International Issues in DIS

1
International Issues in DIS

• Week 13 - Lecture 1

2
DIS cross country boundaries

• Either with one set of application and database
  servers for a group of countries
• Or, with separate application and database
  servers in each country.
• Often not cost efficient
• Transactions and queries cross country boundaries
• Applications should be designed for
  implementation in different countries

3
We will consider the issues not the solutions

• As the solutions often depend on the
• Operating system
• DBMS
• Application

4
Textbooks and software originates in the USA

• The US market is big, and international aspect
  have often been overlooked
• Textbooks dont cover these issues
• But major software houses are much better now
  than five years ago

5
Issues

• Currency
• Calendar
• Character sets
• Language
• Legal Accounting
• Privacy legislation
• Cultural commercial

6
Currency

• Software often handles this reasonably well
• Local currency symbols are ambiguous
• for US, Australian and New Zealand dollar
• Newspapers often use US, A, NZ
• ISO standard is USD, AUD, NZD

7
Field sizes in many packages are a trap

• Often defined for USD but Indonesian Rupiah,
  Italian Lira and Turkish lira need four more
  digits, plus the currency symbol
• USD999,999,999,999.00
• ITL9,999,999,999,999,999
• IDR99,999,999,999,999,999.00
• TRL999,999,999,999,999,999.00
• Database, screen reports have to allow for these

8
Currency conversion

• Transactions normally recorded in the currency in
  which it is negotiated
• Usually this is the currency of the country
  making the sale, but not always
• Countries with volatile currencies often use a
  major currency for big contracts

9
Accountants have conversion rules

• Reporting usually requires reporting in the
  currency of the country concerned or the
  organisations global currency usually the
  currency of the dominant country
• But it cant be done at reporting time and should
  be stored on the transaction
• Often each transaction is converted at the rate
  on the date of the transaction, but some
  organisations use a set rate for the month or
  quarter
• Assets and liabilities have to be converted
  annually
• Debts raised in one currency and settled in
  another result in exchange differences

10
Calendar

• The de facto international standard is the
  Gregorian calendar - the Julian calendar with
  minor modifications by Pope Gregory in 1582
• There are others - Chinese, Japanese, Thai,
  Persian and Arabic calendars
• Oracles DBMS will convert if asked

11
Other calendar problems are

• The different numeric date formats
• Australian DDMMYYYY
• The US MMDDYYYY
• European YYYYMMDD
• Probably the DD MMM YYYY is the most unambiguous
• But then language affects the month abbreviations

12
Week numbering is sometimes used in systems

• The English tradition is that the week starts on
  Sunday
• But the ISO standard has the week starting on
  Monday
• Depending on which day the 1st Jan falls on, this
  can result in different week numbers

13
Encoding schemes

• Computers were initially designed to have one
  byte with 8 bits and 256 combinations represent
  the different characters
• The 7 bit ASCII and IBMs 8 bit EBCDIC were the
  two most common standards
• But most countries dont use the standard Latin
  character set
• Oracle lists some 178 character sets, see
  http//otn.oracle.com/tech/globalization/pdf/Unico
  de.PDF for a good paper on character sets.
• Another useful paper on character sets is
  http//www.unicode.org/standard/principles.html
• http//www.microsoft.com/globaldev/getwr/steps/wrg
  uide.mspx for an overview of internationalising a
  system

14
Encoding schemes

• We now have Unicode ISO/IEC 10646
• Originally was to be 16 bit, but now
• Either 8 bit, 16 or 32 bits
• All include the 256 character Latin 1 character
  set
• Windows, XML and much other system software
  use/support Unicode
• Each character is unique and given a plain text
  name, leaving the font to assign a glyph

15
Sorting is another issue

• Binary sorts often do not work with other than
  the 26 character Latin alphabet
• An ORDER BY would produce ABC, ABZ, BCD, ?BC, as
  ? has a higher numeric value than B
• Linguistic sorts are therefore required, but
  these may not work with multi-byte character
  sets???
• A Linguistic sort will handle special cases such
  as the Spanish ch which is treated as one
  character and should sort between c and d

16
Language

• Ideally, if a database is to be updated by people
  from a number of countries, each should be able
  to do so in their own language.

17
But there are problems

• Imagine if the staff in multiple countries
  recorded their names in their own language (and
  character set) and we asked for a list of all
  staff could we read it?
• But the same system would need to print
  employment contracts in the local language
• One approach is to record some details twice in
  the local language and in a common language

18
More on language

• Messages can be generated by the client O/S, the
  Server O/S, DBMS and the Application. Many do
  allow language selection. In house systems have
  to provide their own facilities
• Screen prompts, report headings and other text
  also need to adapt
• Conversion at the Web server or in the Browser is
  an emerging approach
• Has to adapt to the user
• O/S commands and DBMS keywords often have to be
  in English. DBMS tables names etc have to be in
  single byte characters

19
Legal and accounting differences

• To an IT person they seem obtuse and trivial
• But they are important and the organisation can
  be subject to heavy penalty
• The systems architect needs to work with legal
  and accounting people, and get sign off on the
  solution
• Make sure you understand the issue as sometimes a
  simple IT solution can save accounting a lot of
  work

20
Legal requirements

• Taxes. Many countries have GST or VAT but they
  are all different
• Layout of tax invoices
• Calculation of taxable amounts
• Who or what is taxable or exempt
• Government reporting is different
• Italy requires a daily list of all transactions
• Eastern European countries still require the old
  communist reporting strict cash format, with
  huge fines for minor infringements

21
EEC Privacy regulation

• Each country implements it differently
• Restricts personal information from going to
  countries without similar controls
• Requires secure data transmission and secure
  processing
• Limits type of data that can be collected
• Prevents data from being used for other purposes
• Other countries have privacy legislation but
  often it has a different emphasis i.e. in the US
  data collected about people belongs to the
  collector

22
Cultural Commercial differences

• Some examples are
• How correspondence is addressed, the format of
  the name, the layout of the address, the size of
  the envelope
• Method for settlement of a debt
• US uses direct credit
• Australia uses credit cards, cheques and
  increasingly electronic means like BPay
• Europe has the Giro