ISP-filter may fail on NBN
- 17 December, 2009 15:51
- Comments 2
The Government’s Internet filtering proposal failed to address clean-feed content in high-speed environments like the National Broadband Network (NBN) and is ultimately doomed, a technical group claims.
This week, Broadband Minister, Senator Stephen Conroy, announced the Federal Government will go ahead with a mandatory ISP-level filtering legislation. The decision came after the Government reviewed results of the Internet filter trial, which was monitored and reported on by Enex Testlab.
In response, technical expert group, SAGE-AU, pointed out the Enex Testlab report showed filter tests were done at 8Mbps and below, despite the criteria for testing being 12Mbps and higher connections. The report also excluded sample sizes and the amount of traffic subjected to filtering.
With the NBN promising speeds of up to 100Mbps and 12Mbps as a minimum standard, SAGE-AU president, Donna Ashelford, claimed the filtering trial was inadequate in demonstrating efficacy of Internet filters and potential speed degradation on the impending network.
“Let’s compare the broadband network with road traffic in random breath testing. If you have a nice slow road, nice big shoulders and with not much traffic, pulling someone in to have a test wouldn’t cause too much drama,” she said. “If you do that on an eight-lane freeway, certainly you will cause more speed impact than on a road that is wide and uncrowded.”
Considering most of the ISP participants in the filter trial were small-scaled companies, Ashelford claimed the sample size was limited and tests results were ineffective in reflecting the impact on the wider broadband network. She called for further trials to be performed before the Government implemented its proposed Internet filtering laws.
“The technology should be tested under real-world speeds and loads that will relate to the NBN,” she said.
Nominations for the 2012 ARN IT Industry Awards open on Tuesday, June 12.
- Bookmark this page
- Share this article
- Got more on this story? Email ARN
- Follow ARN on twitter
- Premier Media Group Fast Study
- Market Potential-Strategy Guide to the Active Archive Market
- Aberdeen Group: Building Business Resilience Through Active Archive
- Spectra Logic and Australian National University Success Story - March 2012
- In Search of the Long-Term Archiving Solution —Tape Continues to Be a Major Player
-
REVIEW: HTC Sensation - a powerful beast wrapped in a sturdy, aluminium shell
-
First look: Samsung Galaxy S III
-
Spotify tunes into Australia
-
Telstra and Navman Wireless extend GPS tracking partnership
-
World’s eyes on Aussie NBN: Conroy
Get exclusive access to ARN's news, research and invitation only events. 
Comments
Anonymous
question why were no NBN services from Telstra or Optus provided
good points, can someone clarify; Were NBN speeds available from any of the ISPs or Telstra in this "live" filter pilot? Surely Optus or Primus at the time of the trial have NBN home capability and plans they could have used.
Roddy
NBN a problem? Nah...
Luckily, the recent ISP trials were not representative of the IT industries capabilities in this area anonymous.
They were representative of how many Aussie ISPs were ready to spend how much of the available federal funds for testing various methodologies.
Telstra did their own tests and Optus would have to answer the question regarding their own trial.
The EOI did stipulate testing at those NBN speeds, which does not prove that faster speed tests were not done, only not documented at speeds above 8Mbps.
It is actually the consolidated ISP "traffic gateway" loads and requests per second which are the real measure of the performance and scalability of ISP filtering systems.
Thus:
http://www.dbcde.gov.au/online_safety_and_security/cybersafety_plan/internet_service_provider_isp_filtering/isp_filtering_live_pilot/isp_filtering_-_frequently_asked_questions#15.0
"ISP-level filtering devices available in the market can handle 10 gigabits per second of traffic or more per device and most solutions have the capability to cluster the devices in groups to support an aggregate of 80 gigabits or more.
The technology allows considerable scope to scale for supporting large quantities of traffic. Devices in the development pipeline are expected to have considerably higher capacities."
We recently reviewed systems handling 16Gbps of gateway router traffic per mid range appliance look-up server with blacklist URL filtering, limited though to 8 servers per cluster today (load balancing switches spread the requests)
Each device peaked out at just under 290,000 URL request lookups per second.
That vendor is just implementing for a 20 million user telco, mandatory blacklist filtering, 4 gateway location implementations. They only need a couple of the look-up appliances per gateway at this time, as the 6 mill user gateway did not exceed 300 requests per second at peak times.
They also set a team of hackers onto accessing and compromising the lookup servers for 6 weeks. No luck, no bonus. More secure than the main routers...
They also tried all sorts of URL lengths, URL malformities etc to cause anomalous behaviour... No luck and no bonus. The report though did mention that they limited the URL lengths to 8192 bytes I believe, in the test, but the string conversion from the URLs was apparently unlimited.
The URL blacklist was limited to 100,000 URLs, and only 10 of the top traffic sites globally were tested, with URLs from google, yahoo, youtube, facebook etc trialled.
The boxes never peaked... The NOC engineers did not even notice a performance blip when the systems were turned on.
Why the cascading/load balaning scalability? Because this telco expects to be ramping their services up to 100Mbps connections as times goes on, and to see significant mobile computing URL request activity growth in the next few years...
Nice report, real telco, bigger by far than all Aussie internet users together, it did work.
And the users? Well no one complained, not even questions, as no one noticed...
That may assist you to get a broader understanding of what major system vendors can deliver, and that the NBN will not be an issue.
We reviewed the results and verified the methodology and the numbers. Cheers
Post new comment