Transcript ppt file

Meta-Headers: Top-Down
Networking Architecture with
Application-Specific Constraints
Murat Yuksel
University of Nevada, Reno
Reno, NV
[email protected]
http://www.cse.unr.edu/~yuksem
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
1
Motivation: The trends

The variety of applications possible is
increasing, especially in wireless




wireless peer-to-peer, mobile data, community
wireless
The size is increasing:

million-to-billion nodes

vehicular networks, sensor networks, MANETs
The dynamism is increasing:
What is unavoidable?: More dynamism, more
disruption tolerance, more entities, and more
varieties
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
2
Motivation: The big picture




Mobile, ad-hoc, dynamic
Unstructured
Cross-layer & layered invariants
Presentation
Application
Application
Transport
Transport
Transport
(TCP, UDP)
(TCP, UDP)
Network
Network
Data Link
Data Link
Session
Physical
(a) OSI
(IP)
(Ethernet 802.3)
Network & MAC
(IP, Mobile IP,
802.1x)
Physical
Physical
(Fiber, Cable)
(RF, Fiber, Cable)
(b) Wireline
(c) Wireless
Application-Specific
Application
Application
?
Network
& Routing
?
Physical
Hardware-Specific

Static
Structured
Layered invariants
Network-Specific

(RF, FSO, Fiber, Cable)
(d) MANET, peer-to-peer
Economics always has the bigger force: economically attractive
applications will keep forcing more vertical components into the stack!
We need a systematic way of implementing vertical components
to avoid an unhealthy monolithic stack architecture.
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
3
Motivation: Response to the trends

Wireless research has been responding with




optimizing via cross-layer designs
adding custom-designed vertical components to the
stack
Old hat: layered vs. cross-layer tradeoff
Bottom-up cross-layer has been the main
approach

Scarcity of wireless resources dominated the
economics
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
4
Motivation: Response to the trends

A paradigm shift: wireless resources are
becoming massively available






Community wireless
WiFi hotspots
Google WiFi, AT&T Metro WiFi
Spectrum resources may still be scarce but
connectivity is already ubiquitous
The key metric to optimize is becoming
application utility rather than the wireless
resources
App-specific vertical designs are needed..
We need top-down cross-layer designs in addition to the
traditional bottom-up ones.
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
5
Why not continue merging layers?

Merging layers:



Which layers must be absolutely isolated?


A greedy approach
Makes it hard to standardize – bad for sw engineering
Application, Network, Physical?
Integrating lower level functions with a higher
layer function will prevent them becoming a
substrate for other higher layer protocols

Cellular provisioning in the US – jailbreaks
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
6
Motivation:
Application Layer Framing (ALF)

Layering was a main component of the e2e architecture..
“a major architectural benefit of such isolation is that it
facilitates the implementation of subsystems whose
scope is restricted to a small subset of the suite’s
layers.”
Clark and Tennenhouse, SIGCOMM’90

But, Integrated Layer Processing (ILP) was there too!



To achieve better e2e efficiency and resource optimization
ILP never become a reality due to the lack of a systematic way
of doing it.
An ALF-based approach is needed:
network protocol services at lower layers can best be
useful when applications’ characteristics and intents are
to theMiami,
lower
IEEEconveyed
GLOBECOM FutureNet,
FL, layers.
Dec 2010
7
Meta-Headers: A vertical design tool

A packet meta-header:




vertically travels across the network stack
establishes a vertical communication channel among
the traditional layers
co-exist with the traditional per-layer packet headers
Applications can communicate their intent
across all the protocol layers by attaching the
meta-headers to data.
<meta-headers, message>
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
8
Headers vs. Meta-Headers
Application-specific packet
meta-headers
message
Application
H4 MH4 MH3 MH2 MH1
message
Layer 4
H3 H4 MH4 MH3 MH2 MH1
message
Layer 3
H2 H3 H4 MH4 MH3 MH2 MH1
message
Layer 2
H1 H2 H3 H4 MH4 MH3 MH2 MH1
message
Layer 1
message
Application
MH1 MH2 MH3 MH4
message
Layer 4
MH1 MH2 MH3 H4
message
Layer 3
MH1 MH2 H3
H4
message
Layer 2
MH1 H2
H3
H4
message
Layer 1
H3
H4
message
Explicit Meta-Headers
Traditional packet headers
Application-specific packet
meta-headers
H1
H2
Implicit Meta-Headers
Traditional packet headers
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
9
H4 MH4 MH3 MH2 MH1
Protocol 2
message
Layer 4
Layer 3
H3
H4 MH4 MH3 MH2 MH1
message
H4 MH4 MH3 MH2 MH1
message
Layer 4
Layer 3
H3
H4 MH4 MH3 MH2 MH1
Service 1
message
Demultiplexing with
meta-headers
Protocol 1
Demultiplexing with
traditional headers
Meta-Headers: Demultiplexing
Service 2
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
10
Informing Applications about Lower
Layer Services


