PrestigeSoft

1
PrestigeSoft

• PrestigeSoft is 7m US-based company
• Within the PrestigeSofts niche, time-to-market
  is very critical
• Entire programming unit in Sofia, Bulgaria
• Develops low-end, low-cost consumer software
  packages
• Most projects are completed on time for market

2
PrestigeSoft

• Success factors
• Learn how to work with each other
• The products themselves are very small
• Ability to take advantage of the ten-hour time
  difference between the US and Bulgaria for rapid
  iterations

3
Orchestral Technologies

• A major US-based company that specializes in
  graphical software
• Have joint venture in Moscow, Russia
• Moscow staff perform the traditional job of
  offshore programming.
• Write codes from spec and develops small
  components for larger products in California
• Expanded to St. Petersburg

4
Orchestral Technology

• Business Developments
• The Russian site was tasked to developed a large
  (one million lines of code) brand new product
  (release 1.0)
• A global team was lunched into this large scale
  collaborative project with sites
• St. Petersburg (Russia), California
  (headquarters), Denver (US, consultant), Regina
  (Canada).
• Design was shared by California and Moscow
• Code was shipped back and forth (across 11 time
  zone) everyday during the coding phase

5
Orchestral Technology

• Factors
• Software configuration team was set up at each
  site
• A policy was instituted whereby code would lock
  up at 8pm at either California or Russian offices
  and nothing further could be added to it.
• Transfer of the codes after an hour by satellite
  to the other team
• Programmers coordinate code sharing via detailed,
  formalized notes to each other
• Programmers dont work on exactly same code

6
Orchestral Technology

• Result
• Completed on time
• At a lower cost
• No immediate follow-up bug release
• Acquired Software African development site was
  added to the setup

7
IBM Global Development

• Five-site, Five-country, development project with
  the round-the-clock aspirations
• The project involved developing small components
  for the IBM VisualAge application development
  environment
• The idea was to develop many small components
  called beans, where each bean will be developed
  rapidly
• Fatware vs. component-based beans
• The IBM JavaBeans was envisioned to be plugged
  together and used to develop functional business
  applications

8
IBM Global Development

• Organization Structure
• Multi-sites with four equal size units of 31
  professionals
• Joint venture of Tsinghua University in Beijin,
  China
• IBM Tata joint venture in Bangalore, India, later
  wholly owned IBM Global services India.
• IBM joint venture firm with the institute of
  Computer Science in Minsk, Balarus, and
• Privately owned SWH Group in Riga, Latvia.
• Why 31 developers at each site?

9
IBM Global Development

• Phalanx
• Close formation of specimen carrying overlapping
  shields
• A close-knit or compact body of people,
  especially, one unified by a common goal
• Example, a bone of a finger.

10
IBM Global Development

• IBM Phalanx (Team Structure)
• In each site there where five core specialist
  area of five people each in such areas
• The vision was to create a reusable team
  structure
• The central group is in Settle, US with 24-person
  to
• Initiate, review, allocate, and provide
  specialized services
• The Settle team played the dual role of both
  architect and users.
• Project management and other services were also
  centralized in the US site including QA and user
  interface specialists.

11
JavaBeans project
Riga SWH Group
HQ Raleigh, North Carolina, USA
Beijing Spin-off with Tsinghua University
Minsk Joint venture with Inst. of C.S.
Bangalore Joint venture with Tata
Source Erran Carmel
12
IBM Global Development

• Development Method
• Follow-the-Sun
• Product specification and development was to be
  iterative, until completion
• US command unit would set up a work specification
  for each JavaBean and assigned it to an offshore
  site, to be turn into code by the end of the day
  and ship it back to the US for successive rounds
  of review and feedback
• After reviewing and testing the code the US unit
  would specify new changes and send those back
  across the ocean for another iteration
• This was to done on daily bases

13
IBM Global Development

• Challenges and Outcome
• Daily turn around was too hard to achieve on
  daily bases for both sides
• Changed to two code drops from each remote site
  per week
• About half a dozen beans were assigned to each
  site
• So while Beijing is waiting for the review and
  feedback on component 1, they worked on beans
  2..6.
• At steady state, the project was juggling about
  two dozen beans between the hub and the remote
  sites

14
IBM Global Development

• Changes in Team Structure
• The original project champion (US-based) left in
  mid 1997.
• The new project manager saw the entire project in
  a different way. That
• A true global development should not be strongly
  centralized but more more of a network
  organization structure
• In which global partners coordinate more
  activities among themselves rather than through a
  central unit
• A large control unit in the US defeats the
  purpose of a low-cost development project

15

• Changes
• The US command unit was moved from Settle to IBM
  development center in Raleigh, North Carolina
• Reduced from 24 people to only 3 (Global manager,
  budget officer and a chief technical architect)
• US site is now a project management unit
  responsible for oversight and some review
  function
• With the new structure the non-US sites began to
  initiate the projects

16
IBM Global Development

• Results and Challenges
• The US hubs overall workload was reduced
  significantly
• The Minsk site became responsible for the
  acceptance testing and integration for all beans
  (in addition to some of its own development)
• Hub delegate tasks to other sites, and
  follow-the-sun-was de-emphasized and no longer
  practical.
• Occasionally time zone advantages still took
  place.

