Transcript Adolfo Rosas

Future for internet streaming media
Adolfo M. Rosas
EU workshop January 19, 2010
Future research priorities in the Networked Media
Systems research area
Context
•
•
•
•
•
•
•
•
•
•
•
•
Internet video & media streaming is a driving force for traffic consumption today.
Video will be 90% of total internet traffic by 2013. (CISCO)
Some commercial services were designed along the past 15 years to close the
internet ‘media gap’: CDNs.
Video & media objects have been ‘small’ until very recently (2 years).
Consumption of internet media has moved from downloads to streaming.
Sudden availability of file sharing (P2P) created a ‘video culture’ in internet.
Some companies invested a lot to push this culture : Youtube.
P2P file sharing and P2P streaming could not close the media gap.
Commercial P2P services could not close the media gap.
Telcos fight P2P: DPI, throttling, …
IPTV never made it to internet: now moving streaming protocols from complex to
easier solutions. RTP+MPEG2TS+RTSP  TCP+HTTP
Premium media (BluRay , HD) out of the legal services in internet due to
deficiencies in security and misalignment of majors and ISPs biz models .
Internet media today works (barely) just because the ‘pipes’ are big now but still
dumb…
Short term
• There is a need to simplify the streaming media state of the art
(mostly inherited from IPTV) to fit the internet.
• Must stabilise media transports : mapping any new coding schema
(MPEG4AVC,On2VP6,…) onto ‘pure’ HTTP is a must.
• Correctly map scalable coding on the easy transports
• Investigate ways to provide QoS to internet HTTP flows end to end
• Evolve HTTP to be a suitable media transport
• Work to enable multicast safely on wide area multihomed networks
Mid term
•
•
•
•
•
Define schemas for the dynamic creation of media networks on top of internet :
overlaid media networks. Build upon the work done by P4P and ALTO.
Being able to interconnect dynamically CDN nodes from different owners is important
P2P streaming has been a hot research topic, but much applied innovation is needed
to create open services end to end.
Content handling policies (including but not limited to access) must be made
interoperable.
Forget about CAS & DRM , implement ‘open managed media’.
Long term
•
•
•
•
•
•
•
Evolve the very internet ‘fabric’ to content aware machinery.
Evolution of IP protocols and the switching machinery is heading towards
content aware routers and content aware IP stacks.
It is key to have media transport supported natively by the network so no
extensive planning needs to be done at the application layer (out of the
network).
Network should have the mechanisms to optimise the data paths and
content delivery.
Applications will be required only to especify a few QoS settings.
There is a the need to create new business models for the consumption of
internet media.
We are at a serious risk of ‘digital life monopoly’ by a few companies that
ofer appealing media services.