Networking and the Internet 6

1
Networking and the Internet (6)

• Last Week
• Compilers and linkers interpreters (Willis,
  ch.7, sect 7.7)
• Interrupts and status processor modes (Coope,
  chapter 9)
• Module assignment
• HTML Frames and Practical
• Checkpoint review of the module so far
• Introduction to Glossary in topic 0 of Learning
  Network
• Week 6 Focus
• Data integrity in a networked environment (first
  part)
• HTML Practical
• Workshop for assignment due Thursday May 7th

2
Review of the Module

• So far we have covered
• Uses of networks in Business
• Introduction to e-Business
• What is the ideal B2C product?
• Systems for Networks
• Operating Systems and File organization
• CPU design and Performance issues
• Compilers and linkers interpreters
• Interrupts and vectors processor modes
• Online Computing (Interactive versus
  Transactional)
• Time-sharing, multi-programming, transactions
• HTML
• basics, links, tables, Frames, practical
  exercises

3
The Rest of the Module

• Still to Come
• Data integrity in a networked environment
• More on the seven-layer model of Computer
  Networks
• LAN and WAN technologies, Network security
• Data transmission
• Network types
• Relieving bottlenecks
• Communication protocols and formats for
  transmission
• Voice-Data integration
• Anything else we should be covering?
• What changes are needed?

4
The Glossary

• The class of 2004 suggested having a single
  source for all the abbreviations and terms used
  on this module
• My start was improved by input from Stephen White
• And Ive added more as new topics have come up
  since
• This Glossary is held on the Learning Network and
  at http//www2.winchester.ac.uk/bm/Courses/BS2911/
  BS2911-Glossary.doc
• Please suggest any other entries needed
• Ideally provide an entire entry term and
  definition
• It doesnt have to be perfect, as I may be able
  to polish it
• e-Mail is probably the best way to do
  this(though we could build it by a conference or
  a Wiki on the Learning Network)
• We can print it when weve covered your key
  concerns

5
Goals of Distributed Systems

• To get the best out of both servers and client
  systems
• Fast response times to user
• Data integrity in the system as a whole
• High availability and reliability
• Low maintenance costs
• Server offers
• Secure, fast access to data
• Integrity and effective back-up
• Access independent of workstation
• Client workstation offers
• User-friendly interface
• Local processing where central data not needed
• MIPS at lower cost than on mainframe systems
• Meaningless Indicator of Processor Speed

6
Distributed Data Models

• UNIX allows you to link data structures
• the mount command attaches a directory hierarchy
  as part of another hierarchy
• Can mount directories from other systems (a bit
  like mapping a network drive in Windows)
• Requires message interchange to maintain data
  provided by the Network File System or
  equivalent
• Many database systems provide automatic
  replication between systems
• For example, Oracle, DB2, Lotus Notes
• Locking is a serious problem with peer database
  engines
• how do you stop 2 copies of a field being updated
  at once?
• Coordinating adds network delay even to local
  updates

7
Distributing Function

• Most systems contain a mixture of
• Data Management
• Business logic
• Data presentation and user interface
• This can be distributed asymmetrically, for
  example
• Data Management on a server
• Business logic on the server
• Presentation logic on a workstation
• or you could split business logic, e.g. to
  validate input before sending it to the server
• Benefit of asymmetric distribution is that data
  integrity is under the control of a single system
• Great until the server fails!

8
Programming Models

• Function Shipping, as in NFS or CICS
• Application issues data request to environment
  (or OS)
• Environment generates messages to another system
  to request action on the data
• Transaction Routing, as in CICS
• Client system invokes a transaction thats not
  local
• CICS routes the request to another system
  (usually a server) where associated program is
  run in its entirety
• Remote Procedure Call (RPC)
• Logic on workstation invokes procedure on the
  server
• Procedure runs, and returns control (and results)
  to caller
• Environment handles RPC messages
• Permits complex structures of networked logic

or vice versa
Network File System
9
Integrity What makes it hard

• In a single system, events can be synchronized
• On uniprocessors, nothing is really simultaneous
• Inter-process communication is affordable
• The system can avoid concurrent updates by
  locking at record level, or by serializing update
  activities
• On a distributed system, simultaneity is possible
• Anything that can go wrong will eventually go
  wrong
• So we usually have an owning site for any
  data,and ship all requests there
• But theres still a problem when two separate
  bits of data are updated within a unit of work
• Credit account A with X, and debit B with X (or
  do neither)
• If accounts owned on different machines, one
  action might fail after the other has completed

10
Logic split between Systems
Inter-systemcommunication

• B is effectively working for A in this example
• During running of As program, code calls
  function on B
• B does the required activity (e.g. debit), then
  returns
• In this case, A waits until B returns
  control(but what if it ran on doing local
  things?)
• What if link fails after B has changed its data?