How will upper layers know about the service
primitives of the layers lower than the one
below?
Reactive – Meta-Headers in Reverse Direction



detect lower layer services in an on-demand manner
as connections arise
meta-headers rewritten by lower layers in reverse
direction
Requires a closed-loop – connectionless or multireceiver services may not work
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
11
Informing Applications about Lower
Layer Services (cont’d)

Proactive – Pre-informed Designer




inform layer k designers about services of layers k-2
and below apriori
too much complexity as the number of lower layer
services increases – rank ordering might help
May not be desirable by ISPs
Regional service discovery via broadcasting –
connectionless
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
12
End-to-End Coordination
1
Applicationspecific packet
meta-headers
Application at
source prepares
meta-headers with
default options and
sets flags to probe
for available
services
message
Application
MH1 MH2 MH3 MH4
message
Layer 4
MH1 MH2 MH3 H4
message
Layer 3
MH1 MH2 H3 H4
message
Layer 2
MH1 H2 H3
H4
message
Layer 1
H1 H2 H3 H4
message
Traditional
packet headers
SOURCE
2
5
Application at
source readjusts
meta-headers for
joint vertical
optimization of
end-to-end
performance.
4
Feedback loop for
conveying end-to-end
multi-hop L1-L4 services,
possibly as a sequence of
options over multiple hops.
Optional
feedback
loop for
conveying
available
L1-L3
services
Optional feedback
loop for local
optimization of
last hop(s) of the
end-to-end path.
Meta-headers are filled
with summary of
available end-to-end L1L4 services, and fed back
to the source application.
Application
MH1 MH2 MH3 MH4
message
Layer 4
MH1 MH2 MH3 H4
message
Layer 3
MH1 MH2 H3 H4
message
Layer 2
MH1 H2 H3
H4
message
Layer 1
H1 H2 H3 H4
message
DESTINATION
3
Meta-headers may
or may not get
converted to
traditional headers.
Meta-headers are filled
with available L1-L3
services, and
optionally fed back to
the source application.
MH1 MH2 MH3 H4
message
Layer 3
MH1 MH2 H3 H4
message
Layer 2
MH1 H2 H3
H4
message
Layer 1
H1 H2 H3 H4
message
ROUTER
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
A dynamic
end-to-end
negotiation..
13
An optimization perspective
Meta-header
probes
questing lower
layer services
Meta-headers
filled with
available
services
Application-Specific Application-Specific
View of the Network Constraints
E
Vertical
optimizations are
possible
B
(application-based
cost)
Lagrange
multipliers
(pieces of Q2
and Q3)
(quality
constraints)
Top-Down Value Choice
Optimization Framework
More dynamic
Meta-headers as
Lagrange
multipliers
Application
Lagrange
multipliers
(pieces of E)
Q3
Q2
(per-layer
state)
W3
(per-layer
state)
Network State
Information
Value
Choices
(implicit)
(per-layer
constraints)
Network Resource
Constraints
W2
(implicit)
(per-layer
constraints)
Network
Link State
Information
Link Resource
Constraints
Links
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
14
Summary




A top-down networking architecture with metaheaders
Vertical optimizations at finer temporal and spatial
granularity
A variety of top-down optimizations:
 Top-down routing (layers 5, 3)
 Top-down QoS/value management (layers 5, 3, 2)
 Top-down dynamic transport (layers 4, 3, 2)
A new class of optimization problems aiming to
improve joint performance of multiple layers while
respecting the isolation among them.
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
15
THE END
Thank you!
This work is supported in part by the U.S. National Science
Foundation awards 0721600 and 0721609.
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
16
An optimization perspective
Vertical optimizations
are possible:
More dynamic
Meta-headers as
Lagrange multipliers
Meta-header
probes
questing
paths
Application
Application-Specific
View of the Network
Application-Specific
Constraints
E
B
Lagrange
(path quality multipliers
constraints) (pieces of Q)
(application-based
path costs)
Lagrange
multipliers
(pieces of E)
Meta-headers
filled with
available paths
Top-Down Routing
Optimization Framework
Routing
Choices
W
Q
(implicit)
(link states or
path-vectors)
(link weights)
Network Topology
Information
Network Resource
Constraints
Network
IEEE GLOBECOM FutureNet, Miami, FL, Dec 2010
17