speaker: stevesouders.com(chief performance guy @ Yahoo! now @ Google)

slide presentation: slide presentation

here’s an unbelievably great tool to help you figure out the impact of various HTML options on page load times!

ibm page detailer – shows page request over time – another great tool for profiling page load times!

most time is spent on frontend

load packet sniffer then the browser

80-90% of end user response time is spent on frontend

14 rules
- make fewer http requests
- use a cdn
- add an expires header

yslow is a tool to analyze performance

very nice guy! looks like keith neds

book – high performance websites

vel08st 20% off before may 5
velocity
june 23-24

new book – even faster websites
- split the initial laod
- load scripts w/o blocking
- don’t scatter scripts
(more)

scripts block parallel downloads!
stevesouders.com/cuzillion
cuz there are a zillion pages to check
open source
tool for quiclly building fast pages

list of functions not used before onload – huge!

Do manually using firebug
load scripts w/o blocking
example – msn.com does parallel scripts!

put script inside an iframe

decision points
- same domains?
- order matters?
- show busy symbol?

Nothing on page rendered until all inline scripts have executued!

Firefox blocks parallel dl when dl stylesheets

ie doesn’t unless ss followed by inline script

created @ a ski resort in Utah in 2001
manifesto for agile soft dev

interactions
working software
customer collaboration
response to change

perfect doc doesn’t add value to end user ????

no plan survives 1st contact with the enemy

principles of agile
- highest priority is to satisfy customer thru eaely & continuous valuable software
- deliver working soft frequently (weeks to months)
- working soft is the primary measure of progress
- welcome changing requirements
- harness change for competitive advantage
- simplicity, the art of maximizing the amt of work not done, is essential
- continuous attention to tech excellence & good design enhances agility – make code extensible as you go
- at regular intervals team discusses how to become more effective & adjusts (code refactoring, process refactoring)
- business people & developers work together daily
- face to face is most effective form of communication
- build projects around motivated people – give them what they need to get job done

user stories – users should be able to login
points assigned by developer
owners look @ points & decide what needs done
do not assign point to bugs & refactoring!

Time boxing (sprints in scrum) – decide how much time to do chunks

communication – must have it. Irc, IM, campfire, calls, you must enable this

daily stand up meetings (15 min max) – early part of day

pair programming

select all/delete – makes code more extensible

build & integrate constantly – automate this process! – use lava lamps!

Testing – you need a complete test suite
stupidity testing – build test for each bug u find
test driven – write test before the code
behavior driven – ?

What about doc @ the end? – test suites are the best

speaker: jonathan schwartz – Sun CEO
Role of blogging in communicating with world & employees (32K)

referred to safe harbor stmt (2nd time I heard of these this week!)

if u r going to lead u must communicate

mysql bought for $1B – why?
- they were making good money
- getting ready to go public
- synergy
- 70K downloaded per day!
So they bought leads! Lots of leads!

50M java runtime downloads/month

the internet as os
utility conputing

network – web – cloud – ???

characteristics of utilities
- transparent pricing
- substitutability

high performance computing (hpc)
500 teraflops
larger than all nsf facilities combined!
business advantage of hpc
running analytics against all the social data being collected
they just bought virtual box – download free
“the worlds largest supercomputer”

when u make computers twice as fast people won’t buy half as many. They will likely buy twice as many.

writing – creating the context in which others think

1/5 energy in data centers is moving air
movable data centers – build inside a shipping contaner!
Power is the most expensive part of data centers
power prices fluxuate
so need ability to easily move dc

speaker: dan lyons – fake steve jobs
tech writer for forbes but he was bored so he did a fake blog on steve jobs

2 twins 2 yrs old – they wake him up every day @ 6am to watch videos

I want this videoed so 12 years from now I can wake my teens up @ 6am to show them the video!

