ApplianceBased Computing

1
Appliance-Based Computing
2
Reference

• Appliance Data Services Making Steps Towards and
  Appliance Computing World by A. Huang, B. Ling,
  J. Barton and A. Fox

3
Introduction

• A digital appliance is defined as a
  single-purpose device.
• Examples include the following
• Network-enabled household appliances like
  NetFridge
• Single-purpose computing facility such as a
  storage server that can be plugged in and used
  with minimal configuration.
• Single-purpose handheld devices such as digital
  cameras and PDAs

4
Introduction

• Appliances are often thought of as
  special-purpose devices that are less
  computer-like than general-purpose PCs but they
  are not embedded devices.
• General purpose PCs may be more useful when
  connected to the Internet, but they are useful
  without the Internet.
• An ActiveBadge is useless without infrastructure
  software for location.
• An appliance such as Netfridge would fall in
  between.

5
Introduction

• We can classify appliances by their degree of
  sharing.
• Mobile devices (e.g., PDA)s have little or no
  sharing.
• Environmental devices (e.g., household
  appliances) is explicitly shared among a few
  people.
• Enterprise devices such as storage are shared
  among many people.

6
Introduction

• The intention of digital appliances is the
  following
• Make our life easier
• Provide powerful improvement over their
  non-digital counterparts.
• Infrastructure enabled
• Able to leverage computational power, network
  b/w, content and aggregate user base of services
  in infrastructure
• Example Viewing Web content on handhelds made
  possible by transformation proxies.

7
Problems with Digital Appliances

• Usability
• Often the digital appliance is more difficult to
  use.
• May have too many features.
• Lack of an infrastructure
• Quote from Dan Carp, CEO of Kodak
• The industry has made picture-taking more
  difficult and more complicated by cramming onto
  digital cameras more features, more buttons and
  more bells and whistles than most people want or
  needThe one lesson that 100 years of consumer
  marketing should have taught us In the picture
  business, simple tramps megapixels, every time.

8
Illustrating the Problem
Move this picture from my camera to my Web page
Display the notes taken on my PDA on this wall
monitor
9
Illustrating the Problem
10
Illustrating the Problem

• Current process for putting pictures from digital
  camera on web page
• We take the picture
• Transfer it to the PC
• Transform it
• Upload it to destination
• Edit the HTML
• Desired
• Take the picture
• Specify the destination
• Next thing you know its at the desired location.

11
Appliance Computing World

• Vision of Appliance Computing World
• An appliance computing world is one in which
  people move data effortlessly among artifacts to
  accomplish a variety of simple and advanced tasks
• Attributes
• Some attributes are inherent to the way people
  interact with appliance computing world. These
  are discussed next.

12
Attributes of Appliance Computing World

• People move data using concrete artifacts
• People think of devices in terms they understand
• People dont care how pictures are stored, but
  only that camera takes a picture, film stores it,
  end of story.
• People dont care what format the file is in.
  They want it to be on a webpage.
• Devices are simple, single-purpose appliances
• Moving functionality from the computer onto a
  camera complicates things, as opposed to making
  life easier.

13
Attributes of Appliance Computing World

• People perform a variety of traditional tasks, as
  well as a new set of advanced tasks with their
  devices
• The new cool digital device needs to perform at
  least the tasks an old, traditional device was
  able to perform and then some
• Tradeoff between this, and the previous two
  attributes

14
Principles That Enable Appliance Computing

• Today dealing with digital devices is difficult
  since extracting information from a digital
  device usually requires interacting with a PC.
• The user has to deal with both file location and
  file format conversions. For example, a user may
  have to take TIFF files (from a scanner) and
  convert it to GIF or JPEG for a web page.
• Bring devices to the forefront
• User deals only with source/destination, and not
  with intermediate computation.
• A files actual location and format should be
  hidden from everyday users.

15
Principles That Enable Appliance Computing

• Some features can make a device easier to use.
• Examples The voice command on cell phones or the
  picture previewing on digital cameras.
• This can still be a burden. For example, the
  digital camera user has to learn how to set the
  modes on the camera to preview a picture.
• Keep the number of user-controllable features
  users must learn to operate a device to a minimum
• User should not have to know anything about
  preview camera mode to turn camera on and off.
• There is tension between ease of use and placing
  potentially useful features on the device.

16
Principles That Enable Appliance Computing

• The emphasis is on making the steps required to
  perform a high-level task simple.
• The implication of the previous two principles is
  that devices are simple.
• What if a user wants more complex functionality?

17
Principles That Enable Appliance Computing

• Place the software required to accomplish tasks
  in the Internet infrastructure.
• Logically centralized software (easy upgrades)
• Functionality is moved from the PCs and devices
  to the supporting infrastructure.
• This can make upgrades and administration
  simpler.
• It is easier to reason about reliability and
  availability.

