Content Distribution Network – going forward...

Rasiah Senthurchelvan

What is Content Distribution Network?

A content distribution network (CDN) is a system of distributed servers (network) that deliver webpages and other Web content to a user based on the geographic locations of the user.

Issues Encountered at the Exponential Growth of Internet:

1) Slow response, 40% of customers do not wait more than 3 seconds for a page to be loaded

2) Difficulty encountered in streaming video/audio, Internet television (IPTV): a) High Bandwidth b) Latency c) Jitter

3) Spikes in Internet traffic from flash crowds

4) Single server serving numerous client requests

a) load

b) look up requested info and deliver it back

c) queue

d) traffic e) latency to deliver content

f) the site

Servers nearest to the website visitor respond to the request.

• CDN copies the pages of a website to a network of servers that are dispersed at geographically different locations

The contents of the page are cached

• Upon user request, CDN redirects from the originating server to the proximate server

• Previously not cached content is cached upon request for the same is made and delivered

How CDN works

Fig. 1: A CDN Infrastructure (ref. 3)

How is surrogate server selection performed?By calling,

1) Nearest function A server with a lower round-trip time is considered nearer than one with a higher round-trip time A server with low packet loss to the client is nearer than one with high packet loss

2) Available function of load and network bandwidth A server carrying too much load or a data center serving near its bandwidth capacity is unavailable to serve more clients

Types of CDNs DNS based Request Redirection (RR) Service-oriented Router (SoR) based RR NetServ node Architecture

DNS Request

Making a DNS request is a lot like looking up a phone

number in a phone book: the browser gives the domain

name and expects to receive an IP address back.

18.15.105.3

End user

1.Req: www.example.com

6.Res: IP addr 18.15.105.3

Authoritative Name Serverns.example.com

18.15.105.2

4.Req: www.example.com5.Res: 18.15.105.3

Root Name Server

2.Req: www.example.com3.Res: ns.example.com (18.15.105.2)

Client Name server

6.Res: IP addr 18.15.105.3

Internet

7. Connect to the server(request for redirector)

www.example.com servers

Figure 2:DNS operation: stepsfor a client to resolvethe address.

Authoritative Name Server

Local Name ServerEnd User

Data Center 2

Data Center 1

Edge ServersRequest for

IP address

request

CNAME

List of IP addressOf Web servers

One of IPaddresses

Request w/CNAME1

23

4

56

Figure 3: DNS based CDN

What do we know about Domain Name System?

Name resolution phase of Web access uses Domain Name System Redirecting clients to the nearest server is to perform the server selection function during name resolution phase

So what's the big appeal of this approach?

The DNS provides a service whose primary function is to map domain names such as www.example.com to the IP address(es) of corresponding machine(s)

How could DNS be used for Request Redirection?

its simplicity – it requires no change to existing protocols its generality – it works across any IP-based application regardless of the transport-layer protocol being used.

The DNS-based RR limitations while selecting and redirecting client connections:

DNS has to travel through many tiers of name servers to identify the nearest content server for a particular client Users are required to find the nearest surrogate server's IP address before initiating a connection with the content server As a result, time taken to initiate a connection is comparatively high Due to the low TTL values, clients are required to query local name servers in short time intervals in order to adapt to the network change Multi tiers of name server resolving create lags in the middle of data transmission

Figure 4: Cache Hit Rate at given average query frequency and TTL

Cache Hit Rate vs. Average Query Frequency (ref. 5)

SoR

Server AServer B

Server C

Server Dclient

CDN CDN

CDN

CDN

Service-oriented Router based Request Redirection

The client sends the data stream to the SoR, a specially designed router which is placed at the edge of network

SoR selects the proper surrogate, A,B, C or D, based on the content of the packet and the status of the server

Redirects the packet stream to the selected server.

In addition, an SoR redirects packets by performing server load balancing .

Figure 5:SoR based RR overview

Packet Queue

Packet Headers and Content Surrogate servers

information for the given content

Advertisement details

Advertisement details

Query Surrogate servers for a given content

Packet Headers and Content

Neighbor information

Data of the packet

Forwarding hop

SoR_Forwarder()

SoR_Receiver()

SoR_Agent_Receiver()

SoR_Agent_Forwarder()

Redirection_Decision_maker()

Incomingpackets

Advertisements

Outgoingpackets

Advertisements

Multi-Map structures

Figure 6:Service-orientedRouter Infrastructure

Neighbor Information Table

Routing InformationTable

End user

NetServRouter

NetServRouter

RegularRouter Content

provider(1) User requests http://youtube.com/getvideo?id=foo

(2) You Tube sends video file

(3) You Tube sends on-path signal to deploy MicroCDN module

(4) NetServ routers download the module

(6) Another user requests http://youtube.com/getvideo?id=foo

(7) YouTube redirects user to nearest NetServ node running MicroCDN

(8) User requests http://netserv1.verizon.com/youtube/foo.flv

(5) NetServ routers notify that the module is active

(9) NetServ router relays the video content, while fetching the file and caching it

Figure 7: On-demand content caching using MicroCDN

RegularRouter

Reference:

1) Muhammad Jaseemuddin, Arun Nanthakumaran, and Alberto Leon-Garcia, “TE-Friendly Content Delivery Request Routing in a CDN”, IEEE ICC 2006 proceedings, pg. 1 – 8

2) Sabato Manfredi , Francesco Oliviero, Simon Pietro Romano, “Distributed Management for Load Balancing in Content Delivery Networks”, GLOBECOM Workshops (GC Wkshps), IEEE, 2010, pg. 579 - 583

3) Yun Bai, Bo Jia, Jixiang Zhang, Qiangguo Pu, “An Efficient Load Balancing Technology in CDN”, 2009 Sixth International Conference on Fuzzy Systems and Knowledge Discovery, pg. 510 - 514

4) A. Shaikh, R. Tewari, M. Agrawal, “On the Effectiveness of DNS-based Server Selection”, Infocomm 2001

5) Kazunori Fujiwara, Akira Sato, Kenichi Yoshida, “DNS traffic analysis – Issues of IPv6 and CDN”, 2012 IEEE/IPSJ 12th International Symposium on Applications and the Internet, pg. 129 – 137

Ref...

6) Janaka Wijekoon, Erwin Harahap, Hiroaki Nishi, “SoR based Request Routing for Future CDN”, Nishi Laboratory, Graduate School of Science and Technology, Keio University, Japan, pg. 1 – 5

7) X. He, S. Dawkins, Y. Zhang, “Routing request redirection for CDN interconnection draft-he-cdni-routing-request-redirection-“, 2012.

8) Erwin Harahap, Janaka Wijekoon, Rajitha Tennekoon, Fumito Yamaguchi ̈, Shinichi Ishida, Hiroaki Nishi, “Distributed Algorithm for Router-based Management of Replica Server in Next-CDN Infrastructure”, 2013 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery, pg. 266 – 272

9) S. Srinivasan, J. W. Lee, E. Liu, M. Kester, H. Schulzrinne, V. Hilt, S. Seetharaman, and A. Khan, “NetServ: Dynamically Deploying In-network Services,” in ACM ReArch ’09 (CoNEXT workshop), 2009. Pg. 1 – 5.

10) J. W. Lee, R. Francescangeli, J. Janak, S. Srinivasan, S. A. Baset, H. Schulzrinne, Z. Despotovic, and W. Kellerer, “Netserv: Active networking 2.0,” in IEEE FutureNet