8/15/2019 Low Latency Data

1/37

Comet:

Low Latency Data For Browsers

Alex Russell

alex@jot.com

Project Lead, The Dojo Toolkit

mailto:alex@dojotoolkit.orgmailto:alex@dojotoolkit.orgmailto:alex@dojotoolkit.org

8/15/2019 Low Latency Data

2/37

Please Ask Questions and InterruptThese slides are online at:

http://alex.dojotoolkit.org

But you didnt come to ETech to read slides

http://alex.dojotoolkit.org/http://alex.dojotoolkit.org/

8/15/2019 Low Latency Data

3/37

Ajax is medriven

8/15/2019 Low Latency Data

4/37

Social appsare also driven

byothers

8/15/2019 Low Latency Data

5/37

Eyes On The UserThe goal is responsiveness

Ajax improves responsiveness for:

Single-user CRUD

Write-only interactions

Read-only apps where large lag is acceptable

The web is inherently multi-userAjax is only half the answer

8/15/2019 Low Latency Data

6/37

To any user,

the serverisother users

8/15/2019 Low Latency Data

7/37

The Multi-User WebSingle interaction updates are not enough

Users in the same space need live updates of:

Their own changes

The changes others make

Updates to context affect available actions

Stale context may mean the wrong decision

8/15/2019 Low Latency Data

8/37

If the web is aconversation...

8/15/2019 Low Latency Data

9/37

...then stalecontext kills

8/15/2019 Low Latency Data

10/37

Latency MattersConversation mediums are defined by latency,interrupt, and bandwidth

Snail-mail

Email

IRC

SMS

IM

Phone

Face-to-face

8/15/2019 Low Latency Data

11/37

Example: Wikis As

ConversationsWikis are conversation enablers

Traditionally medium-to-high latency

Not well suited to high-volume changes

Locking/overwrite issues

Ajax allows more context to go stale

What is changing in the Wiki while I edit?Who wants to break my lock?

Have attachments been added?

Is the text of the a e itself chan in ?

8/15/2019 Low Latency Data

12/37

Conversations Are

Ordered EventsGranular interfaces require granular events

Granular conversations are more immediate

IM vs. Email

Social apps are event busses

Social web apps just batch changes today

No effective way to subscribe to server eventstoday

To fix the context, syndicate the events

Does SOA ring a bell? How about JMS?

8/15/2019 Low Latency Data

13/37

Broadcast Is Synchronization ForShared Nothing

8/15/2019 Low Latency Data

14/37

Comet

8/15/2019 Low Latency Data

15/37

Comet: Server Push DataNew term, old tech

Unburdened by previous definitions and tools

Long-lived HTTP connections instead of polling

Similarities to Ajax:

No plugins

Plain old HTTPAsynchronous

Broad browser support

Pa load can be an thin textual

8/15/2019 Low Latency Data

16/37

Applications Implementing

CometGMail + GTalk

JotLive

Renkoo

Meebo

cgi:irc

KnowNowothers?

8/15/2019 Low Latency Data

17/37

Why Now?

8/15/2019 Low Latency Data

18/37

How Is This

Different FromAjax?

Servers push data in additionto clients requesting it

8/15/2019 Low Latency Data

19/37

Stale Ajax

Context and manipulatedcontent go stale at differentrates T

ime

Server

Ajax

Interaction

8/15/2019 Low Latency Data

20/37

Keeping UpPushing state updates keepsAjax interactions and pagecontext in sync T

ime

Server

Com

etUpdatesState

8/15/2019 Low Latency Data

21/37

How Comet Fights LagAvoids HTTP & TCP/IP set-up and tear-down

Single connection is re-used (in some forms)

Ajax + Polling latency:

Time since last polling request +

TCP and HTTP request setup +

Data transmission time/latencyComet latency:

Data transmission time/latency

Lots of zombie connections! (C10K+ problem)

8/15/2019 Low Latency Data

22/37

Transfer Only Necessary Data,Exactly When Its Most Relevant

8/15/2019 Low Latency Data

23/37

Demo

8/15/2019 Low Latency Data

24/37

Implementation StylesLong-poll

Examples: Meebo, Live Page

Reconnect after every datagram

Server might package multiple datagrams together

Simple to implement w/ XMLHTTP object

Multipart XMLHTTPNo known system does this portably today

Similar to the forever-frame technique

Different delimiters for IE and FF, no Safari

8/15/2019 Low Latency Data

25/37

Contd.The forever frame

frame or iframe

Browser incremental rendering hack

blocks sent to iframe are evaluated aftersome sort of flush token tickles the browser

Highly portable

Allows connections to sub-domains

document.domain

Important in designing workable architectures

8/15/2019 Low Latency Data

26/37

Client Server

Time

connectionopen

data

connection close

connectionopen

connection close

data

connectionopen

The Long Poll

Data transfer ends theconnection

Clients re-connect after every

datagram

8/15/2019 Low Latency Data

27/37

Client Server

Time

connectionopen

data

connection close

connectionopen

data

data

data

data

data

data

Forever Framesand MultipartConnection only closes onerrors or connectionrecycling

Data is encoded in envelopes

8/15/2019 Low Latency Data

28/37

Todays Web Servers Wont Cut It

Copyright Simon James

8/15/2019 Low Latency Data

29/37

Comet Can Reduce LoadBut not on your current web infrastructure

Polling is a latency/load trade-off

Comet is an architectural complexity trade-off (today)

Most of todays web servers use threads or processes

Threads consume fixed resources per request

Do not free them until end of connectionComet does not free connections quickly

Polling frees resources quickly, but makes many times asmany requests

8/15/2019 Low Latency Data

30/37

Client Server

Time

Client Server

Time

Load Profiles

Client Server

Time

Long PollPolling Forever Frame

8/15/2019 Low Latency Data

31/37

Event-Based ToolsOS level

kqueue (FreeBSD)

epoll (Linux)

Network level

POE (Perl)

Twisted (Python)Yaws (Erlang)

event_mpm (Apache 2.2, unstable)

Jetty (Java)

8/15/2019 Low Latency Data

32/37

The Two

ConnectionLimit

8/15/2019 Low Latency Data

33/37

WorkaroundsMultiplex!

Events for multiple components must come over

the same connection

Prevent creation of multiple tunnels

DNS hackery

document.domain + subdomains, wildcard DNS

Flash

XMLSocket + Flash 8 ExternalInterface

8/15/2019 Low Latency Data

34/37

Is Comet Good For My

A ?Do users collaborate on shared data?

Can presence data improve the conversation?

Can your users benefit from fresher data?

Can it not be attempted any other way?

Can my architecture handle it?

How long do users stay in a single page?If lag is acceptable, can polling work/scaleinstead?

8/15/2019 Low Latency Data

35/37

Early Design LessonsWork with interaction designers

Learn from your desktop competition

They had the same design problems

Be consistent

Let users know whythe data is changing

Let users know who changed the dataCommunicate connection failures clearly

Push data updates, not functionality changes

8/15/2019 Low Latency Data

36/37

Evolution, Not Revolution

8/15/2019 Low Latency Data

37/37

Questions?