf Media Gateways, based on traditional circuit switched port cabinets, housed in each of the two central facility equipment rooms to support the sizable TDM interface requirements at the main location. At the remote facility, IP Rack Module Media Gateways, co-located with the PCS, were proposed for the limited local analog station and PSTN analog/digital trunk interface requirements.
Avaya
Avaya proposed a single system Aura S8800 Server platform: fully redundant active/passive S8800A Servers distributed between the two equipment rooms at the central facility. The primary equipment room also housed servers for SIP Enablement Services (SES) and Application Enablement Services (AES), though it should be noted that an upcoming system release will eliminate the need for these peripheral servers by using a single virtualized S8800A Server currently supported for mid-size customers. The soon-to-be-released virtualized server offer for large line size customers will also support embedded media & utility services and voice messaging.
Each equipment room in the Avaya proposal housed multiple G450 Media Gateways for analog station and PSTN analog/digital trunk requirements. For the optional remote facilities, Avaya proposed a S8300 Local Survivable Processor (LSP) option housed in a media gateway for seamless, full feature survivability: a G430 Media Gateway was proposed at the 50-line site and a G450 Media Gateway at the 200-line site. Note that a self standing S8800A is also available as an LSP option for customers with larger remote line size requirements (up to 2400 stations). Avaya also had the option of supporting basic feature survivability at the small remote branch by proposing its Standard Local Survivability (SLS) feature fully embedded in the local media gateway equipment at no extra cost to the customer.
Cisco Systems
Cisco proposed a single system Unified Communications Manager cluster design. The configuration included load sharing active Subscriber servers in each of the two central facility equipment rooms and at the optional remote facility. Also housed in the primary equipment room at the central facility was a Publisher server (for CDR function) in support of distributed cluster. IP station users were registered to each of the provisioned Subscriber servers (primary, secondary, and tertiary mode) for call control redundancy.
Cisco 2900 ISRs were proposed for PSTN analog/digital trunk requirements and housed in the two central facility equipment rooms and at the remote locations. Analog station interface requirements were supported using distributed Cisco VG-224-24 media gateways in each of the equipment rooms at the central facility and at the 200-line remote facility; the Cisco 2900 ISR, alone, was sufficient to satisfy the minimal analog station requirements at the optional remote 50-line facility. Note that Cisco also had the option of proposing its IOS router Survivable Remote Site Telephony (SRST) feature to satisfy the basic feature requirements at the small remote branch, but indicated that the full-featured Subscriber server as proposed was cost effective for the configuration.
Digium
Digium proposed multiple networked Asterisk Business Edition systems to satisfy both the geo-redundancy and remote survivability requirements of the RFP. Discrete Asterisk systems were housed in each of the central facility equipment rooms and at the optional remote facility: a single system image was supported across the systems. Digium selected HP Proliant DL 360 G6 servers for the hardware platform, though other supplier industry standard servers could also have been deployed.
The Asterisk system design included three types of functional servers in each central facility equipment room and at the remote facility: OpenSIP's SIP Registrar Server; Asterisk Call Management Server (call set-up/tear-down, SIP registration); and Asterisk Application Server (central registration and location database operations, also supporting messaging and conferencing services).
Fully redundant server pairs for each function were housed in each central facility equipment room; single (non-redundant) servers for each function were housed at the remote facility. The Digium proposal also included a dedicated server equipped with a customized Digium PCI telephony card in each of the central facility rooms and at the large remote branch facility in support of PSTN digital trunk interface requirements. Analog station/trunk interface requirements at the central and remote facilities were supported using third-party media gateway equipment from AudioCodes (Mediant 1000 model)
Microsoft
The Microsoft proposal was based on its recently-announced Communications Server "14" platform. The single system configuration design included three different functional servers housed in each of the two central facility equipment rooms: Consolidated Front End Server (endpoint registration, IM, presence, call control, conferencing services, location information services, call routing and mediation services for peripheral gateways); Back-end Server (database services); and a Monitoring Server (proxy server function between endpoints and the Back-end server).
The distributed server groups across the two equipment rooms were programmed to operate in load sharing mode and also support communications services at the optional remote branch facility. The equipment in either room is capable of supporting the entire system configuration and communications requirements in the event of a failure of equipment in the other room. A single Edge Server housed at the central facility was included to provide an interface to remote station users for registration, feature support, access to conferencing services, federation for third parties running Communications Server, and presence/ IM support for public IM service providers. Third party media gateway equipment from AudioCodes (Mediant 1000 model) was housed in each of the central facility equipment rooms for analog station and PSTN analog/digital trunk interface requirements.
The recently announced Survivable Branch Appliance was proposed to satisfy remote facility survivability requirements. Microsoft selected the AudioCodes version of the new Appliance, noting that HP, NET, Dialogics, and Ferrari have also announced their own versions. The Appliance includes: an embedded Windows Server to run Communications Server "14" software; mediation services; media gateway interfaces; and a SIP registrar. All call feature functionality is available to remote station users in local survivability mode except updating of call forwarding settings and Response Group routing.
Mitel Networks
Mitel proposed a multiple system network solution to support the geo-redundancy requirements at the central facility. Two systems were programmed to work transparently in load sharing mode across the central facility equipment rooms with a single system image: Either system is capable of supporting the full configuration port capacity if the other should fail.
