This report documents the results of a Cisco Systems Inc. (Nasdaq: CSCO) IPTV infrastructure test. Back in 2006, Light Reading commissioned the European Advanced Networking Test Center AG (EANTC) to conduct an independent test of a complete end-to-end solution for service providers focusing on video services.
The test was groundbreaking in several ways. We verified an end-to-end IPTV and triple-play solution focusing on the network service requirements of broadcast TV and video-on-demand (VOD) applications.
EANTC asked Cisco to design a network for up to 1 million customers and to provide a full POP, backbone, and service-area infrastructure for the test, serving 60,000 emulated users. The test was based on a mock RFP (request for proposal) detailing requirements for a fictional, but realistic, service provider triple-play network. Cisco’s ability to propose a suitable network design was part of the test exercise.
Our test covered four main areas:
Massive scaleability of broadcast TV and VOD services
QOS (quality of service) in an integrated business and residential customer network with real-time video, voice, and data service
High availability with sub-second failover time for all of the services
User experience for video services under high network load and packet loss on the first mile
The Cisco solution showed excellent results in all test areas:
Sustained zero packet loss operation serving 60,000 triple-play customers with 20 high-definition and 200 standard-definition video channels, 22-Gbit/s VOD traffic, and additional Internet traffic
Zero loss 1:60 replication of multicast traffic at 99 percent load
Out-of-service times under node and link failure situations of less than 70 ms for any unicast service and less than one second for the multicast service
Correct QOS implementation even under full load with 70 percent real-time traffic in the low latency queue
Substantial improvements of video service performance with Cisco's unique call admission control (CAC) for unicast and multicast services; we verified that CAC prevented oversubscription in the network by too many video viewers
Outstanding video quality experience with up to 10 percent packet loss on the last mile when Cisco’s proprietary set-top box enhancements re-requesting missing packets from a network buffer server (VQE) were implemented
It took nine months from the first preparations to publication of the test. A few hiccups delayed the start a bit: In total, four weeks were required to configure the routers under test. A team of 10 engineers from Cisco, EANTC, and Spirent Communications plc (NYSE: SPM; London: SPT) were involved in the test full time, from April to June 2007, sometimes being augmented with another 30 Cisco engineers for hardware and software troubleshooting. Preparations culminated in two weeks of day-and-night testing in Cisco labs, completed in time for presentation of this detailed report to Light Reading readers just before NXTcomm.
It was a tremendous effort – but well worth it: IPTV infrastructure is clearly very complex and requires a different, more holistic focus in implementation and testing than past broadband access solutions. It seemed a simple exercise in the beginning: Take standard Multiprotocol Label Switching (MPLS) capabilities, stir in some aggregation hardware with lots of queues, add multicast, and away you go. But what we found is that service providers today cannot run IPTV services "out of the box." In order to successfully implement and make money from these services they must partner with a vendor that will do extra and sophisticated network design work, and that can supply additional technology to prevent network over-subscription and other problems.
Cisco delivered on these requirements in our test – providing a new generation of hardware specifically suited for large-scale, real-time video deployment. Its proprietary video enhancement software worked well; and Cisco’s design proposal worked superbly in our mock service provider network, providing us with a high level of confidence that the solution will easily scale to the design goal of a million users and beyond. In other words, Cisco did extremely well in a hugely challenging test. (See Testing Cisco's IPTV Infrastructure.)
— Carsten Rossenhövel is Managing Director of the European Advanced Networking Test Center AG (EANTC), an independent test lab in Berlin. EANTC offers vendor-neutral network test facilities for manufacturers, service providers, and enterprises. Carsten heads EANTC's manufacturer testing and certification group and is responsible for the design of test methods and applications.
— Jambi Ganbar is Senior Test Engineer at EANTC, focusing proof-of-concept tests of triple-play core, aggregation, and access networks. Before joining EANTC, Jambi had worked for MCI and CAIDA.org.
- I lack an analysis why the IPTV packets was lost. If this happens every time the Residential Internet traffic is oversubscribed, it is a very serious problem which is not discussed by the authors of the report and might result in higher percentage of drops than indicated in the report.
- It would also be interested to see how a solution performs in a more realistic topology, with more hops and other/fewer fiber stretches. Passing traffic in such simple topology and passing so few routers, it would be strange if it did not achieve the results reported in the tests.
- What is the price per subs for this solutions? Some indication should be fine. This is interesting in the light of that the author says that the CRS'es are lightly loaded. What happens if they are heavily loaded? Can it be justified to have such big routers a solution
The blogs and comments are the opinions only of the writers and do not reflect the views of Light Reading. They are no substitute for your own research and should not be relied upon for trading or any other purpose.
To save this item to your list of favorite Light Reading content so you can find it later in your Profile page, click the "Save It" button next to the item.
If you found this interesting or useful, please use the links to the services below to share it with other readers. You will need a free account with each service to share an item via that service.