17
IBM Global Development

• Result and challenges
• Simultaneous development in five nations with
  inexperienced partners.
• Project objective was to build leading edge
  technology (i.e. software components) while
  experimenting with a new management model
• JavaBeans are small components that are easier to
  define and for which the development time is
  short.
• Once finished it is able to communicate with
  other beans.
• If too many beans are been developed at the same
  time integration and coordination could become a
  problem

18
IBM Global Development

• Global Development Justifications
• Compelling reason was to take advantage of low
  cost development centers
• 8,000-15,000/programmer/year
• 10 of the cost in US
• Plus travel costs, etc.
• To reduce time to Market using follow-the-sun
• Estimated 35 cycle time reduction before the
  project structure changed
• Nevertheless, beans development allows for very
  rapid development
• Is the global development project a fruitful
  project?

19
IBM Global Development

• Hidden Goal
• Cost and time-to-market for this project were not
  the only strategic drivers of the IBM JavaBeans
  project
• IBM was seeking a strong presences in all of
  these emerging markets in order to develop a
  strong base of developers who have experience
  working with IBM on important projects.
• This would allow IBM to continue to expand in
  these markets
• China, in particular, was an important market for
  IBM.
• All five sites were connected via local internet
  provider to IBM global network.
• Tools used include Lotus Notes and other
  collaborative technologies.

20
Largest American IT Professional Services
FirmsMassive offshore centers

• Mexico, India, Argentina, Brazil, Latvia,
  Hungary, others
• Calcutta, Korea
• India, Canada, Brazil, Ireland, others
• Manila

21
Are Global Software Teams Successful

• Success is measure by four key parameters
• Was the product delivered on time?
• Was cost reduced?
• Was the product Innovative?
• Was the product relatively free of bugs and
  according to specifications?

22
Are Global Software Teams Successful

• Was the product delivered on time
• Follow-the-sun development may not reduce the
  cycle time for some projects
• True follow-the-sun requires a great amount of
  communications and understanding
• True follow-the-sun with daily hand offs is very
  difficult for programming, particularly programs
  involving any complexity
• Works with customer support issues and bug fixing

23
Are Global Software Teams Successful

• Were development costs reduced?
• Travel and hotel
• Communication set up

24
Are Global Software Teams Successful

• Was the product innovative?
• Higher level design
• Full ownership of key products
• New initiatives from global site that later turn
  into a product

25
Are Global Software Teams Successful

• Was the product relatively free of bugs and
  according to specifications
• Home country often take responsibility for
  testing
• Development standards are often higher with
  global teams

26
Are Global Software Teams Successful
Largest American Firms Massive offshore centers

• India About 400 million per year 1800 people
  in 99
• Mexico
• Chennai India 700-1000 staff, launched in 2001,
  handling CAD/CAM, e-mail processing, and
  application development

GE
27
Are Global Software Teams Successful

• India
• Exporting 6 billion in software services and
  products.
• Producing about 50,000 new, well-trained software
  engineers per year

28
Are Global Software Teams Successful

• Israel
• Exporting 15 billion in high-tech.
• 2000 high-tech firms (larger than any nation
  except for USA).
• More off-shore American RD centers than any
  other nation.
• 40,000 IT professionals

29
Are Global Software Teams Successful

• Russia
• Much smaller IT sector than the 3 Is
• ranks 3rd in the world in scientists and
  engineers
• MNCs Intel, Motorola , Sun, others
• Concentrated in, Moscow, St. Petersburg, and
  Novosibirsk
• about 100-160 firms none are large
• Wages 20gt India
• Weaknesses
• Lack of experience in managing the complexity of
  high level quality processes, especially over
  distance.
• Poor English skills
• Access to Russia -- The need for visas.
• Expensive bandwidth
• Lack of quality certification firms
• Legal issues software piracy, export and import,
  taxation, labor law, company registration,
  currency control.

30
Are Global Software Teams Successful

• Ireland
• Isle of localization and call centers
• 25000 software professionals with 3000 now coming
  out per year

Source McCaffrey 99, Cochran 01
31

• Are Global Software Teams Successful?

32

• What is your predictions for Global Software
  Teams?