Testing was conducted on the above listed Unified Communications components configured to provide UC for a network topology consisting of a main office and three branch office locations. The four-site deployment was specified to support up to 2,500 aggregate users with 2,000 of these users equipped with full voice messaging and unified communications capability. We included full redundancy for server components at the main office and three branch office locations.
Testing was conducted on unified communications components configured to provide UC for a network topology consisting of a main office and three branch office locations. The four-site deployment was specified to support up to 2,500 aggregate users with 2,000 of these users equipped with full voice messaging and unified communications capability. We included full redundancy for server components at the main office and three branch office locations.
The main office network infrastructure utilized a 100Mbps Ethernet Power over Ethernet-enabled network using an Avaya C364T-PWR. The branch offices were connected by different routed virtual LAN networks
We conducted H.323 to SIP endpoint tests and verified that latency and voice quality were within acceptable standards for voice. MOS 4.0 was our minimum, but the Avaya 9640 IP phones using the G650 Media gateway actually achieved better quality at a 4.5 MOS score. These tests, as well as SIP signaling analysis and verification of traffic encryption, were conducted using the ClearSight Analyzer, a distributed protocol and traffic analyzer from ClearSight Networks. Voice quality assessment was evaluated by capturing live traffic off the network and by doing post capture analysis processed through the ClearSight Analyzer.
SIP traffic load tests (which emulated endpoints) were conducted using the Emprix Hammer Transport and Touchstone's WinSip. An impressive 6,000 busy hour call attempts with 1,000 concurrent users was achieved in testing.
Voice quality assessment for G.722 wideband codec employed in the Avaya 9640 IP hard phones was evaluated using Touchstone's WinEyeQ product and subjective scoring by our listening panel. The voice quality was described as "near high fidelity" by our testers. The ability to hear the high and low tones is greatly enhanced using the G.722.
Security scans including open port scans, protocol interaction with mutated traffic, common vulnerability exploit tests, denial of service and SIP server torture tests were conducted using Miercom's own testing suite combined with a Mu Dynamics Mu-4000 Security Analyzer. The vulnerability scans were run first to provide information that could be used in a subsequent compound or complex attack. We did not utilize network countermeasures in the network for the preliminary security testing (such as a firewall or UTM product) as we wanted to exercise the inherent security of each of the UC components.
The components in Avaya UC package did not reveal many chinks in the armor to allow for further exploits. Worse case scenario of denial-of-service tests revealed that, although communications may be temporarily disrupted, the components of the UC system would not fail or require reset which is a significant accomplishment. Follow-on testing was conducted using a UTM appliance from Avaya's partnering vendor Juniper Networks and was verified to thwart the DoS and compound exploits that we applied to the unprotected Avaya UC system. Bottom line, with the correct network security countermeasures employed, we could no longer disrupt the VoIP traffic.
We conducted this review looking at six areas including Features and Functionality, Integration, Reliability, Management and Monitoring, Security and Performance.
The criteria under consideration in each category are outlined below.
Features and Functionality
Support of and effective use of all communications modes: IM, fax, voice telephony and e-mail, was tested. Credit was given to advancements in text-to-speech and speech-to-text conversion. Multi-language support including voice-to-text was tested and verified to work well (we tested Russian, Spanish and English). UC features including use of VIP (important caller) lists and scheduled meeting or away times, as they integrate into the use of presence, was tested. Client-to-client presence propagation was also tested.
We tested how well the Avaya UC client would interact with MS Office Communication Server and IBM Lotus Sametime. Observations included the effectiveness of advertising an Avaya user's presence to the non-Avaya IM clients. Phone presence propagated well to the third-party applications. UC presence did not propagate well outwardly, but worked well importing presence from the third-party applications.
The UC clients' integration with underlying clients was included in the evaluation. For example, the one-x Communicator interfaced well with the resident third-party e-mail and IM program on the desktop client. The overall IP PBX integration with the Exchange or Lotus Notes server providing access to voice mail, visual voice mail (caller identification information and .wav file in an e-mail attachment) was included.
We tested system component reliability including endpoints and the servers themselves exercising the inherent failover capabilities by disconnecting network links and interrupting power in some cases. The 9640 IP phones were capable of multi-zoning to register to back-up servers for resiliency. We interrupted connectivity to the primary SES SIP server and confirmed the 9640 phones would indeed reregister and continue to function using the alternate SIP server that it could reach. The S8730 server was tested to see if it would fail over to a hot standby S8700 with no loss of telephony or other UC communications. A full system failover of the Main Office S8700, causing branch offices to fail to local S8300s resident in G450 gateways, was also verified. Voice mail resiliency could not be tested.
Management and Monitoring
We evaluated the ability of Avaya's management and monitoring tools to pinpoint problematic voice communications on the customer network. The Avaya Integrated Management Suite provided a clear view of the managed network and isolated problem nodes and connections that reported poor voice quality. The tool works by placing test traffic on the network and it reports latency, packet loss and jitter which is projected to an estimated MOS.
Included were tests for vulnerability of each of the UC components in the Avaya solution using scanning tools and commercially available security testing solutions. We evaluated the behavior of the components while under attack and our ability to disable any component was attempted. Complex attacks were conducted using noted potential vulnerabilities. Server hardening guidelines and user guides were considered.
Tests included SIP load tests with Empirix and Touchtone equipment to stress the IP PBX and SES SIP server in the network to achieve 1,000 concurrent users with 6,000 BHCA. We evaluated voice quality for H.323-to-H.323, to SIP using G711 CODEC as well as SIP-to-SIP calls using G.722 wideband CODEC. Time for presence to update and propagate was evaluated, with Avaya typically achieving less than 1 second update from UC client to client (for example updating the status from "on-phone to available") even in a mixed Avaya IM and third- party IM client environment.
< Return to test: Avaya offers wide array of Unified Communications wares >