Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
1
1
FG INET: Intelligent Networks
An-Institut Deutsche Telekom Laboratories
Prof. Anja Feldmann, Ph.D.
[email protected]://www.net.t-labs.tu-berlin.de/
2
INET: Research Group
r Locationm Telefunkenhochhaus, 16. Stock
r Office hoursm Tuesday 12:30 – 13:00
m After the lecture or per e-mail
r Contactm Best per e-mail ☺
r Teaching contactm Nadi Sarrer / Oliver Hohlfeld
r Web site
http://www.net.t-labs.tu-berlin.de/
2
3
T-Labs
r Institute at TU Berlin funded by Deutsche Telekom AG
r Co-locates researchers from TU Berlin and project managers of Deutsche Telekom
r Two partsm Strategic research
• 7 Research Groups + 30 Researchers + 40 Ph.D. studentsso far: INET, Quality and Usability (QU Lab), Security in Telecommunications (SecT), Service Centric Networking (SNET)
m Innovation development• Project work, e.g., IP-TV, Overarching AAA
4
Dr. Stefan Schmid
r Biography: m PhD in Distributed Computing Group (DISCO) @ ETH Zürich
m Postdoc at Chair for Efficient Algorithms (TU München) and Chair for Theory of Distributed Systems (Uni Paderborn)
m Senior Research Scientist @ Deutsche Telekom Laboratories
r Research Interests:m Robust and scalable architectures for distributed systems
(dealing with churn, self-stabilization, etc.)
m Network virtualization (e.g., embeddings or service migration)
m (Distributed) algorithms
m Peer-to-peer systems (e.g., BitThief), wireless networks (e.g., robust medium access), social networks (e.g., Wuala)
m Incentives, mechanism design, game theory
3
5
Example: Network Virtualization
r Virtual networks
m Decoupling of services from resource infrastructure
m Virtual networks embedded dynamically,
where and when most useful!
m No knowledge of future requests:
online embedding algorithms needed!
r Use case: Agile mobile provider
m Move servers “closer to requests”
m Improves Quality-of-Service…
m ... but need to take into account migration costs!
m How to be „competitive“?
m We also develop prototypes…
6
Florin Ciucu, Ph.D.
r Biography:
m B.Sc. in Informatics, Faculty of Mathematics, University of Bucharest
m M.Sc. in Computer Science, George Mason University, USA
m Ph.D. in Computer Science, University of Virginia, USA
m Senior Research Scientist at Deutsche Telekom Laboratories
r Research Interests:
m Performance analysis of computer and communication systems
• Overflow and delay analysis in queueing networks
• Optimal resource allocation (e.g., buffer size in routers)
• Capacity of multi-hop wireless networks
m Randomized algorithms
• Medium Access Control (Aloha, CSMA/CD)
• Scheduling in multi-hop wireless networks
• Traffic policing
4
7
Example: Buffer Size in Routers
r Buffers needed to handle the frequent situation
r Interesting tradeoff: Large vs. Small
r Topics of interest
m The impact on end-to-end multimedia quality (e.g., video)
m How small buffers should be? (Nobody knows after 40 years)
yessmallsmalllargeSmall
nolarge, e.g,
$15000largesmallLarge
All-Optical Networks
CostsDelaysLossBuffer Size
incoming_data_rate > router_data_rate
8
Current research topics
r ISP-application collaborationm Content aware traffic engineering
m Caching and content distribution networks
r Programmable networksm Berlin Open Wireless Mesh, OpenFlow,
Software defined radios
r Broadband access evolutionm Licensed/unlicensed spectrum, Mobile user
experience
r Community-inspired optimization
r Cloud Networking
5
9
Internet and traffic engineering
Source: Arbor Networks 2009
10
Internet and traffic engineering
Source: Arbor Networks 2009�Offline Process
Adjust routing or peering, dimension the network
Traffic Engineering:
6
11
The new Internet
Source: Arbor Networks 2009
�New core of interconnected content and consumer networks
12
The new Internet
Source: Arbor Networks 2009
�New core of interconnected content and consumer networks
Google, Google,
AkamaiAkamai, ,
RapidSharRapidShar
ee, , ……
7
13
The new Internet
Source: Arbor Networks 2009
�New core of interconnected content and consumer networks
Moving Target I:
Popular Applications
14
The new Internet
Source: Arbor Networks 2009
�New core of interconnected content and consumer networks
Moving Target I:
Popular Applications
Moving Target II:
Bottlenecks
8
15
The new Internet
Source: Arbor Networks 2009
�New core of interconnected content and consumer networks
Moving Target II:
Bottlenecks
�ISPs lost control of their traffic
Moving Target I:
Popular Applications
16
The new Internet
Source: Arbor Networks 2009
�New core of interconnected content and consumer networks
Moving Target II :
Bottlenecks
�ISPs lose control of their network
Moving Target I:
Popular Applications
“Telekom’s chief executive,
said Google and others should
pay telecoms groups for carrying
content on their networks”
9
17
Challenge
Content-aware Traffic Engineering
ISPs re-gain control of their traffic by biasing host selection
18
Improving content access timeCase study: CDN
10
19
Client
External
DNS
Provider
DNS
Internet Service Internet Service
ProviderProvider
(ISP)(ISP)
Host
1
2
3
4
6
PaDISPaDIS
5
Provider-aided Distance Information System
PaDIS
20
Client
External DNS
Provider
DNS
Internet Service Internet Service
ProviderProvider
(ISP)(ISP)
Host
1
2
3
4
6
PaDISPaDIS
5
Full View Full View
of the ISP of the ISP
NetworkNetwork
Content can be
downloaded from any eligible host!