Facebook is like webkinz for adults

speaker: dash – rob cury
dash express – first internet connected gps

crowd sourcing traffic
we know where to build the next starbucks based upon searches in cars!

speaker: google matt cutts @ google 8 years working on spam – web spam

trust & reputation reduce spam

captcha’s – stopping spammers
spammers are human – make them mad! (waste their time)

speaker: matt mulaway of wordpress
started automatic
akismet & wordpress

168M users with 20 employees

monotone – dynamic creation of blog template based upon characterisics in a photo

speaker: Clara Shih from SalesForce.com

created it on nites & weekends
works for appexchange
she use to work @ google
facebook pays employees $600/month bonus to live within 1 mile!
safe harbor ststement before all salesforce presentations – buy @ your own risk!
decade legacies
70 – mainframes
80 – client servers
90 – internet
00 – social networking?

moto – no software
saas – new way

no longer owned by IT
big trend we are seeing

liveops – a virtual call center company

online social networks will enter the enterprise

crm is the first social network
$75B/year industry

relationships & referrals are key to sales

life is all about relationships!

she attended oxford
she is from hong kong
she was bit by the new salesforce security activation! (login on a new computer & you have to have access to email!)

who owns the data I pull in from my facebook contacts?

ringside – allow developers to manage a federated social network by using best of breed

hooversconnect – closed sn of cio’s by invitation only

sf 2 sf – poking could mean sending a proposal to a company

built in php
hosted @ dreamhost
now would build on force.com using apex

• tagging is unstructured, social, organic (they change over time – consider Katrina)
• incredible amounts of data
• 11,000 tags created on a library of congress photo project in 24 hours
• magic – inferred tags (created by a russian mathematician)

what is an ontology & how is it useful on the web?

references: john sowa, adam pease

agenda
- problems with tagging systems
- semantics & logic
- application

folks – people
nomy – management

will folksonomy lead to semantic web?

tagging pattern
- resource
- entity inspects the resource
- applies a tag

collaborative tagging – folksonomy – social bookmarking (del.icio.us)
problems – ambiguity

conceptual domain

tag clouds
folksanomies

possible app (disambiguating)
- user submits tags
- system consults ontology db
- system displays choices (were ambiguity exists)
- user selects from choices

linking folksonomies to ontologies is a worthwhile endeavor

Speakers: mysql, amazon simpledb, microsoft

• amazon guy pushes – let us handle the tech details. Trust us!
• how are u dealing with your address book data? scripts & synchronizing
• amazon – simpledb
• google – mapreduce
• big vs adaptable
• meshlive is all about abstracting away details of synchronizing data
• control vs ease of use/cost?
• Does it matter where my bits are?
• Does it matter who has access to my bits?
• What about my clients bits?
• Performance, cost, integrity, security, trust
• all data in one place or synchronize over multiple places?
• Relational or cloud db?
• Note – many here using iphone!
• Will sql be here tomorrow? It will go the way of assembly language – abstracted down the stack.Data portability

Will the db be abstracted away into “the cloud”? This is what Amazon is betting on. Some don’t like the details of SQL. I think it will come down to a flexibility vs usability issue. Time will tell.

Summary of Thursday’s keynote speakers.

jonathan zittran author of “the future of the internet & how to stop it”

• us 1.0 – we will all get along and if not just move west
• us 2.0 – we need checks & balances
• platforms (facebook, google) need to be able to say to the man in the suit with shades sorry we cannot do what you are asking
• we need more elements of citizenship

michelle baker (CEO of Mozilla)

• now  $55M company
• opening the mobile web
• there is just 1 web
• firefox is the best way to get there
• what can I get to and what can I do with it – not what device am I using
• The use cases should be the same regardless of the device
• apcmag.com/mozilla_ceo_speaks_out_on_future_of_firefox.htm

CTO of yahoo

• Y! OS – making yahoo! More social (OS = Open Social)
• announced release of beta for developers to access search monkey
• letting users choose where they want to share what
• unifying all yahoo profiles
• rewiring yahoo to open up web services for consistent development, deployment, etc
• making yahoo more social
• open social = operating system?

