On the Efficiency of Collaborative Caching in ISP-aware P2P Networks

1
On the Efficiency of Collaborative Caching
inISP-aware P2P Networks

• Jie DaiHai Jin et al.
• H.K.U.S.T. U.T. H.U.S.T.
• IEEE Infocom, Shanghai, China, April 10-15, 2011

Presenter Su Hu
2
Warm-up

• P2P Overlay network

For the first 4 downloaded pieces, the pieces are
selected at random
3
Warm-up

• Challenges ISPs
• Tremendous data volume
• Costly inter-ISP traffic
• 1) Not at the same layer

P2P Overlay
Application data
Internet access service
ISP Underlay
4
Warm-up
Profit

• Challenges ISPs
• 2) Users pay for bandwidth, why throttling ?
• Shared bandwidth

local loop,
bandwidth definition,
ADSL architecture
5
Outline

• Warm-up
• Abstract
• Introduction
• Related Work
• Inter-ISP Traffic Model Cache Allocation
• Improving Cache with ISP Peering Agreement
• Performance Evaluation
• Conclusion
• Summary

a, q, ß, ?, ISP index etc.
6
Abstract

• Why Collaborative Cache
• Reduce the inter-ISP traffic
• Existing design ignores
• Dynamic P2P traffic patterns, ISP peering, cache
  server capacity .
• Analysis of resource allocation with awareness of
  Inter-ISP traffic and ISP policies

7
Abstract

• Our work
• Characterize inter-ISP traffic patterns
• Develop cache allocation framework focus on
  minimizing inter-ISP traffic.
• Incorporate both locality-aware/unaware ISP
  peering agreements
• The research help us understand
• Traffic characteristics of existing P2P
• Design of collaborative ISP cache mechanisms

8
Outline

• Warm-up
• Abstract
• Introduction
• Related Work
• Inter-ISP Traffic Model Cache Allocation
• Improving Cache with ISP Peering Agreement
• Performance Evaluation
• Conclusion
• Summary

9
Introduction

• Background 1 The Tussle
• P2P 70 of the Internet traffic
• Can ISP throttle P2P packets?
• ISP want to maintain customer bases
• Background 2 How to resolve it
• Disparity
• Locality-aware peer selection P4P TopBT
• vulnerable due to the dynamic of P2P

Proximity-driven biased neighbor select
10
Introduction
Reduce access latencies to web page

• Our Solution- caching
• Web cache
• Collaborative Caching lead to win-win
• Inter-ISP
• Experiences of User

Cache for P2P
Redirect traffic to cache server at edges of ISP
Reduce the latency of P2P packet
11
Introduction

• New Characteristics from web cache
• Mitigate the inter-ISP traffic
• Inter-ISP traffic pattern, collaboration between
  P2P ISP
• Cache server resource allocation
• ISP peering agreements

Both Storage (cache hit ratio) bandwidth
(servers uploading capacity) constraints are
important.
The collaboration between ISPs over the public
Internet corresponding cache server
12
Introduction
Video distribution platform

• Propose a Optimization framework
• Theoretical model of i-ISP traffic
• Resource allocation scheme
• The effects of ISP peering on our solution
• Collaborative cache scheme tailored to ISP
  peering

ISP scales , channel popularity
Reduce i-ISP traffic both locality-aware/unaware
peer selection
Positive on mitigation i-ISP traffic
13
Outline

• Warm-up
• Abstract
• Introduction
• Related Work
• Inter-ISP Traffic Model Cache Allocation
• Improving Cache with ISP Peering Agreement
• Performance Evaluation
• Conclusion
• Summary

14
Related Work

• 3 classes of ISP-friendly design
• Peer-driven
• PPLives latency based mechanism, TCP ping
• ISP-driven
• P4P ISP advertise preferred paths to P2P app.
• Why ISP caching?
• Not impair the P2P robustness
• Transparent to end user
• Upon locality-aware system

