CoBlitz: A Scalable Large-file Transfer Service (COS 461) KyoungSoo Park Princeton University.

CoBlitz: A Scalable Large-file Transfer Service

(COS 461)

KyoungSoo ParkPrinceton University

KyoungSoo Park 2

Large-file Distribution• Increasing demand for large files• Movies or software release

• On-line movie/ downloads• Linux distribution

• Files are 100MB ~ tens of GB• One-to-many downloads

How to serve large files to many clients? • Content Distribution Network(CDN)?• Peer-to-peer system?

KyoungSoo Park 3

What CDNs Are Optimized For

Most Web files are small (1KB ~ 100KB)

KyoungSoo Park 4

Why Not Web CDNs?

• Whole file caching in participating proxy• Optimized for 10KB objects• 2GB = 200,000 x 10KB

• Memory pressure• Working sets do not fit in memory• Disk access is 1000 times slower

• Waste of resources• More servers needed• Provisioning is a must

KyoungSoo Park 5

Peer-to-Peer?

• BitTorrent takes up ~30% Internet BW

1. Download a “torrent” file

2. Contact the tracker

3. Enter the “swarm” network

4. Chunk exchange policy

- Rarest chunk first or random

- Tit-for-tat: incentive to upload

- Optimistic unchoking

5. Validate the checksums

torrenttracker

peers

updown

Benefit: extremely good use of resources!

KyoungSoo Park 6

Peer-to-Peer?

• Custom software• Deployment is a must• Configurations needed

• Companies may want managed service• Handles flash crowds• Handles long-lived objects

• Performance problem• Hard to guarantee the service quality• Others are discussed later

KyoungSoo Park 7

What We’d Like Is

Large-file service withNo custom clientNo custom serverNo prepositioningNo rehostingNo manual provisoning

KyoungSoo Park 8

CoBlitz: Scalable Large-file CDN

• Reducing the problem to small-file CDN• Split large-files into chunks• Distribute chunks at proxies• Aggregate memory/cache • HTTP needs no deployment

• Benefits• Faster than BitTorrent by 55-86% (~500%) • One copy from origin serves 43-55 nodes• Incremental build on existing CDNs

KyoungSoo Park 9

How It Works

Agent CDNClient

Only reverse proxy(CDN) caches the chunks!

CDN

CDNCDN

CDN ClientAgent

CDN

chunk1

chun

k 1

chunk 2

chunk 3

chunk 2

chunk 5

chunk 5

chunk 1

chunk 1

chunk 4 chunk 5 chunk 5

chun

k 4

chunk1 chunk2

chunk 3 chunk3

chunk5 chunk4

CDN = Redirector + Reverse ProxyDNS

coblitz.codeen.org

OriginServer

HTTP RANGE QUERY

KyoungSoo Park 10

Smart Agent

• Preserves HTTP semantics• Parallel chunk requests

Client

sliding window of “chunks”

donedone

done

HTTP

CDN

CDN

CDN

CDNno action

CDN

no actionno action

waitingwaitingwaiting

done

waitingdone

waitingwaiting

Agent

KyoungSoo Park 11

Chunk Indexing: Consistent Hashing

Static hashing f(x) = some_f(x) % n

But n is dynamic for servers - node can go down - new node can joinCDN node (proxy)

Problem: How to find the node responsible for a specific chunk?

Xk : Chunk request

X1

Consistent Hashing F(x) = some_F(x) % N (N is a large but fixed number)

Find a live node k, where|F(k) – F(URL) | is minimum

… N-1 0 …

X2

X3

KyoungSoo Park 12

Operation & Challenges

• Provides public service over 2.5 years• http://coblitz.codeen.org:3125/URL

• Challenges• Scalability & robustness• Peering set difference• Load to the origin server

KyoungSoo Park 13

Unilateral Peering

• Independent proximity-aware peering• Pick “n” close nodes around me• Cf. BitTorrent picks “n” nodes randomly

• Motivation• Partial network connectivity

• Internet2, CANARIE nodes• Routing disruption

• Isolated nodes

• Benefits• No synchronized maintenance problem• Improve both scalability & robustness

KyoungSoo Park 14

Peering Set Difference

• No perfect clustering by design• Assumption

• Close nodes shares common peers

Both can reach Only can reach Only can reach

KyoungSoo Park 15

Peering Set Difference

• Highly variable App-level RTTs• 10 x times variance than ICMP

• High rate of change in peer set

• Close nodes share less than 50%• Low cache hit• Low memory utility• Excessive load to the origin

