Global Internet Content Delivery

1
Global Internet Content Delivery
Bruce Maggs

• Akamai Technologies
• and
• Carnegie Mellon University

2
The WebSimple on the Outside
ContentProviders
EndUsers
Internet
3
But Problematic on the Inside
Peering Points
ContentProviders
End Users
Network Providers
UUNet
NAP
Qwest
NAP
AOL
4
The Akamai Solution
ContentProviders
EndUsers
Akamai Servers at Network Edge
NAP
NAP
5
Network Deployment
13000Servers
1000Networks
65Countries
6
Content Delivery Using Akamai

• Welcome to xyz.com!
• Welcome to our Web site!
• Click here to enter

http//www.xyz.com/logos/logo.gif
http//www.xyz.com/jpgs/navbar1.jpg
7
Typical Page Content
8
Akamai DNS Resolution
select cluster
End User
select servers within cluster
9
Over 1300 Web Sites are Now Akamaized
10
Differences for Live Streaming
Regions
11
Page Assembly
Container Page TTL5d
Site owners create container pages that can be
populated with varying content
XYZ news, content, promotions, etc. TTL5d
TTL8h
TTL15m
Breaking headlines TTL2h
12
Syntax Example
Akamai
Weather Forecast dc includedcwthr.htm/ boston includebostonwthr.htm/ if geo nyc include nycwthr.htm/ else include uswthr.htm/

• Similar to SSI so its easily understood but
  also breaks performance bottlenecks when
  distributed across 12,000 servers
• Used as an API to third-party applications on
  Akamais network

13
Connecting from the Edge to the Source
End User
Source Server
14
Connecting from the Edge to the Source
End User
Source Server
15
(No Transcript)
16
Akamai Cluster

• Servers pool resources
• RAM
• Disk
• Throughput

17
View of Clusters
buddy
suspended
hardware failure
odd man out
suspended datacenter
18
Real-Time Reporting and Analysis

• Traffic Analyzer
• Real-time viewing of customer traffic
• Reports geographic distribution of traffic

19
Historical Reporting

• Traffic Reporter
• For viewing of historical logs
• Customized data-mining of customer traffic

20
9/11 Summary

• Akamais network had capacity for all content
  providers requesting service
• Total bits served on September 11 was
  approximately 3.5 times normal
• Traffic was higher on September 12
• (But not as high as January 7, 2002)

21
Agent Performance Measurements
22
Portal P FreeFlow traffic
23
News Site A FreeFlow Traffic
24
News Site A EdgeSuite Traffic
25
News Site B FreeFlow Traffic
26
News Site B FreeFlow Streaming
27
Sports Site S FreeFlow traffic
28
Steve Jobs Keynote
29
Network Utilization
No affect on Network Egress (TX backbone example)
09/10/01
09/11/01
BB1 interface to Sprint
09/10/01
09/11/01
BB1 interface to Digex
30
The Reason Why .
Akamai Server Interfaces 1. Original request
sent to website 2. Redirect to akamai server
within POP 3. Minimizes egress transit and
improves overall performance AKAMAIZED
SITES C-SPAN go Network CBS ChicagoSunTimes CNN
Interactive WashingtonPost CNET Reuters MSNBC
Yahoo NBC Internet
09/11/01
09/10/01
Pleasanton
09/10/01
09/11/01
Houston
09/11/01
09/10/01
Dallas
31
Engineering Methodology

• C programming language (gcc).
• Reliance on open-source code.
• Automated unit and system builds and tests.
• Staged rollout to production.
• Independent release management.
• Burn-in on invisible system.

32
Research Questions (Hard)

• How do you know that new software for a
  distributed system of tens of thousands of
  servers will work when it is deployed?
• How do you even know if a system of tens of
  thousands of servers is behaving properly?
• How do you know when such a system is under
  attack?