15
Related Work

• Existing P2P cache design
• Focus on independent server cache
• Improving the byte hit ratio
• Ignore ISP collaboration cache server bandwidth
  constraint
• Existing collaborative cache design
• Dans work
• Rate allocation among cache servers
• Ignore inter-ISP traffic model, practical
  constraints in real P2P

This paper inter-ISP traffic model, server
storage and bandwidth constraints ,peer
selection, ISP peering
16
Outline

• Warm-up
• Abstract
• Introduction
• Related Work
• Inter-ISP Traffic Model Cache Allocation
• Improving Cache with ISP Peering Agreement
• Performance Evaluation
• Conclusion
• Summary

17
I-ISP Traffic Model Cache Allocation

• Inter-ISP traffic model
• P2P video streaming
• locality-aware locality-unaware
• Optimization framework of allocation resource
• Inter-ISP traffic mitigation
• Two sets of server strategies
• Collaboration between P2P app. cache server
  

18
I-ISP Traffic Model Cache Allocation

1. Inter-ISP traffic model

Assume streaming length is same, only depend on
streaming rate

• Notation
• P2P video streaming

video channels
number of concurrent users in P2P v system
number of concurrent users in video channel i
Assume peer out-degree equals in-degree
streaming rate of video channel i
size of video channel I
in-degree of individual peers
19
I-ISP Traffic Model Cache Allocation

1. Inter-ISP traffic model

• Notation
• Existing ISPs
• ISP1 is most popular, ISPk is lest
  popular

number of ISP in which peers view video ?
Storage capacity by cache server in ISP k
uploading bandwidth by cache server in ISP k
percentage of channel i stored in c server in
ISP k
uploading bandwidth to channel i by c server in
ISP k
number of concurrent users of channel i in ISP
k
20
I-ISP Traffic Model Cache Allocation

1. Inter-ISP traffic model

Probability that any user view channel i

• Channel popularity distribution
• 
  
• q
• i
• ISP user distribution
• ß 0, same user amount each ISP
• higher the ß, more unbalanced the ISP user
  

(1)
P2P object be accessed over long term
Zipf-Mandelbrot distribution
the probability
the probability
Probability that any user is in ISP k
(2) ß different scenarios of ISP user
populations
21
I-ISP Traffic Model Cache Allocation

1. Inter-ISP traffic model

• 
  
• Inter-ISP traffic rate model (n-c)
• 1. Locality-unaware peer selection
• mnumber of neighbor in same ISP
• Hyper-geometric distribution

(3)
(4)
Evenly selected, Neighbors decides mainly by ISP
user numbers
22
I-ISP Traffic Model Cache Allocation

1. Inter-ISP traffic model

M defectives in N, extract n samples, and the
probability of k defectives

• H(n , M , N)
• p(xk) C(k , M) C(n-k , N-M) / C(n , N)
• k max(0 , n-NM) , , min(n , M)
• N xi M xik n din

p2p streaming server is the
external sources.
(5)
23
I-ISP Traffic Model Cache Allocation

1. Inter-ISP traffic model

• Inter-ISP generate by channel I in ISP k
• 1) more popular channel more inter-ISP traffic
• 2) ISPs have similar scales,
• 3) ISPs have widely different scales,

(6)
24
I-ISP Traffic Model Cache Allocation

• Inter-ISP traffic rate model (n-c)
• 2. Locality-aware peer selection
• number of persistent external links

1. Inter-ISP traffic model

Give priority to nearby peer (evaluate by the ISP
peer in)
i-ISP traffic per peer
i-ISP traffic per peer
(7)
25
I-ISP Traffic Model Cache Allocation

1. Inter-ISP traffic model

• Locality-aware

• Locality-unaware

30
5-10

1. 80, both have similar inter-ISP
   traffic
2. -gt 0 , both coefficients values
   -gt 1
3. the left coefficients is always larger than the
   right

26
I-ISP Traffic Model Cache Allocation

1. Cache resource allocation mechanisms

Peers in any channel are evenly distributed along
the channel ?