Probably the BEST keynote of the conference and I lost all my notes! I was burning too much battery the day before so Matt suggested that I turn off my wireless and take notes in Word. That sounded like a good idea, except for the Word part, so I started this blog page, then disconnected my wireless. When I was ready to commit all my words to the blogosphere, I turned wireless back on, pressed the publish button and ERROR CONNECTING TO SERVER with the side effect of GOOD BYE DATA! Matt was quick to remind me that’s why he suggested putting my words in Word!

Oh well, it really was a good keynote talk. The one thing I recall is Marc’s latest business venture – nign.com. A place to create social networks. To date there have been 250,000 networks created. They have an interesting pricing model – advertising or you can pay to keep the ads off. Looks like a lot of their sites are alumni associations. I think GOMP will prove to be a great idea for keeping our graduates connected to each other and to BJU!

speaker: Clay Shirky, author of “Here Comes Everybody: The Power of Organizing Without Organizations”

time spent on entire wikipedia project to date: 100M manhours
time spent tv watching = 2000 wikipedia projects per year!
100M manhours of ad watching per weekend!

Cognitive surplus
doing anything is better than doing nothing
people like to consume, produce & share

internet community watches 1T hours of tv per year! if we only get 1% shift to creating & sharing the impact will be 10,000 wikipedias!

Story of 4 year old girl who after her father installed a new HD TV was caught behind the system looking for the mouse! a media system that ships w/o a mouse ships broken!

deep trends of web 2.0
- internet as platform
- harnessing collective intel
- data is the intel inside
- software above the level of a single device
- software as service

opportunities
- enterprise use of web 2.0
- cloud computing
- mobile phones & ubiquitous sensors – data will be collected from sensors (mac has a motion sensor – users are giving that info to a quake network to compute the next big one)

question to ask: how can we make our customers more efficient?

we are starting to treat the Internet as the OS

speaker: gavin bell

cognitive dissonance – people are now comfortable with the fact that a link on one site will take them to another site so get over it!

flow – user is fully engaged with the process

challenge people – give users something a little beyond their experience with reward for accomplishing – people enjoy challenges they can reach

curiosity
pivots
reinforcement
social software – we can have meaningful social relationships with 150 depends on the size of our cortex

continuous partial attention

human memory is associative, not linear
we remember 7 +/- 2

email is one of our primary social interactions online – tons of social info is in our inboxes

verbs are focussed on what users want to do rather than what we want them to do

focus on experiences not features

openID – sound controverial! see if a student wants to research

have you considered filing email bankruptcy?

everyone uses email
have you considered building an email app?

Tripit – name of company of presenter

created from a recurring trend – no one buys all travel products from same place – one trip but lots of sources of info (tix, rental car, hotel, other tix/reservations

forward all travel emails to plans@tripit.com

- parses the data & reformats
- integrates with other related things
- ical & atom support
- iphone, treo, blackberry

trying to interface with travel companies was too complex (technically and politically)

types of email app services
- broadcast (mailing lists – google alerts)
- cmd line interface (i want sandy, tripit, twitter – from web, sms, I’m, facebook, desktop…))
- data transfer – fickr (email to xxx@photos.fickr.com), blogger
- semantic parsing – trackmyshipments, tripit

problems
- deliverability (keep your sending ip addresses clean!) – proper dns config, use spf/dkim, privacy policy, rfc compliance, esp whitelists, bounce handling
- spam (spammers like to go fast so slowing things down in an rfc compliant way can help)
- privacy issues (opt out mechanism, listen to your communitites, be transparent & responsive)
- standards

use spf & dkim!!!!!!

panelists – they know how to make money from the Web 2.0 stuff

consumers are adopting this stuff w/o evening knowing they are (tagging, blogging, posting videos, etc)

mortaza husan – ceo of peanut ads – interview 10,000 xbox users via social networks and ask them questions; work with 70 social networks

erin hunter – comScore – largest provider of digital marketing, page views, key metrics, measures use of videos on web, ad impressions, 2 million users in the network

conrad – podcast – measure audience response, 30K publishers, help businesses better understand their audiences

derrick schmidt -

how is the web 2.0 consumer different from web 1.0 user?

more interactive, dynamic, lauch then measure, “the conversation”, attractive because of the large audiences but not in control, we are all digital now

how do we make it easier for the marketer to reach the consumer – many more touch points

can no longer reach 20M users via a single event (TV Show); people are much more scattered; need to advertise on 100 sites to reach the same users!

participation & distribution

more are willing to participate

content is everywhere

what technologies are you watching?

