Transcript Active networks, (and reliable multicast)

Communication networks, active
networking and reliable
multicast
C. PHAM
Mardi 12 juin, 2001
Outline





Introduction
Nowadays network technologies
Active networking
Application: Active Reliable Multicast
Conclusions
2
Outline





Introduction
Nowadays network technologies
Active networking
Application: Active Reliable Multicast
Conclusions
3
The need for communication
4
The way people are communicating…
Internet
5
Internet milestone
Something really fast
ATM, QoS,
RVSP, DiffServ,
IPv6, MPLS…
2000
Internet 2, NG
Internet as you know it
1990
1983
1974
??
1995
ANSNET from MERIT, MCI, IBM
ARPANET has 200 transit nodes
TCP/IP for internetworking
1969
ARPANET was born with 4 transit nodes
1968
First transit node by BBN on DDP 316
1960
DoD project for a reliable, flexible network
6
User perspective of the Internet
from UREC, http://www.urec.fr
7
What it is in reality…
from UREC, http://www.urec.fr
8
Outline





Introduction
Nowadays network technologies
Active networking
Application: Active Reliable Multicast
Conclusions
9
Links: the basic element for networking

Backbone links



optical fibers
10 to 160 GBits/s with DWDM techniques
End-user access





V.90 56KBits/s modem on twisted pair
512Kbits/s to 2MBits/s with xDSL modem
1Mbits/s to 10Mbits/s Cable-modem
64Kbits/s to 1930KBits/s ISDN access
9.6KBits/s (GSM) to 2MBits/s (UMTS)
10
Routers: key elements of internetworking

Routers




run routing protocols and build routing table,
receive data packets and perform relaying,
may have to consider Quality of Service
constraints for scheduling packets,
are highly optimized for packet forwarding
functions.
11
General architecture of an IP router
IP input processing
IP output processing
IP packet
IP packet
Filter Action
Routing
agent
Forwarding
table
Packet scheduler
IP output processing
IP packet
IP packet
Packet scheduler



receives input packets,
sends packets to output buffers,
transmits packets (with QoS?).
12
Desires put on the general Internet

High-bandwidth


Ubiquity of the network access (wireless,
RTC, xDSL, mobile…)


for remaining connected everywhere
Quality of Service


for bandwidth-consuming applications
for high-quality multimedia reception
Dynamicity, adaptability

to take into account recent technologies
13
Challenges for the Internet









high-speed www
video-conferencing
video-on-demand
interactive TV programs
tele-medecine
high-performance computing, grids
virtual reality, immersion systems
distributed interactive simulations
remote archival systems…
14
The reality…(1)

High-bandwidth accesses are not available
for everybody



high-bandwidth is achievable in the core
network with optical fibers and DWDM
techniques but,
most end-users have an access ranging from
56Kbits/s to 2Mbits/s and,
it will be the case for many years!
15
The reality…(2)

An ubiquitous network access



generally implies heterogeneity and
asymmetric performances,
how to take into account this heterogeneity?
The heterogeneity of bandwidth makes
QoS


a difficult quest on an end-to-end basis,
seems that QoS is the networking forever
Graal…
16
The reality…(3)

New technologies require years to be
deployed

need for standardization



IPv6, MPLS
new services and protocols are costly to
deploy
many proprietary implementations, no
interoperability of services and new
technologies

DiffServ, TagSwitching, LabelSwitching…
17
Towards a better Internet…




Interoperability of systems
Rapid deployment of new services,
accelerating infrastructure innovation
Take into account the heterogeneity of
needs and network accesses
Customization of services, applicationoriented processing features
18
Towards the concept of…





Introduction
Nowadays network technologies
Active networking
Application: Active Reliable Multicast
Conclusions
19
What is active networks?





Programmable nodes/routers
Customized computations on packets
Standardized execution environment and
programming interface
No killer applications, only a different way
to offer high-value services, in an elegant
manner
However, adds extra processing cost
20
Motivations behind Active Networking

From the user perspective


From the operator perspective


applications can specify, implement, and
deploy (on-the-fly) customized services and
protocols
reduce the latency/cost for new services
deployment/management
From the network perspective

globally better performances by reducing the
amount of traffic
21
Active networks implementations

Discrete approach (operator's approach)



Adds dynamic deployment features in
nodes/routers
New services can be downloaded into router's
kernel
Integrated approach



Adds executable code to data packets
Capsule = data + code
Granularity set to the packets
22
The discrete approach

Separates the injection of programs from
the processing of packets
A1
A2
active code A1
active code A2
Data
Data
23
The integrated approach

User packets carry code to be applied on
the data part of the packet
data
code
data
data
data
data

code
High flexibility to define new services
24
An active router
AL packet
IP input processing
some layer for executing code.
Let's call it Active Layer
IP output processing
IP packet
IP packet
Filter Action
Routing
agent
Forwarding
table
Packet scheduler
IP output processing
IP packet
IP packet
Packet scheduler
25
Interoperability with legacy routers
APPLI
APPLI
AL
AL
TCP/UDP
TCP/UDP
IP
IP
traditional IP routing
IP
IP
AL
AL
TCP/UDP
TCP/UDP
IP
IP
26
Some open problems…

Security and integrity


Performances



how to be sure that user code are safe?
how to add active computation without
weeping out performances?
Standardization of programming interface
How to bill the CPU time?
27
Some active network applications

Customization of services


Filtering







Web-caching, on-the-fly compression/encryption
Auction, Distributed Interactive Simulations, HLA
Firewall
Congestion control
QoS
Network management
Reliable multicast
Middleware collective operation
28
Where to put active components?