Each central facility equipment room housed an industry standard server, provisioned with redundant Mitel 3300 MXe model processor blades, supporting Mitel Communications Director (MCD) generic software and the Mitel Applications Suite (MAS). Each equipment room also housed a Mitel 3300 CX for PSTN digital trunk gateway requirements and a Mitel 3300 AX for analog station/trunk gateway requirements.
In addition, the proposal included a Mitel Border Gateway (MBG) in the primary equipment room for remote (teleworker) stations and SIP trunk support. At the 50-line remote facility Mitel proposed its 3300 CX for seamless, full feature survivability and local media gateway (analog station/trunk) requirements. At the remote 200-line facility, a redundant Mitel 3300 MXe was proposed for seamless, full feature local survivability and local PSTN digital trunk requirements; a Mitel Analog Services Unit (ASU) was also included for local analog station/trunk requirements.
NEC
NEC also proposed a multiple system network solution to support the geo-redundancy requirements at the central facility: two intelligently networked UNIVERGE SV8500 systems (using the NEC FCCS option) distributed across the equipment rooms. Unlike the Mitel load sharing design, in normal operating mode all endpoints at the central and remote facilities would be under the control of the SV8500 system housed in the primary equipment room. All affected endpoints would failover to the backup system in the other equipment room during disaster recovery mode if the primary system fails or control signaling is not available. A drawback to this approach is that during the failover re-registration process, active calls will not be preserved.
The SV8500 systems housed in each of the two central facility equipment rooms were based on fully redundant (active/passive) call control deployed on special-purpose servers; NEC peripheral application servers were also housed in the primary equipment room with the primary SV8500 system. Rack mountable NEC MPC chassis equipment for supporting a mix of analog station, analog trunk and digital trunk media gateway blades were evenly distributed across the two central facility equipment rooms (all configured as part of the primary equipment room SV8500 in normal operating mode).
For the optional remote facility, NEC proposed its Survivable Remote Media Gateway Controller (SRMGC) to provide seamless, full feature communications in case of WAN or signaling failures. In addition, NEC proposed its MPC media gateway equipment at the remote facility in support of local analog station and PSTN analog/digital trunk requirements.
ShoreTel
The ShoreTel proposal was based on a single fully distributed system design. The configuration included a mix of custom appliance ShoreTel Voice Switches, including the following models: ShoreTel 220T1, ShoreTel 30, ShoreTel 50, ShoreTel 90, ShoreTel 90v and ShoreTel 24A. Multiple ShoreTel Voice Switches act as a single unified communications solution, but each has the intelligence to function independently if there is a disruption to the LAN/WAN, keeping telephony services available to assigned/registered endpoints. Each Voice Switch provides call control and feature support for a specified number of IP/SIP endpoints and includes embedded media gateway services for analog station/trunk or digital trunk ports; the proposed 90v model also has embedded voice mail support.
The system configuration consisted of multiple Voice Switch models equally distributed across the two central facility equipment rooms (N+1 redundant design for each room based on 90v model). If any Voice Switch should fail, devices currently associated with it will automatically re-associate with another local Voice Switch (based on available spare port capacity) or to a N+1 switch located anywhere in the network); failover is supported across both the LAN and WAN. The remote facility housed additional Voice Switches (unaffected by any WAN or central facility signaling failures) providing primary support to local station users for all call control, media gateway, and voice messaging services.
Siemens Enterprise Communications
Siemens proposed its carrier grade dual-node OpenScape UC Server solution based on a dual redundant server architecture. The Siemens configuration was programmed to operate in load sharing between the nodes: all endpoints are registered with both nodes to ensure uninterrupted communications services if either node fails. The nodes were distributed across the two central facility equipment rooms and programmed to operate in load sharing mode. Call control in each node is based on a fully redundant IBM xSeries Cluster Server. In addition, each equipment room housed a server for media services/ management tools, and one room housed a Licensing and Deployment Server for mass provisioning of endpoints.
Note that all of the above core system functions, along with several applications (messaging, contact center, and UC), will be able to run on a single virtualized industry standard server following commercial availability of the new system software release just announced at VoiceCon Orlando. For media gateway requirements, each central facility equipment room housed a mix of third party equipment from Mediatrix: 4124 model (analog station); 1204 model (analog trunk); and 3531 model (digital trunk).
The new OpenScape Branch, available in several models based on SIP station capacity requirements (ranging from 50 to 6,000 lines), was proposed to support seamless survivable communications services at the remote facility. The appliance offers the following embedded functions: integrated SIP proxy; media services; Session Border Controller (SBC); and security functions (firewall, VPN). OpenScape Branch, in its initial release, supports most, but not all, of the standard generic software features in survivable mode: Bridged Call Appearance, Call Park, and Programmed Call Coverage are the more prominent features not supported. Local media gateway services for analog station and PSTN trunk requirements were supported using a mix of Mediatrix equipment (see above for models)
The following table summarizes the system design proposals from three perspectives: how central facility geo-redundancy was supported; how remote survivability was supported; and deployment of 3rd party equipment (servers, media gateways, and SIP phones)
Allan Sulkin is president of TEQConsult Group and a regular contributor to No Jitter.
