Caching the Uncacheable [Long Version]

Caching The Uncacheable

Leveraging Your CDN To Cache Dynamic Content

Hooman Behesh+, VP Technology

Dynamic Content Is Really Interes<ng!

What Is Dynamic Content?

•  Stuff that’s not sta+c! •  With web traffic, generally the base HTML –  Big deal because it’s blocking –  And some+mes a large object à longer download

Blocking

What Is Dynamic Content? •  Stuff that’s not sta+c! •  With web traffic, generally the base HTML –  Big deal because it’s blocking –  And some+mes a large object à longer download

•  Could be other things too –  AJAX calls –  API calls

•  More…

Classically, with dynamic content…

Caching

Dynamic Content Caching Problems

•  Serving stale pages – Lack of good invalida+on framework

Caching vs.

Invalida<on

We tried…

•  Real-‐+me visibility – Real-‐+me analy+cs/stats – Real-‐+me logging/tracking

Caching

CDNs and Dynamic Content

•  Generally, handling dynamic content has been a maRer of transport – Op+mize from-‐origin delivery – “DSA” (Dynamic Site Accelera+on) – Middle mile op+miza+ons – TCP tweaks

Dynamic Content, Tradi<onally

CDN Node

Client

Origin

Some TCP Tweaks

CDN Node CDN Node

Client

Origin

Lots of TCP Tweaks

•  We some+mes do micro caching of HTML –  Short TTL for HTML content via Cache-‐Control – Not full proof

•  Ex: news stories faux-‐pas!

•  ESI (Edge Side Includes) –  Par+al caching – Hard and onerous

Actually…

•  Dynamic content is more cacheable than we think

•  Sta+c for shorter periods of +me

•  Unpredictable invalida+on –  Standard HTTP caching rules aren’t good enough

A Lot BeLer!

CDN Node CDN Node

Client

Origin

No CDN

CDN (sta<c)

Blocking CDN

(sta<c)

CDN (sta<c)

CDN (sta<c

+ dynamic)

So Many Benefits!

•  Performance –  Faster +me to first byte –  Faster start render – Happy users!

•  Offload –  Less work for our servers –  Less bandwidth at origin

What would make it beLer?

Programma<c Invalida<on

•  Invalida+on API

•  Granular –  Specific URL – URL groups – All –  Etc…

Purge •  As a page gets published/updated, a purge command also gets published

•  Must be Instant!

•  Big problem with classic CDNs –  mul+-‐minute purges! –  Ex: News story, part 2

Instant?

•  IS NOT: –  12 minutes! –  quick acknowledgment!

•  IS: –  <1sec –  predictable and determinis+c behavior –  (has to also include propaga+on)

Power of the Purge!

•  Purge dependencies – Surrogate Keys – Using tags/labels to purge en+re chunks of content at once

Not totally obsolete!

Example: CMS + Purge

WordPress: Before

CDN Node

WordPress: Before

CDN Node

WordPress: Before

CDN Node

WordPress: Before

CDN Node

WordPress: Before

CDN Node

WordPress: AVer

CDN Node

WordPress: AVer

CDN Node

HTTP/1.1 200 OKContent-Type: text/htmlContent-Length: 55,666Cache-Control: Totally Long Time!

WordPress: AVer

CDN Node

WordPress: AVer

CDN Node

WordPress: AVer

CDN Node

WordPress: AVer

CDN Node PURGE

WordPress: AVer

CDN Node PURGE

WordPress: AVer

CDN Node

(Has to be instantaneous!)

WordPress: AVer

CDN Node

WordPress: AVer

CDN Node

Example: customer1.js

Before

CDN Node

Origin http://www.3rdparty.com/customer1.js

(Referer: www.customer1.com)

Before

CDN Node

Before

CDN Node

Origin

HTTP/1.1 200 OKCache-Control: max-age=60Last-Modified: Wed, 24 Sep 2014 19:51:30 GMTContent-Type: application/javascriptDate: Thu, 25 Sep 2014 12:22:20 GMTServer: ApacheContent-Length: 7835

http://www.3rdparty.com/customer1.js

Before

CDN Node

Origin

Before

CDN Node

Origin

Before

CDN Node

(Ager 1 min)

Before

CDN Node

Origin http://www.3rdparty.com/customer1.js VALIDATION

If-Modified-Since: Wed, 24 Sep 2014 19:51:30 GMT

(Ager 1 min)

Before

CDN Node

304 Not Modified

(Ager 1 min)

Before

CDN Node

304 Not Modified

(Ager 1 min)

CDN Node

Origin

HTTP/1.1 200 OKCache-Control: max-age=60, s-maxage=2592000Last-Modified: Wed, 24 Sep 2014 19:51:30 GMTContent-Type: application/javascriptDate: Thu, 25 Sep 2014 12:22:20 GMTServer: ApacheContent-Length: 7835

CDN Node

Origin

CDN Node

Origin

CDN Node

(Ager 1 min)

CDN Node

(Ager 1 min)

This happens many many <mes! (many many happy visitors!)