In the core network?



routers already have to process millions of
packets per second
gigabit rates make additional processing
difficult without a dramatic slow down
At the edge?


to efficiently handle heterogeneity of user
accesses
to provide QoS, implement intelligent
congestion avoidance mechanisms…
29
PSTN
ISDN
xDSL
GSM, UMTS
10Mbits/s
core network
Gbits/s
Server
100Mbits/s
wireless LAN
1Mbits/s, 10MBits/s
visio-conferencing
Outline





Introduction
Nowadays network technologies
Active networking
Application: Active Reliable Multicast
Conclusions
31
Unicast
Sender

Problem


Sending same data
to many receivers
via unicast is
inefficient
Example

data
data
data
data
data
data
Popular WWW sites
become serious
bottlenecks
Receiver
Receiver
Receiver
32
What is multicast?
Sender


Not n-unicast from
the sender
perspective
Efficient one to
many data
distribution
data
data
data
data
Receiver
Receiver
Receiver
33
Multicast

History




Ethernet



Long history of usage on shared medium
networks
Data distribution
Resource discovery: ARP, Bootp, DHCP
Broadcast (software filtered)
Multicast (hardware filtered)
Multiple LAN multicast protocols

DECnet, AppleTalk, IP
from Gordon Chafee, http://bmrc.berkeley.edu/people/chaffee
34
IP Multicast Introduction

Efficient one to many data distribution



Location independent addressing


Tree style data distribution
Packets traverse network links only once
IP address per multicast group
Receiver oriented service model




Applications can join and leave multicast groups
Senders do not know who is listening
Similar to television model
Contrasts with telephone network, ATM
from Gordon Chafee, http://bmrc.berkeley.edu/people/chaffee
35
IP Multicast

Service





All senders send at the same time to the same group
Receivers subscribe to any group
Routers find receivers
Unreliable or reliable delivery
Reserved IP addresses


224.0.0.0 to 239.255.255.255 reserved for multicast
Static addresses for popular services (e.g. SAP)
from Gordon Chafee, http://bmrc.berkeley.edu/people/chaffee
36
Example: video-conferencing
from UREC, http://www.urec.fr
37
video-conferencing (2)
224.2.0.1
Multicast address group 224.2.0.1
from UREC, http://www.urec.fr
38
Multicast difficulties

At the routing level





management of the group address (IGMP)
dynamic nature of the group membership
construction of the multicast tree (pruning…)
multicast packet forwarding
At the transport level



reliability, loss recovery strategies
flow control
congestion avoidance
39
Reliable multicast

What is the problem of loss recovery?




feedback (ACK or NACK) implosion
replies/repairs duplications
adaptability to dynamic membership changes
Design goals



reduces recovery latencies
reduces the feedback traffic
improves recovery isolation
40
Reliable multicast: a big win for grids!!
SDSC IBM SP
1024 procs
5x12x17 =1020
data & program distribution
collective & gather operations
224.2.0.1
synchronization barriers
NCSA Origin Array
256+128+128
5x12x(4+2+2) =480
CPlant cluster
256 nodes
application user
Multicast address group 224.2.0.1
41
Solutions

Traditional




end-to-end retransmission schemes
scoped retransmission with the TTL fields
receiver-based local NACK suppression
Active contributions



cache of data to allow local recoveries
feedback aggregation
subcast
42
A step toward active services: LBRM
43
Active local recovery


routers perform cache of data packets
repair packets are sent by routers, when
available
data
data
data5
data1
data2
data3
data4
data5
data1
data2
data3
data4
data5
data1
data2
data3
data5
44
Active feedback aggregation

Routers aggregate feedback packets
data4
only one NACK is
forwarded to the
source by the router:
global NACK
suppression
45
Active subcast features

Send repair packet only to the relevant set
of receivers
46
Active Reliable Multicast Mechanisms

Answer general questions such as





Answer specific questions such as



is active networking beneficial for multicast?
where active components should be placed?
in what proportion?
how fast do they need to be?
what mechanisms (global vs local NAK suppression,
subcast facilities) for what performance?
scalabity of the proposed solutions?
Design of new multicast protocols
47
Network model
F active routers among N.
B receivers in a local group
2 kinds of receivers: linked and free
M. Maimour & C. Pham
48
Benefit of global aggregation on throughput
M. Maimour & C. Pham
49
Benefit of the source subcast facility
M. Maimour & C. Pham
50
Impact of active router density
M. Maimour & C. Pham
51
Conclusions

Active networks are





nice
smart
good
great…
…and we are looking at several aspects
such as



high performance environments
active QoS design
reliable multicast protocols
52
References


D. L. Tennehouse, J. M. Smith, W. D. Sincoskie, D.
J. Wetherall, and G. J. Winden. A survey of active
network research. IEEE Communications Magazine,
pages 80--86, January 1997.
L. Wei, H. Lehman, S. J. Garland, and D. L.
Tennenhouse. Active reliable multicast. IEEE
INFOCOM'98, March 1998.

M. Maimour, C. Pham. A Throughput Analysis of
Reliable Multicast Protocols in an Active Networking
Environment. TR. http://resam.univlyon1.fr/~cpham/Paper/TR/TR01-2000.ps.gz
53
Web links

ANTS


Tamanoir and active reliable multicast


http://wind.lcs.mit.edu/activeware
http://resam.univ-lyon1.fr
Active Networking in France

http://www.loria.fr/~festor/raf/raf.html
54