11
Logic split between Systems
Inter-systemcommunication
Server
Client

• A makes a request of server B
• It then waits while B runs the program
• When B returns the results, A continues working
• In principle, A could have run on doing local
  things,but eventually its going to have to wait
  for Bs response
• Integrity risk if link fails after B has changed
  its data

12
Data Integrity

• Consider a banking transaction B pays A 100
• System must decrement Bs account
• AND increment As account
• If it cant do both, it must do neither
• In real life, the actions cant be atomic, so
  transaction monitor (such as CICS) groups them
  into a unit of work
• Starts unit of work
• decrements Bs account, logging previous state
• Increments As account, logging previous state
• Commits changes (or backs out if something fails)
  at end of unit of work

13
Two-phase Commit

• One way to preserve data integrity is to back out
  any changes made if the unit of work aborts, e.g.
• System A logs the before state of the account
  it owns, then credits the account (speed limited
  by disk and need to move data between systems)
• System B logs the before state of the account
  it owns, then debits it, then notifies A that
  its ready to commit
• System A reacts to Ready by confirming commit to
  B no delay for disk activity or volume data
  transmission
• System B gets confirmation and purges its log
• Similar logic applies however many nodes are
  involved (you effectively run along a chain of
  nodes)

14
Window of doubt

• What if link goes down after A has received
  Ready to commit from B?
• Could we end up with after state at A and
  before at B?
• CICS deals with this by maintaining logs until
  the loop is closed, even if this is days later
• If B comes back and NAKs the commit, A can
  back out the change
• If B comes back and confirms, A can purge its log
• What about other transactions that hit the same
  data in the meantime?
• Can probably maintain integrity of values in the
  file
• But some decisions taken might be based on values
  that will later be backed out

See glossary
15
Synchronous or Asynchronous

• Everything so far has been synchronous
• Client cant do anything until the server has
  responded
• Thats important if youre taking decisions on
  the way
• But business often includes activities that can
  run at their own speeds
• We dont do everything by phone
• If no conversation is needed, we write or leave a
  message
• Even applies to payment we can send a cheque by
  mail
• Distributed processing can also work through
  Messaging
• We do need a way to guarantee that the message
  will arrive
• And that it will arrive once only
• Queuing is helpful to keep messages in sequence

16
Advantages of MQ Approach

• Simple
• Allows easy connection of heterogeneous systems
• Asynchronous operation is built into the model
• Coping with failure is inherent
• You never know when the message will arrive,so
  you have to design around non-instant delivery
• Network failure is simply an extended case of
  this
• Depends on integrity of infrastructure
• Since many customers and systems use same
  vehicle, problems will get ironed out quickly
• Well cover Messaging and Queuing next time

17
Module Assignment

• The Site (two-thirds of assignment marks, split
  equally between design and coding)
• Pick a topic that interests you
• Produce a site with 4-6 pages
• Work in straight HTML no Dreamweaver, Front
  Page etc
• Cascading style sheets good, but not required
• Put the site on to an ISP if you like, but also
  hand it in on the USB stick this will prove
  that its portable
• Write up your development process (one third)
• Summary of what you set out to achieve
• Review of what you discovered during development
• Reflection on what you did

18
The Mechanics of Writing HTML

• Need at least two windows open
• A browser to test your page in e.g. Internet
  Explorer
• An editor to update it in Notepad is fine
• Internet Explorer View Source opens file in
  Notepad
• Can then modify the file, Save it,
• Then check the effects in IE
• If you use frames, View Source on menu ?
  right-click View
• Reality is that most pages are based on existing
  ones
• Get file into Notepad
• Save As with new name (make sure extension is
  right )
• Modify the content as you wish including the
  Title
• You never have to bother typing ltHTMLgt ltHEADgt etc