• mobile
• blogger
• widgets
• (looking for all the touch points)
• video
• twitters
• search – does it mean brand is more or less important? search can have a branding affect!

fragmentation creates challenge and opportunity

we self organizing into common groups

makes measurement more difficult

20 years ago the average person watched 12 TV stations; today it is 13; we can only handle so many touch points!

discovery and delivery will have to be more automated (job security for computer science students!)

recurring theme – advertisers can no longer spend $20M on TV advertising and get the same impact because consumers are fragmented in where they spend their time

click thru vs branding – getting people to think of you when they need you

Avenue A / Razorfish Consumer Behavior Study, July 2007 digitaldesignblog.com

• tag clouds
• access mobile data services
• post to blogs
• subscribe to RSS
• use social network
• read blog
• viewed a video online

Internet’s reach exceed TV from 7am to 8pm (comScore Media Matrix 2007)

beaware that cookie deletion is growing (overstate reach, understate frequency) 30% delete cookies each month

16% of clickers account for 80% of clicks

“all purchasing is local” – is this true? (response to a question about globalization impact on online advertising)

speaker: Blaine from Twitter (left 2 weeks ago)

summary: how to keep people connected with eath other real time

twitter is a good example of this real time information

Jabber – an open source, xml based instant messaging platform (basis for Twitter)

Sept 2000 Jabber.com signed a contract to provide mobile interface

what is the real time web? Social Objects, the things we exchange, writing, photos, audio, video, location, personal data

we want a notification based system rather than a polling system

“are we there yet” vs “let me know when we are there”

goals

• real time
• low cost
• asynchronous
• simple

HTTP asks the question “what happened in the past?” (wrong question)

HTTP is hard to scale for frequent polling

HTTP Ping/Push – also has problems

SMTP – too much spam!

Comet (used by gmail) – requires polling, one connection per user

Jabber – fulfills all the goals

First Steps

• not p2p
• client to server
• looks like email
• client maintains persistent connections
• it is all just xml
• 2 streaming xml documents
• addressing looks just like email
• Jabber Federation
• server to server
• dialback authentication
• explicit whitelist by default
• messages are the primary payload
• can send html and Atom
• streaming conception of the web

over 200 specs in jabber – ignore them unless you need them!

core & IM: RFC 3920, 3921

commands: presence, send/receive messages,

define bot behavior

• what does your bot do?
• conversational
• informational
• recorder

define api behavior

• what does your api look lie
• atom?
• custom xml with namespaces

rosters – let the libraries handle them until you need to scale your app

PubSub – mechanism for publishing & subscring to feeds

like presence subscriptions (tell me when xxx is online) but for data

Social Network Federation – what happens when twitter goes down? same problem we would have if the world was made up of one email server!

the solution – PubSub mechanism; breaking down walled gardens

sounds like they are in the early stages of trying to build a network of adapters to this technology??

use atom over xmpp

Scalability

• scaling is not well documented
• jabber stops working well at 35K contacts due to roster presence behavior
• solution – use components (XEP-0014); allows you to turn off roster

Tools

• Ruby: xmpp4r & xmpp4r-simple
• Java: Smack
• Python: twised-words
• Perl: Net::Jabber
• Javascript: JSJaC
• Jabber Servers
• ejaberd
• openfire
• debugging: Psi
• PubSub: Idavoll

Jabber enabled sites

• livejournal
• twitter
• jaiku
• gtalk
• chesspark
• fire eagle

Enterprise Mashups – Hype and Reality

speaker (John at ProgramableWeb.com) (see slide share to get the slides)

Summary: the use of mashups, web services, WOA , openAPI’s is growing rapidly allowing enterprises to build ad hoc systems very quickly (days instead of months); we are in the early stages of this technology; will it become the next “Excel” of business (where business users build their own web apps rather than waiting on IT?)