• Inter-ISP traffic rate for ISP k

(8)

• Minimize
• Subject to

• Maximize
• Subject to

(10)
(9)
27
I-ISP Traffic Model Cache Allocation

1. Cache resource allocation mechanisms

• Theorem 1
• For max i-ISP mitigation, optimal resource
  allocation

(11)
(12)
(13)
28
I-ISP Traffic Model Cache Allocation
Continuous knapsack, solution Non-decreasing
with index Use greedy algorithm , give storage
as needed for channel with higher priorities,
(11) Achieve upper of as
min ( , ) using (12) , (13)

1. Cache resource allocation mechanisms

• Theorem 1
• Proof
• Maximize
• Subject to

(14)
29
I-ISP Traffic Model Cache Allocation

1. Cache resource allocation mechanisms

• Theorem 1 Remark
• Design guidelines of collaborative cache
  mechanism
• P2P system parameters
• number of users, channel popularity, file
  size, streaming rate of channel
• ISP cache server needs to collaborate with P2P
  app.

Reduce end-to-end latencies, Mitigate i-ISP
prevents throttling by ISP
Precisely indentify the content requests of P2P
packets needs help of P2P app.
30
I-ISP Traffic Model Cache Allocation

1. Cache resource allocation mechanisms

• Algorithm 1
• Optimization-based Collaborative Cache framework
  for i-ISP mitigation
• P2P app. actively transmits system states to ISP
  cache server.
• Compute , , allocate ,,
  as ,
• Cache server cut request to external, if average
  uploading rate to channel , satisfy the
  request
• Monitor P2P states, adjust resource according to
  T1.

Population-based I, Concurrent users x.
31
Outline

• Warm-up
• Abstract
• Introduction
• Related Work
• Inter-ISP Traffic Model Cache Allocation
• Improving Cache with ISP Peering Agreement
• Performance Evaluation
• Conclusion
• Summary

32
Improve Cache with ISP Peering Agreement

1. ISP Peering Agreements

• Concept
• ISPs provide free connectivity to transit user
• Alleviate costly transit traffic
• 2 positive outcomes
• Large group of traffic-free candidate neighbor
• Strategically select P2P content to store and
  deliver
• ISP peering relation is Reflexive Symmetric

Free i-ISP traffic is not need to cache
(15)
symmetric Matrix E
33
Improve Cache with ISP Peering Agreement
Only peers in peering ISP help to mitigate i-ISP
traffic, no collaboration between cache servers

1. Impact of ISP Peering

(5)

• Not-full collaboration between peering ISPs
• Cache server not deliver to peers of peering ISP
• Locality-unaware peer selection

(16)
(17)
34
Improve Cache with ISP Peering Agreement
Compared to (6), here need to
also subtract the probability of being peering ISP

1. Impact of ISP Peering

• Not-full collaboration between peering ISPs
• Locality-unaware peer selection (cont.)
• Locality-aware peer selection

(18)
i-ISP traffic per peer
i-ISP traffic per peer
Multiply not
(19)
35
Improve Cache with ISP Peering Agreement

1. Impact of ISP Peering

• For both scenarios i-ISP traffic reduced due to
  expansion of free neighbor candidates.