19
Using Graphics
Never use BMP

• Pictures inserted with image tagltimg srcurl
  altalternative textgt
• Use JPEG format for photos, GIF for line drawings
• Ideally manipulate the image to the size you need
• Most systems have Microsoft Photo Editor lets
  you crop and resize images
• Can also resize with width or height attributes
• Dangerous to use both, as you can distort the
  image
• Use one or the other, or take care to preserve
  aspect ratio
• If image cant be displayed, browser will reserve
  space using these attributes, and show value of
  alt attribute
• Better to scale down than up, but file size isnt
  affected large images will still be delivered
  very slowly

20
Reminder Creating Hyperlinks

• Any hyperlink reference needs to say what it is
  linking to
• In HTML, we use an Anchor tag with the HREF
  attribute
• ltA href"http//www.pdq.edu/freds.htm"gtQuick
  Univlt/Agt
• ltA href"freds.htm"gtLocal referencelt/Agt
• You can reference any resource on the Internet
  Just give its URL
• Any lack of specificity means look locally, so
  missing out the domain means it is in the local
  file system
• Always leave out as much as you can
• Makes it much easier to move the site
• Can also use relative references, e.g. to a
  sub-folder ltA href./subdir/fred2.htm"gtrelative
  referencelt/Agt

21
Some HTML Primers on the Web

• The original NCSA primer has been removed, but
  theres a printable archive on our web-site
• Two good primers are available on-line at
• http//www.w3schools.com/html/html_intro.asp
• http//www.web-nation.com/lessons/html-pri.htm
• This is Tony Drewrys clear exposition of HTML
  tagshttp//www.cems.uwe.ac.uk/tdrewry/html/index
  .htm
• A very wide-ranging site, covering everything
  from simple HTML to XML and databases
  http//www.w3schools.com/
• Any more favourites?

22
Remaining slides are revision

• Frames topic from Week 5

23
Netscape Frames

• Typical Example of how HTML has grown
• Netscape extended HTML to address a navigation
  issue
• Other browsers climbed on to the bandwagon
• HTML standards committee added the facility
• Idea is that we can attach pages within logical
  frames inside our page
• Gives common look and feel to site
• Makes navigation information permanently
  available
• Lets you include alien material in your website
• BUT
• Burns up valuable screen real-estate

24
Frames

• Frames are a useful way to organize content of
  screen
• You can have a fixed bit, perhaps for navigation
• and what the user does there can control what
  appears in another frame
• E.g. we might have choice of what appears in the
  RH pane
• Normal way to organize modern web-sites
• Details on the web at http// sharkysoft.com/tutor
  ials/frames/
• Summary in your notes

• Basic structure is a page that defines
• how it is split into frames,
• hyperlinks to the resources that go into each
  frame
• Awkward shapes can be produced by binary chopping
• Each frame can be named
• You can control which named frame gets updated
  with next resource
• or even create a new browser window
• Best way forward is to try it

25
Learning about Frames

• As usual, the best thing is to try it with a
  simple example
• Download http//www.winchester.ac.uk/bm/courses/BS
  2911/frames.zip
• Extract everything in it into a folder on your
  machine
• Edit the file framed.html using Notepad
• Can you work out what its meant to do?
• You may need to look inside the components it
  links to
• Open framed.html
• Does it behave as you expected?
• What happens when you resize the window?
• Try to understand exactly whats happening

26
Frames Exercise

• Create a page called xxx.htm (you choose the xxx)
• Divide it into three frames
• Split the page vertically into two equal parts
• Left initially displays a trivial page you create
  call xxxL.htm
• Split the Right-hand half horizontally 7030
• Top 70 of the RH, displays a trivial page called
  xxxTR.htm
• Bottom 30 of the RH is for navigation store as
  xxxBR.htm
• In Bottom-right frame
• Provide hyperlinks to load target pages into the
  left-hand and top-right frames
• And heavier-duty hyperlinks to replace whole page
• Associate graphics with each hyperlink, e.g.
  arrows
• You can base all future frame-sets on this!