Transcript 3) CCN - MMLab

i-3 content-centric networking
Taekyoung Kwon (TK)
[email protected]
Some slides are from Van Jacobson@PARC
1
Why Content Networking (CN)?
2
Internet
• Original motivation for Internet is to share
computing resources
– Remote login, file transfer
• The Internet communication model is host-tohost conversations
– An IP address indicates a host
• It has been successful for 40 years
– TCP/IP has been so flexible for diverse host based
applications
3
Why content networking (CN)?
• Internet traffic is already content-oriented
– CDN, data center, web cache, redundancy elimination…
– multimedia, web, P2P, IPTV…
• Users/applications care “what to receive”
– They don’t care “from where or from whom”
– So, host-based comm. model is outdated
• Storage cost is getting cheaper sharply
– Compared to networking cost
• Other advantages of CN coming up later
4
Internet traffic breakdown
5
IP networking
• Host-centric design
• Lookup-by-name
– Indirection (from name to locator)
• DNS
– Host/link/DNS availability concern
• DNS is extendible and highly available
– Distributed design
– Thanks to caching
• Locators can be aggregated
– Network prefix
• Currently almost 450k
– Routing scalability is better than CN
6
Content networking (CN)
• Route-by-name
– No indirection, better availability
– Content name (or ID) is a routing entry
– Huge scalability concern
• In-network caching
• Global-scale pure CN may not be feasible
– At least billions of contents
– Some aggregation may be possible
• E.g. hierarchical names like URLs
• Other merits such as authentication
7
Storage cost
• In-network storage
8
IP networking vs. CN
• Network prefix
• Content name
Destination
Next Hop
Content Name
Next Hop
192.168.0.0/16
Router C
/a.com/b.jpg
Router C
/a.com/b.jpg
9
Content name or ID
• Content names (Cnames)
– May replace the IP addresses
– Content identifiers (CIDs)
• Cname/CID design choices
– Hierarchical vs. flat
– Semantics vs. semantic-free
• Persistency
• Location independence
– Variable length vs. fixed length
• examples
–
–
–
–
cnn.com/sports/news.avi
sonypictures.com/spiderman3.html
yahoo.co.kr/image/logo.jpg
0xF034BC….024A,
• E.g. hash of content data, name, public key
– Or hybrid
10
Advantages of CN
• Better delivery efficiency
– Multicast, mobility, QoS/QoE,…
• Caching at CN-capable routers
– Shortest path to the (potentially cached) content
• Inter-ISP traffic reduction
• Web host provisioning
– Server-less computing
• Policing
– track the history of content requests
• Content authenticity
– signature
– Phishing and pharming are not possible
11
Where to put Cname?
• In TCP/IP
– Application layer header
• E.g. HTTP, SIP
• Deep packet inspection
– IP option header
• New L3 header
– A clean slate approach
12
Content Centric Networking (CCN)
Named Data Networking (NDN)
13
CCN
• “Networking Named Content,” ACM
CoNEXT 2009.
• Van Jacobson
– Palo Alto Research Center (PARC)
• NDN project at NSF
14
CCN Philosophy
• Solve the cognitive mismatch
– User/app wants “what”
– Network wants “who”
– Mapping between two models requires a lot of
convention and configuration (middleware,
wetware)
• Users specify the objective
• No distinction between bits in a memory
and in a wire
• Data security and integrity are the
architectural foundation
15
CCN basics
• Content name
– Hierarchical, variable-length, semantics
• No IP address
• Consumers send Interest Packets
• Content holders send back Data Packets
Source: Van Jacobson@PARC
16
CCN hourglass model
Source: Van Jacobson@PARC
17
A user wants some content
Source: Van Jacobson@PARC
18
Content is downloaded
Content is
cached!
In-network
caching
Source: Van Jacobson@PARC
19
Another user requests the same
content
Source: Van Jacobson@PARC
20
Name tree
• Components
• Tree traversal to discover
data
• Default traversal is
LeftmostChild
• Relations: prev, next
• /parc.com/videos/Widget
A.mpg RightmostChild
Source: Van Jacobson@PARC
21
CCN forwarding
Source: Van Jacobson@PARC
22
CCN: Strengths and Weaknesses
Pros
• Better availability
• Better delivery
• Inter-ISP traffic
• Accountability, Policing
• Integrity
• Authentication
• Serverless computing
• Multicast
• Mobility
Cons
• Huge scalability concern
• More in-network
processing
• Potentially more
signaling for routing
23
Multiple copies
CCN router
IP router
* No loop in CCN!
