Ohjelmistoprosessien kehitt

1
Ohjelmistoprosessien kehittäminen ja tutkimus
Nokia-yhtymässä

• Kari Känsälä
• Senior RD Manager
• Nokia Research Center
• Blanko '98 -päivät, Oulu, 22.-23.10.1998

2
Sisältö

• Yleisiä piirteitä Nokian ohjelmistokehityksestä
  ja -prosesseista
• ohjelmistot Nokian tuotteissa
• Nokian ohjelmistokehitys numeroina
• tyypillinen nokialainen ohjelmistokehitysprosessi
• ohjelmistoprosessien haasteita Nokian piirissä
• Ohjelmistokehityksen ja -prosessien sekä
• niiden tutkimuksen (R), kehittämisen (D) ja tuen
  (S) organisointi
• liiketoimintayksiköt (Business Units, BUs) vs.
• Nokia Research Center (NRC)
• NRCn nykyinen ohjelmistoprosessien
  tutkimus/kehittäminen kohdealueineen
• kolme Casea (prosessien arviointi, prosessien
  mallinnus, projektien estimointi)
• Yhteenveto
• seuraavat kalvot ovat englanninkielisiä

3
Software in Nokias products

• All main products include software (SW)
• Fixed networks
• Cellular (mobile) networks
• Telephone switches (DX200)
• Professional mobile radios
• Intelligent network products
• Wireless data products
• Mobile phones
• Multimedia terminals, satellite receivers
• Car terminals
• Amount of SW in a product ranges from
• 1000 lines of SW source code (1 KLOC) to
• 30 000 000 lines (30000 KLOC)

4
Software development effort in numbers

• Number of SW developers (developing SW for
  Nokia's products)
• has grown rapidly from
• about 1000 (1990) to
• about 7000 (1998)
• represents now
• more than 50 of all RD effort
• more than 15 of Nokias all employees
• Location of software development
• 70-80 yet in Finland
• foreign sites growing rapidly in Europe, USA,
  Asia and Australia

5
Typical product development process

• Products are typically delivered to customers in
  subsequent releases
• Release cycle may range from 6 months to 2 years
• For each product release we
• gather customer requirements from all over the
  world
• analyze and prioritize customer requirements
• create product specification
• create product design (allocated to hardware (HW)
  and SW)
• create SW in SW projects (in parallel with HW
  projects)
• create SW specification
• create SW design
• implement SW design ("coding module/unit
  testing")
• test SW ("integration testing" or "functional
  testing")
• test product ("product testing" or "system
  testing")
• deliver product to customers

6
Challenges for software development

• Most BUs have an ISO 9000 certified quality
  system covering product development
• Nokia Software Process Initiative (NSPI) aims at
  improving
• productivity,
• quality, and
• predictability of
• SW development, and at enhancing the
• capability of SW development processes
• in Nokia's Business Units (BU)
• SW development process capability is assessed
  using the SW Capability and Maturity Model
  (SW-CMM) of the SEI/CMU
• level 1 basic SW development capabilities
• level 2 excellent SW project management
• level 3 excellent SW process
• level 4 quantitative SW management (based on
  excellent SW metrics)
• level 5 optimizing SW management (based on
  excellent SW management)

7
SW RD in Nokia

• SW development
• for Nokia's products
• primarily in RD units of BUs
• major part of new SW technologies RDS in
  NRC's laboratories

• SW process/engineering RDS
• valid SW process DS exists in all RD units
• most of SW process R, large part of Dand some
  of S done by NRC

NMP, NTC, NCP, NVO
RD RD RD
R D
NRC
R
International RD co-operation
Universities, research institutes
8
NRC's current SW process research

• Focus areas in NRC's SW engineering RDS
• SW architectures and reuse
• user interfaces and usability
• embedded SW methods (especially Octopus)
• SW engineering environments (especially Telecom
  Design Environment, TDE)
• SW process improvement, including
• SW process assessments and metrics
• SW process modeling and definition
• SW project effort estimation and metrics
• requirements management
• SW testing and test management

9
Case 1 SW process assessments

• Background
• most BUs were ISO 9001 certified in the beginning
  of 90ies
• some BUs started to use SW-CMM in 1993-1995
• Status in 1998
• SW-CMM is now the de-facto std of SW process
  assessments in Nokia
• there are 3 Nokia-std CMM assessment methods
  CBSA, CBPA and CBA-IPI
• a large pool of Nokia-internal CMM assessors
  exist
• CMM briefings or basic training has been given to
  hundreds of Nokians
• NRC's role since 1996
• method selection and fine-tuning (CBSA, CBPA)
  using SEI's CMM material
• training (CBSA, CBPA) and training facilitation
  (CBPA, CBA-IPI)
• leading of or participating in assessment teams
  (all 3 methods)
• reference Kari Känsälä, "Utilization of the CMM
  within Nokia." SYTYKE seminar, Helsinki,
  7.9.1998.

10
Case 2 SW process modeling definition

• Background
• ISO 9001 was the basis for "quality systems" of
  BUs in the beginning of 90ies
• RD units of BUs created a set instructions for
  SW development, in many cases as an integral part
  of the quality system
• Status in 1998
• most RD units have a "SW process manual",
  accessible on Nokia Intranet, and (loosely or
  tightly) linked to the quality system of BU
• NRC's role since 1995
• some major BUs (e.g. Nokia Mobile Phones) has
  been supported in establishing a tightly
  integrated SW Engineering Process (SWEP) Manual
• based on the ISO/SPICE v2.0 process reference
  model
• based on SEI's SW process modeling definition
  recommendations
• using Intranet as the UI and Lotus Notes as the
  development platform
• reference Kari Känsälä Atte Kinnula,
  "Restructuring a Software Process." SPI97,
  Barcelona, 1.-4.12.1997. (There will be an update
  in SPI98.)

11
Case 3 SW project effort estimation

• Background
• effort/duration estimation used to be based on
  expert knowledge and task-level bottom-up
  estimation techniques
• a BU started to develop a "home made" effort
  estimation method based on regression analysis of
  existing SW project data
• Status in 1998
• most BUs have defined an explicit procedure for
  effort/duration estimation
• many BUs use Function Points Analysis (FPA)
  related effort estimation methods
• NRC's role since 1994
• a Nokia-specific FPA related effort estimation
  model was created by combining the "home made"
  method of the above-mentioned BU and the FPA
  theory
• several related models were created for other BUs
  and/or for specific purposes
• now an integrated method is created to cover
  existing methods and original FPA
• reference Petri Vesterinen and Kari Känsälä,
  "Customization of FPA at Nokia." COCOMO Forum,
  Los Angeles, 9.-11.1996. (Updates at later COCOMO
  F's.)

12
Summary

• SW is very important for Nokia, and will be even
  more important in the future -gt Nokia's SW
  development processes must be faster and more
  productive -gtSW process/engineering improvement
  must be organized efficiently
• Nokia's Business Units (BUs) take care of
• SW development for their products, and
• own SW process/engineering development and
  support
• Nokia Research Center (NRC) supports BUs in
• new SW technology RD and support, and
• SW process/engineering research and development
• NRC takes care of
• identifying new SW process/engineering
  innovations, and
• modifying them to applicable solutions for BUs
• in selected key areas
• in cooperation with corresponding Finnish and
  foreign organizations