CDN Node

Origin

Customer1 changes config

(Ager 1 min)

CDN Node

Origin PURGE customer1.js

(Ager 1 min)

CDN Node

Origin PURGE customer1.js

(Ager 1 min)

CDN Node

(Ager 1 min)

CDN Node

(Ager 1 min)

CDN Node

Origin

(Ager 1 min)

More than just Invalida<on…

The Influence of Clouds

•  The CDN is an extension of the app •  No longer a black box •  Real-‐+me integra+on with the app •  Infrastructure as code – Your content => You need control

Control

•  Programmability – Configura+on API –  Invalida+on API –  Instantaneous and real +me

Caching Control

•  Granular caching – Dynamic caching – flexible cache keys – etc

•  Ex: Geo-‐based caching

Control at the Edge

•  Moving app logic to the edge •  VCL – Varnish Configura+on Language – Script-‐like configura+on for func+onality at the edge

Visibility

•  Real +me analy+cs – Network stats – HTTP stats (status codes , etc) –  Caching stats (hits, misses, etc) –  Stats API

•  Logging –  Real +me logs –  Streaming to various log des+na+ons

Example: Beacon Termina<on at the Edge

Before

CDN Node

Origin

Log Analysis

http://collector.site.com/beacon.img?a=1&b=2&c=3

Before

CDN Node

Origin

Log Analysis

Before

CDN Node

Origin

Log Analysis

HTTP/1.1 200 OKPragma: no-cacheExpires: Wed, 19 Apr 2000 11:43:00 GMTCache-Control: no-cache, no-storeLast-Modified: Wed, 21 Jan 2004 19:51:30 GMTContent-Type: image/gifDate: Fri, 20 Jun 2014 12:22:20 GMTServer: ApacheContent-Length: 35

Before

CDN Node

Origin

Log Analysis

CDN Node

Origin http://collector.site.com/beacon.img?a=1&b=2&c=3

CDN Node

Origin

CDN Node

Origin

HTTP/1.1 204 No ContentDate: Sat, 21 Jun 2014 23:21:12 GMTServer: Awesome ServerContent-Length: 0

CDN Node

Origin

Syslog / S3 / FTP/etc

Example: Edge-‐generated Content

JSON Data Center ID

CDN Node

Origin http://www.site.com/which_datacenter.js

JSON Data Center ID

CDN Node

Origin

{ ‘datacenter’ : ‘SJC’ }

http://www.site.com/which_datacenter.js

VCL Snippet

JSON Geo IP

CDN Node

Origin

{ “city” : “New York”, “state”: “New York”, “country”: “United States”, “ip”: “173.18.14.237”}

http://www.site.com/geo_ip.js

More Examples

•  Caching with tracking cookies: –  hRp://www.fastly.com/blog/how-‐to-‐cache-‐with-‐tracking-‐cookies

•  API Caching: –  hRp://www.fastly.com/blog/api-‐caching-‐part-‐iii (part 3, with links to previous two parts)

•  Log Streaming: –  hRp://www.fastly.com/blog/+ps-‐for-‐streaming-‐logs

Let’s Sum Up!

Summary

•  Dynamic content can be cached – We need instant purging – We need real-‐+me logs and stats

•  Real-‐+me integra+on of our CDN with our app is cool! –  Extensive/granular API to control the CDN –  Control and visibility at the edge lets us be really crea+ve

•  Never use “Totally Long Time!” in a Cache-Control header!

Thank you!

hooman@fastly.com

Caching the Uncacheable [Long Version]

Technology

Storage-Aware Caching: Revisiting Caching for Heterogeneous....

long, web-only version

Presentation destination changemakers long version

Brown Bear Long Version

(re)Mix010gy (long version)

GeometricMonitoringofHeterogeneousStreams(Long Version ...

Presentation Long Version 3.2.09

Going Indie - Long version

Koine Vision Long Version

Long version natavan aliyeva

Ehcache User Guide · Ehcache is now used for Hibernate...

Caching in ASP - gangainstitute.com€¦ · Caching in...

HFDP Certification Catalog Long Version

Caching and caching dependencies explained in Kentico CMS

Overview Introduction to ASP.NET caching Output caching .....

Our Center Long Version