What – housingmaps.com started it all about 3 years ago

combine web servies into something new (consumer mashup)

• - lightweight app
• - created in days
• - Web Oriented Arch
• - internal + external services
• - done at data, logic or presentation layer

examles

• workflow mashup – 80% of 72K applications e-signed in 2 months, 75% cost reduction
• portal mashup – took 3 days to build, last one took 3 months
• CRM mashup – cleaner data & less data entry time; 2 weeks to create
• BI mashup – reduce IT backlog; their IT department said $500K to build; they did in 4 days (Audi)

takes only days or weeks, rather than months or years

bottom-up, ad-hoc, less structure, grassroots,

What is driving this?

• Open APIs
• SaaS
• Rich Internet Apps
• Dynamic Scipriting
• cloud computing
• widgets

rather than calling OS & DB apis, we are calling web services

history

• ebay 2001
• amazon 2002
• salesforce 2003
• google, skype, flickr, yahoo 2005 (105 apis listed in PW)

they showed number of companies providing web apis by sector (email, widgets, mapping, shopping, photos, community, music, messaging, financial, telephony, enterprise, reference, mobile, messaging)

enterprise mashups

• CRM
• HR
• Doc Mgt
• Conferening
• Billing & Customer
• salesforce has over 40,000 users
• 1 billion api calls/month
• virtualization
• project management
• software development (assembla)
• email serivces (exacttarget, whatcounts, verticalresponse)
• call centers (callfire)

SOA – is this the last mile of SOA?

who creates them – IT users or business users? trend toward business users, but IT users are skeptical

will mashups become the new Excel?

technologies and tools

web oriented architecture (woa)

• REST: get, post, put, delete (web becomes read/write media)
• URI’s, URL’s
• POX (plain old XML)
• RSS, Atom, JSON

SOA vs WOA – both will coexist

SOA – static programming model

WOA – dynamic programming model

SOA uses SOAP

WOA uses REST

Tool Groups

• data mashup tools – beta pipes, rssbus, denodo, newsgator, apatar, snaplogic
• scraping tools – kapow, dapper
• development tools – proto, google, serena, etelos, tibco, teqlo, bungee labs, jackbe
• DIY consumer tools – platial, my maps, popfly
• services – xignite, strikeiron, mashery, webmetrics
• what binds them together: HTTP, REST, SML, RSS, Aton, Ajax, XHTML

snaplogic – inputs, operations, outputs

inputs – databases, feeds, xml

operations – merge, filter, sort, convert

outputs – xml, rss, atom

kapow – data extraction (scraping) tool

suites – IBM lotus mashups, infosphere (the glue), websphere sMash

data as a service – StrikeIron, xignite

where will you build and run them?

• cloud app builders (longjump, coghead, rollbase)
• cloud IDEs (force.com, bungeeconnect)
• cloud computing (joyent, amazon webservices, mosso)
• managed hosting (opsource)
• do-it-yourself

Challenges

• immature marketplace (early adopter phase, much change)
• SLAs for APIs (no sla’s)
• Security
• Data quality and trust

Advice for IT

• beware of the hype but don’t ignore
• Got SOA – make it a mashup platform
• start simply
• think tools
• add goverance as needed

API’s are becoming the glus of Software as a Service (SaaS)

SLA’s are starting – amazon and google maps are now offering

status.aws.amazon.com

vendor software is becoming mashups

• business intellegence (working with google maps)
• ms office + ebay + paypal
• IBM webshere portal + google gadgets
• FaceForce: Salesforce.com + Facebook