Source: Van Jacobson@PARC
24
CCN security
• In CCN, the content itself (not its
container) is trusted
– In TCP/IP, endpoints are authenticated
• Anypoint can retrieve the content
from anywhere and validate it
– Content is publicly authenticatable
• All content is digitally signed
– Binding btw. name and the content itself
• Still PKI is needed
25
Signature in CCN
• Binding btw content name and payload is
authenticated
Name ||
Payload
digital
Signature
(encrypt)
PKI
signature
Name ||
Payload ||
Signature
Data packet
digital
Signature
(decrypt)
Verify!
Name ||
Payload
anypoint
publisher
Publisher’s private key
Publisher’s public key
26
Interest packet
Source: Udugama at Univ. Bremen
27
Data packet
Source: Udugama at Univ. Bremen
28
CCN vs. Related Technologies
(CDN, P2P, ICN)
29
Why CDN? For CPs!
• How do I get my content to my
customer quickly, reliably, and accurately?
• How can I support 20 million hits per
day?
– Flash crowd
• Can I offload any server traffic?
* CP: content provider
CDN: content delivery network
30
CDNs make content more
available
• Push the content to the edge
– Multiple places
• Load balance mirrored content
• Creative DNS solutions
31
First Point – DNS
• Selects from among several
mirror sites operated by
content provider
32
Akamai DNS Resolution
4
xyz.com
510.10.123.5
xyz.com’s
nameserver
akamai.net
8
a212.g.akamai.net
7
6
.com .net
Root
(Verisign)
9
15.15.125.6
ak.xyz.com
10 g.akamai.net
20.20.123.55
11
select cluster
Akamai High-Level DNS Servers
12 a212.g.akamai.net
Local Name
Server
End User
16
Browser’s
Cache
14
3
1
30.30.123.5
13
Akamai Low-Level DNS Servers
select servers within cluster
2
15
OS
33
With/without Akamai
Akamai operates over 73,000 servers in 70 countries in about 1,000 autonomous
systems, which on any given day may handle upwards of 20% of traffic in 2010
34
Akamai vs. LimeLight (As of 2008)
Source: GridsLab at Univ. of Melbourne
35
Problems with legacy CDNs
• Current CDN models
good for offloading
content distribution
• Cache mostly located
at exchange points
– Colocation
• No benefit to access
network operators
– No traffic reduction
– No revenue sharing
Source: Dirk Kutcher@NEC Lab
36
telco CDN (or operator CDN)
• ISPs are not happy with “off-net” CDNs
– Network control issues
– Limited monetization chances
Source: Alcatel-Lucent
37
telco CDN: strengths
• Aka “On-net” CDN
• Optimize QoS/QoE
– E.g. Deep caching
• Low cost
– Same CDN infra for its own content and CP’s content
– Cache OTT content (i.e. reduce traffic cost)
• Help CPs
– Customize services depending on content portfolio
• New Biz models
– In the content-to-customer chain
• Managed vs. OTT content
• E.g. Value-added service for CPs
* OTT: over the top
38
Recent changes in CDN: P2P CDNs
• P2P CDNs
– Exploit user machines mostly
– Little cost
– Often copyright issues
39
Google Global Cache
• CP can be a CDN provider as well
40
BitTorrent operations for “popeye.mp4”
www.bittorrent.com
1
Peer
• File
popeye.mp4.torrent
hosted at a (wellknown) webserver
• The .torrent has
address of tracker for
file
• The tracker, which
runs on a webserver
as well, keeps track of
all peers downloading
file
41
BitTorrent operations for “popeye.mp4”
www.bittorrent.com
Peer
2
Tracker
• File
popeye.mp4.torrent
hosted at a (wellknown) webserver
• The .torrent has
address of tracker for
file
• The tracker, which
runs on a webserver
as well, keeps track of
all peers downloading
file
42
BitTorrent operations for “popeye.mp4”
www.bittorrent.com
Peer
3
Swarm
Tracker
• File
popeye.mp4.torrent
hosted at a (wellknown) webserver
• The .torrent has
address of tracker for
file
• The tracker, which
runs on a webserver
as well, keeps track of
all peers downloading
file
43
BitTorrent: Basic Idea
• Chop a file into many pieces
• Replicate DIFFERENT pieces on different
peers as soon as possible
• As soon as a peer has a complete piece,
it can trade it with other peers
– Tit-for-tat
• Hopefully, a peer will be able to
assemble the entire file at the end
44