PaDIS
11
21
ISP-Application cooperation
r Insight: ISP knows its network
m Node: bandwidth, geographical location, service class
m Routing: policy, OSPF/BGP metrics, distance to peers
r PADIS concept
m Service of AS / ISP
m Input: list of possible dst IPs
m Ouput: ranked list of dst IPs
• E.g., according to distances between src IP and dst IPs
r Applicable, whenever there is a choice!
22
Teaching
r Lectures (Vorlesungen)
r Seminars (Seminare)
r Lab courses (Praktika)
r Projects (Projekte)
r Theses (Diplom / Master / Bachelor)
r PGT: Project Group Meeting (Doktorandenseminar)
r NLS: Networking Lecture Series (external visitors)
12
23
Lectures
r Network protocols and architectures (VL+UE)
m Design principles of the Internet ...
m Base for all other INET classes
r Internet Routing (VL)
r Internet Security (VL)
r Internet Measurement (VL)
24
Seminars
r Internet Routing
r Internet Measurement
r How
m Topics: Current research papers
m Task:
• Summary paper + presentation
• Participation in discussion during the seminar
m En block after the end of term
13
25
Lab courses
r Hands on exercises
r Mesh Lab
m Understanding various wireless concepts (e.g., interference, MAC layer, multi-hop routing) Experiments with mesh routers in the BOWL indoor network
r Router Lab
m Configuring and managing networks
m Internet experiments in a Lab
26
Projects and theses
r Topic:m See Web pages
m Talk to members of INET
m Suggest your own topic
r Work flow:m Literature/background search
m Presentation of idea at students‘ talks series(20 minutes ☺)
m Execution of idea / preparation of thesis document
m Presentation of results at students‘ talks series(20 minutes)
14
Teaching at INET / Anja Feldmann
27Fachgebiet INET – Prof. Anja Feldmann – www.net.t-labs.tu-berlin.de
NPA: Network
Protocols &Architectures(VL+UE) 6LP
SE: Routing(SE) 3LP
RL: RouterLab(PR) 6LP
ML: MeshLab(PR) 6LP
PJ: Projekt(PJ) 6LP or 12LP
IR or IS or IM:
Internet Routing, Security,
Measurement(VL) 3LP
Bachelor Thesis (12LP)
Master Thesis (30 LP)
Ba
ch
elo
r
0-1
5 L
P
Ma
ste
r
12
-27L
P
WiSem SoSem
Course Overview
Example Module Paths
NPA (6LP)
IR, IS, IM, SE
(6/9LP)
ML or/and RL
(6/12 LP)
NPA (6LP)
SE+PJ (9LP)
IR, IS, IM, SE
(6/9LP)
ML or/and RL
(6/12 LP)
NPA (6LP)
IR, IS, IM, SE
(6/9LP)
ML or/and RL
(6/12 LP)
Topics
Internet Protocols:
� Routing and IP� Transport (TCP/UDP)� Applications� Future Internet
� Traffic Measurement
� Workload Modeling� Wireless� Performance Analysis� Network Security
SE: Measure-ment (SE) 3LP