KyoungSoo Park 16

Peering Set Difference

• How to fix?• Avg RTT min RTT• Increase # of samples• Increase # of peers• Hysteresis

• Close nodes share more than 90%

KyoungSoo Park 17

Reducing Origin Load

• Still have peering set difference• Critical in traffic to origin

• Proximity-based routing• Converge exponentially fast• 3-15% do one more hop• Implicit overlay tree

• Result• Origin load reduction by 5x

Origin server

Rerun hashing

KyoungSoo Park 18

Scale Experiments

• Use all live PlanetLab nodes as clients• 380~400 live nodes at any time• Simultaneous fetch of 50MB file

• Test scenarios• Direct• BitTorrent Total/Core• CoBlitz uncached/cached/staggered• Out-of-order numbers in paper

KyoungSoo Park 19

Throughput Distribution

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 2000 4000 6000 8000 10000

Throughput(Kbps)

Fra

cti

on

of

Nod

es <

= X

(C

DF)

Direct

BT - total

BT - core

In - order uncached

In - order staggered

In - order cached

55-86%Out-of-order staggered

BT-Core

KyoungSoo Park 20

Downloading Times

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 200 400 600 800 1000 1200 1400 1600 1800 2000

Download Time (sec)

Fra

cti

on

of

Nod

es <

= X In-order cached

In-order staggered

In-order uncached

BT-core

BT-total

Direct

95% percentile: 1000+ secs faster

KyoungSoo Park 21

Why Is BitTorrent Slow?

• In the experiments• No locality – randomly choose peers• Chunk indexing – extra communication

• Trackerless BitTorrent – Kademlia DHT

• In practice• Upload capacity of typical peers is low

• 10 to a few 100 Kbps for cable/DSL users• Tit for tat may not be fair

• A few high-capacity uploaders help the most

• BitTyrant[NSDI’07]

KyoungSoo Park 22

Synchronized Workload Congestion

Origin Server

KyoungSoo Park 23

Addressing Congestion

• Proximity-based multi-hop routing• Overlay tree for each chunk

• Dynamic chunk-window resizing• Increase by 1/log(x), (where x is win

size) if chunk finishes < average• Decrease by 1 if retry kills the first

chunk

KyoungSoo Park 24

Number of Failures

4.3

5.7

2.1

0

1

2

3

4

5

6

Direct BitTorrent CoBlitz

Fai

lure

Per

cent

age(

%)

KyoungSoo Park 25

Performance After Flash Crowds

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 5000 10000 15000 20000 25000 30000 35000

Throughput(Kbps)

Fra

cti

on o

f N

odes >

X

BitTorrent

In- order CoBlitz

BitTorrent: 20% > 5Mbps

CoBlitz:70+% > 5Mbps

KyoungSoo Park 26

Data Reuse

7 fetches for 400 nodes, 98% cache hit

7.7

35

55

0

10

20

30

40

50

60

Shark BitTorrent CoBlitz

Uti

lity

(# o

f n

odes

ser

ved

/ co

py)

KyoungSoo Park 27

Real-world Usage

• 1-2 Terabytes/day• Fedora Core official mirror

• US-East/West, England, Germany, Korea, Japan

• CiteSeer repository (50,000+ links)• University Channel (podcast/video)• Public lecture distribution by PU OIT• Popular game patch distribution• PlanetLab researchers

• Stork(U of Arizona) + ~10 others

KyoungSoo Park 28

Fedora Core 6 Release

• October 24th, 2006• Peak Throughput

1.44Gbps Release point 10am

1 G

Origin Server30-40Mbps

KyoungSoo Park 29

On Fedora Core Mirror List

• Many people complained about I/O• Performing peak 500Mbps out of

2Gbps• 2 Sun x4200 w/Dual Operons, 2G mem• 2.5 TB Sata-based SAN• All ISOs in disk cache or in-memoy FS

• CoBlitz uses 100MB mem per node• Many PL node disks are IDEs• Most nodes are BW capped at 10Mpbs

KyoungSoo Park 30

Conclusion

• Scalable large-file transfer service• Evolution under real traffic

• Up and running 24/7 for over 2.5 years• Unilateral peering, multi-hop routing,

window size adjustment• Better performance than P2P

• Better throughput, download time• Far less origin traffic

KyoungSoo Park 31

Thank you!

More information:http://codeen.cs.princeton.edu/

coblitz/

How to use:http://coblitz.codeen.org:3125/URL**Some content restrictions apply See Web site for details Contact me if you want full access!