(18)
(19)
36
Improve Cache with ISP Peering Agreement

1. Improving cache with ISP Peering

• Full collaboration between peering ISPs
• The bottleneck
• One ISPs cache server cant store whole P2P
  object
• -- Cache server bandwidth utilization
  insufficient
• Peering combine of global cooperative cache
• Peering-based full collaboration
• Upload rate for i rate of i-ISP can be
  intercept( )

Cache server not only serve for peers in own ISP,
but also to peering ISPs
bandwidth assigned by to for channel i
lt------
37
Improve Cache with ISP Peering Agreement

1. Improving cache with ISP Peering

Any request to i can be served
if sufficient bandwidth

• Full collaboration between peering ISPs
• Maximize
• Subject to

(20)
Upper bound, Centralized solution,
inappropriate for practice
Peering, resource, limit aik to serve max
, propose a distributed collaborative cache
scheme in algor 2
(21)
38
Improve Cache with ISP Peering Agreement

• Algorithm 2
• An ISP Collaboration-based Distributed Cache
  framework for i-ISP mitigation
• Cache server announce surplus bandw and storage
  to peering ISPs.
• After announce of , sorts channel in
  descending order of ,first
  channel , , bandw request to
• Upon receive r from , allocates and
  confirm
• After confirm of , evicts content confirm,
  reallocate to such , broadcast
  surplus info to peering ISP.

39
Outline

• Warm-up
• Abstract
• Introduction
• Related Work
• Inter-ISP Traffic Model Cache Allocation
• Improving Cache with ISP Peering Agreement
• Performance Evaluation
• Conclusion
• Summary

40
Performance Evaluation

1. Trace-Driven Analyses

• Statistical result of measurement on UUSee
• Number of channels 993
• (channel 100 has 100 users at peak time)
• Number of concurrent users 100000
• To fit the cure of peak time users
• a 0.78 q 4 30 ? 5
• Evaluation of Inter-ISP Traffic Pattern
• Factors P2P content popularity, ISP popularity
• L-A(locality-aware)
  L-U(locality-unaware)

41
Performance Evaluation

1. Evaluation of Inter-ISP Traffic Pattern

Fig.1.
42
Performance Evaluation

1. Evaluation of Inter-ISP Traffic Pattern

?
Fig.2.
43
Performance Evaluation
Fig.3.
44
Performance Evaluation

1. Evaluation of Collaborative Cache Mechanisms

Fig.4.
45
Performance Evaluation

1. Evaluation of Collaborative Cache Mechanisms

Fig.5.
46
Performance Evaluation

1. Evaluation of Collaborative Cache Mechanisms

Fig.6.
47
Performance Evaluation

1. Evaluation of ISP Peering Agreements

• 10 3 Peering Scenarios
• 1) Scenario 1
• 1/2 3/4 9/10 extreme unbalanced
• 2) Scenario 2
• 1/6 2/7 5/10 still has original
  property
• 3) Scenario 3
• 1/10 2/9 5/6 extreme balanced

48
Performance Evaluation

1. Evaluation of ISP Peering Agreements

Fig.7.
49
Performance Evaluation

1. Evaluation of ISP Peering Agreements

Fig.8.
50
Performance Evaluation
About percentage of 10 ISPs, so it cant reach 1

1. Evaluation of ISP Peering Agreements

Fig.9.
51
Outline

• Warm-up
• Abstract
• Introduction
• Related Work
• Inter-ISP Traffic Model Cache Allocation
• Improving Cache with ISP Peering Agreement
• Performance Evaluation
• Conclusion
• Summary

52
Conclusion

• Propose an inter-ISP traffic model
• Develop a cache resource framework under resource
  constraint and peering agreement
• Put forward guidelines for cache storage and
  bandwidth allocation design
• Strategy to improve collaborative cache under ISP
  peering
• Future work improving user experience

53
Summary

• Review P2P overlay and challenge with ISP
• Review other existing ISP-friendly design
• Give the notation used in this slide
• Propose the inter-ISP traffic model
• Give the Cache resource allocation mechanisms
• Improve cache mechanisms with ISP peering
• Evaluation of our collaborative cache mechanism

54
(No Transcript)
55
Good Points

• Propose the probability model, summarize the
  formulation of traffic under every strategy,
  formulate the optimization problem
• Rational performance analysis based on experience
  data
• Next how to improve and implement it?

56
Thank you!