18
Appliance Data Services

• Appliance Data Services refers to a general
  application framework on top of which appliance
  computing applications are built.
• The framework implements previously mentioned
  principles

19
Scenario Used

• Jane is traveling through Europe.
• During the trip, she uses her digital camera to
  take pictures and her PDA to jot down
  descriptions of her pictures.
• She periodically goes to an ADS access point at
  an Internet Café.
• Jane logs in by inserting her SmartCard and
  entering her PIN.
• She selects the photoalbum application.
• She is prompted for her pictures followed by the
  descriptions, both of which she transfers using
  the IR ports on her digital camera and PDA.

20
The Architecture
21
Architecture Basic Data Unit

• Basic data unit controlling framework operation
  is a triplet containing the following fields
• User identifier
• Command to be executed
• Data to be operated on
• The triplet is useful for
• Application selection (command-tag)
• Access Control (userid)
• Other service features (command-tag userid)
• Billing, security

22
Architecture Main Stages

• Data Receive Stage
• Collects data, and transfers to next stage once
  userid/command-tag/data received
• Application Control Stage
• Determines desired application, determines all
  needed parameters, sends to Services Execution
  Stage
• Services Execution Stage
• Invokes required service on a template it
  receives
• Services are very modular and composable

23
Architecture Data Receive Stage

• Access Point
• Receive data from Appliances
• Isolates device heterogeneity to a single
  architectural component.
• Possible Implementations
• A commodity PC outfitted with the appropriate
  hardware interfaces or it can be designed as a
  special-purpose network appliance.
• Example The Access Point Jane uses requires a
  Smartcard reader, IR receiver and software and a
  touch-screen monitor to display Janes
  application options.

24
Architecture Data Receive Stage

• Access Point (continued)
• The software on the Access Point is organized as
  a set of device adapters, each enabling the
  access point to speak a different device
  communication protocol.
• This isolation of device heterogeneity to one
  component allows the rest of the system
  independence from device-specific communication
  protocols.
• Key Challenge Extensibility in supporting
  devices and protocols this arises from the lack
  of standardization among device vendors and the
  increase in the variety of devices being
  introduced.

25
Architecture Data Receive Stage

• Aggregator
• Manages session state
• Gathers data sent from the Access Point and sends
  to the data to the next stage once all pieces of
  (userid, command-tag,data) triple are received.
• Simplifies adding support for new devices in a
  robust, easy-to-deploy way
• Scenario
• Jane enters janedoe for the user identifier and
  photoalbum for the command-tag.
• The aggregator maintains this state so that each
  picture and text description from Jane creates a
  triple (janedoe, photoalbum, picture or
  description). This is forwarded to the next
  stage.

26
Architecture App. Control Stage

• The user identifier and command tag are used to
  determine the chosen application.
• Command Canonicalizer
• Converts command-tag from its original data type
  to plain text
• Facilitates the design of devices with simple
  user interfaces
• Example from scenario
• Jane takes a picture and speaks the desired
  command-tag into the camera.
• When the pictures are transferred into an Access
  Point, the command-tag has been specified.

27
Architecture App. Control Stage

• Template Database
• This is where templates are stored.
• Templates are looked up by command-tag userid
• Templates describe applications behavior by
  describing data, and specifying services to be
  performed.
• Minimizes device configuration
• Provides a level of indirection between
  application selection and application
  specifications which separates the concerns of
  applications users and applications creators.

28
Architecture App. Control Stage

• Template Database (continued)
• Templates may be developed by the manufacturer
  and placed within the infrastructure.
• Customers can get new applications (or an
  upgrade) without having to worry about
  configuring the applications on their devices.
• Example
• Kodak wants to make a set of ADS applications
  available to customers who purchase a Kodak
  camera.
• If templates are shipped with the application, it
  makes it more difficult to upgrade or add new
  applications Pirating is easier.

29
Architecture App. Control Stage

• Dataflow Manager
• Coordinate data received from the user to make
  certain that an application has all the data it
  requires.
• Uses application template to place data into
  proper parameter slots.
• Data from different devices at different times
  may be coordinated.
• Example
• Photos could be input from the camera, and their
  description from the PDA at a later time. Later
  these can be combined.

30
Architecture Services Execution

• Application Dispatcher
• Invoke the services specified in the application
  template on the data it receives.
• Example
• For Janes photoalbum application, these services
  include a service that scales her pictures by an
  appropriate amount and FTPs the pictures and
  descriptions to the appropriate site.

31
Developmental Experience

• Set of services and device adaptors to support
  two applications Web Photo album and Guest book
• Web Photo album
• Users create and publish web-based photo albums.
• Guest Book
• Takes input from a web cam, business card scanner
  and PDA to create a web-based guest book
  containing peoples pictures and business card
  information.
• XML application template used to describe the
  input data and services to invoke.

32
Conclusion

• Does this make good business?
• Could potentially be useful as digital devices
  becomes more widespread.
• Just how powerful is this model of computing?