Be Aware You’re Uploading - BAYU Alok Vimawala Sr. Network Administrator University Housing University of Michigan Alok Vimawala Sr. Network Administrator University Housing University of Michigan
What is BAYU? BAYU is a new program started by the University of Michigan It s students, that live in residence halls, when they are detected engaging in peer-to-peer (p2p) uploading
When did BAYU start? BAYU was introduced on October 30, 2007 University officials were discussing putting something in place to make students more aware of their online activities for some time
Who is affected by BAYU? BAYU is targeted towards students living in on-campus housing 15 residence halls and 1 apartment community The program covers approximately 10,000 individual that live on campus
How does BAYU work?
1. Student uploads data using a p2p protocol (bittorrent, gnutella, etc.) 2. The upload is detected by the PacketShaper ® 3. Every 10 minutes, a computer program connects to the PacketShaper ® and retrieves data about uploads detected by the PacketShaper ®
How does BAYU work? 4. The computer program determines the user responsible for the upload by checking with network registration systems in place at the University of Michigan 5. The program notifies the responsible user via
What technologies are used by BAYU? Linux Perl Oracle
Do I really need all of them? The technologies BAYU depends on are not set in stone. It just needs the following: A way to detect p2p uploads A way to relate uploaders to users A way to notify users that they were seen uploading
How often are users ed? Notifications are limited to once every 24 hours If a user was detected and ed 5 hours ago, he will not be ed now The 24 hour window is customizable per user and can be as small as 10 minutes long (every time the program is executed)
What does the say?
Help! I received a BAYU notification What avenues do students have for support? -Web based support - support -Phone support -In-person support
BAYU home page
BAYU help page
What if I don’t want to receive these s? Users can opt-out of receiving the s via a website Users can opt-out only if they have received a BAYU notification to ensure that users don’t “preemptively” opt-out The user will not be notified for the remainder of the term
BAYU opt-out page
support Students can reply to the BAYU notification and the reply goes to real people Another avenue for contact It seemed “natural” for students to want to reply directly to the address that sent the instead of visiting a link in the
Phone support Two different help-desks answer BAYU related calls Help-desks provide a basic level of support Common questions or complaints are addressed by the help-desks
Common questions Turning off notifications I am not sure what these notifications mean Please stop ing me! What triggered the notification -IP Address -Protocol
Common solutions False positive -World of Warcraft -Electric sheep screen saver Command-control traffic Inadvertently registered someone else -Rogue wireless access point -Registered a friend’s machine
In-person support One (soon to be two) walk-up computer support center for students residing on- campus Disabling p2p uploading Uninstallation of p2p applications
Privacy concerns How much does BAYU know? How long does BAYU keep that information?
What does BAYU know? Whether the user was ed within the past 24 hours How often to the user Whether the user has opted-out of receiving s
How much information does BAYU store? One week worth of notification and detection data One term worth of opt-out information One term worth of information about detected users Aggregate data about the program
How much information does BAYU store? Every time the program is executes it also stores the following aggregate data: -Number of users detected -Number of messages sent -Number of messages not sent due to users opting- out and also due to notification frequency -Total number of users detected to date -Total number of users that opted-out to date
Does BAYU work? How does one measure the effectiveness of the program? -Decrease in number of uploads detected? -Decrease in number of copyright take-down notices? -Criteria still evolving
Student response
Outside response
Future work Better detection of protocols Increasing the number of protocols that trigger a notification Widening the scope of the service to possibly cover wireless users Better tools for administrators of the program
Questions? Comments?