Cache Me If You Can

1
Cache Me If You Can
2
You Are Here
3
Cache Proxies for Web
A
4
Hierarchical Caching
B
A
5
Cooperative Caching
A
B
6
Distributed Caching
A
B
7
Streaming Media vs. Web Pages
8
Video Access Pattern

• by S. Acharya and B. Smith
• Study at Lulea University, Sweden
• 55 complete, 45 stop very early
• High temporal locality

9
Bimodal Distribution
10
Zipf Distribution
log (Frequency)
Frequency C/rankm
log (rank)
11
Zipf Law

• Frequency of English Words
• the of to a
• City size

12
Benefits of Caching
13
Reduce Access Latency
)
(
14
Reduce Server Load
15
Buy Some Time
16
Reduce Client Buffer Size
17
Type of Cache
18
Types of Cache

• Static caching
• Dynamic caching
• patching buffer

19
Static Caching

• Cache all or nothing is bad
• Split video into segments of size S

20
If I am writing a survey

• Introduction
• Differences between web and media caching
• Video access pattern
• Benefits of caching
• Different type of cache
• Proxy caching architecture

21
Other Caching Issues

• How to segment?
• Which segments to cache?
• When to fetch data?
• Who to fetch from?
• When cache is full, who to kick out?
• How to measure popularity?
• Can cache adapt to popularity?

22
Rest of the Lecture

• Soccer cooperative caching proxy
• Silo, rainbow and cache token distributed
  caching proxy
• Mocha with congestion control

23
SOCCER

• Self-organizing cooperative caching architecture

24
How to segment?
25
Effects of Size S

• Large S Low utilization
• Small S Lots of gaps

26
Prefix Caching Policy

• 1 Chunk k segments

27
Fetching Cache Data
Server
B
A
Client 2
Client 1
28
Option 2
Server
B
A
Client 2
Client 1
29
Option 3
Server
B
A
Client 2
Client 1
30
Issues

• How to advertise?
• How to choose helper?

31
How to Advertise?

• Balance between
• network load
• freshness of information

32
Scalable Advertisement

• Expanding Ring Advertisement

16 32 64 128
33
How to Choose Helper?

• Consideration for Static Cache
• network distance (1,2,3,4)
• number of streams being served

34
Static Cache

• Cost for retrieving data from node X to node Y
  

35
Dynamic Caching

• Need to consider the size of patching buffer

S
36
Dynamic Caching

• Nused
• Nfetched min(s, Nused ?)

S
37
Dynamic Cache

• Potential Waste Nfetched/Nused
• Integrate into overall cost

38
Helper Selection

• Helper ask other helpers to help if segments not
  found on local cache
• Pick lowest cost helper

39
If Static Cache

• Retrieve to fill current gap
• May pick different helpers for different gaps
• Should avoid frequent switching

40
If Dynamic Cache

• Stick to one helper

41
Evaluations
Single Proxy
Hierarchical Proxies
SOCCER
42
Silo, Rainbow and Cache Token
43
Distributed Caching
Server
B
A
new clip!
44
Distributed Caching
Server
B
A
45
Distributed Caching
Server
B
A
46
Distributed Caching
Server
B
A
47
Cooperative vs Distributed

• Cooperative caching caches independently
• Distributed caching caches as a team

48
Problems

• Who should cache what?
• Which segment to kick out?
• How to redistribute data?

49
RCache

• Np proxies
• video of length Lv
• divide into Ns equal segments
• Each proxy caches each segment with a/Np
  probability

50
Analysis

• Probability that whole video is cached
• Storage requirement aL

51
Two Parameters

• How to divide video into segments?
• What is the probability of storing each segment?

52
Silo
probability of storage
segment size
53
Problems

• Who should cache what?
• Which segment to kick out?
• How to redistribute data?

54
Compute Popularity

• For each video
• For each segment

55
Rainbow Algorithm
less popular
56
Problems

• Who should cache what?
• Which segment to kick out?
• How to redistribute data?

57
Data Redistribution

• When popularity changes, need to redistribute.
• Redistribute on-demand (lazy)

58
Cache Token

• Each segment have two bits
• (T,C)
• T I am suppose to have the segment
• C I have the segment

59
Data Redistribution

• (T1,C1)
• (T0,C0)
• (T1,C0)
• (T0,C1)

60
Silo, Rainbow and Token

• Distributed Caching Architecture
• Fault tolerant
• Scalable

61
Proxy Cache Congestion Control
62
Scenario (10am)
Server
A
Client 2
63
Scenario (1am)
Server
A
Client 2
64
Problem

• Cache interfere with congestion control algorithm
• Make cache adapt aware of quality adaptation

65
Congestion Control Protocol

• RAP AIMD Layered Coding
• sender-driven layered unicast

66
Solutions

• Making cache quality-aware
• Prefetch
• Replacement Algorithm

67
Cache Miss
Server
A
Client 2
68
Cache Hit
Server
repair prefetch
A
Client 2
69
Peeking Inside the Cache
70
Cache Hit Repair
71
Cache Hit Prefetch
72
Cache Hit Prefetch
73
Prefetch Algorithm
playback point
prefetch window
74
Proxy Request to Server

• Multiple requests (for different clients) are
  batched.

75
Server Response

• Missing segments are sent in decreasing priority

3
4
2
1
76
Server Response

• Send as many segments as possible until next
  prefetch request

77
Solutions

• Making cache quality-aware
• Prefetch
• Replacement Algorithm

78
Goal of Replacement Algorithm

• Goal converge to efficient state
• if a stream is popular
• average quality is high
• variation in quality is low

79
The Algorithm
80
Trashing and Locking
81
Choosing Victim

• whit (weighted hit)
• Tplay/Ttotal
• Calculate whit for each layer in a stream over a
  popularity window

82
Example
83
Evaluations

• Two Criteria
• Completeness
• Continuity

84
vs Request Number
85
Different Popularity
86
Summary

• Proxy Caching for video
• Techniques for segmentation, replacement etc.
• Interactions with other aspect of streaming