professor per runeson, datavetenskap, lth … · 2019. 10. 15. · each during in spring and summer...

10
CC alijavam at Flickr Programvaruintensiva system PROFESSOR PER RUNESON, DATAVETENSKAP, LTH [email protected]

Upload: others

Post on 31-Dec-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word

CC alijavam at Flickr

Programvaruintensiva systemPROFESSOR PER RUNESON, DATAVETENSKAP, LTH

[email protected]

Page 2: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word

PROP. 2019/20:1 UTGIFTSOMRÅDE 16

186

Regeringens överväganden

Anslaget beräknas öka med 21 397 000 kronor fr.o.m. 2020 till följd av fördelning av den ökning av universitets och högskolors anslag för forsk-ning och utbildning på forskarnivå som aviserades i forskningspropositionen Kunskap i samverkan – för samhällets utmaningar och stärkt konkur-renskraft. Av samma anledning beräknas anslaget öka med 13 616 000 kronor fr.o.m. 2020 vilket medför att ökningen som föreslogs i propositio-nen Vårändringsbudget för 2019 överförs perma-nent.

Anslaget beräknas även öka med 72 000 000 kronor fr.o.m. 2020 med anledning av den för-stärkning av det strategiska forskningsområdet it och mobil kommunikation som aviserades i forskningspropositionen Kunskap i samverkan – för samhällets utmaningar och stärkt konkur-renskraft.

Förslag till anslagsnivå och beräknad anslagsnivå Regeringen föreslår att 1 022 080 000 kronor an-visas under anslaget 2:14 Linköpings universitet: Forskning och utbildning på forskarnivå för 2020. För 2021 och 2022 beräknas anslaget till 1 038 162 000 kronor respektive 1 054 872 000 kronor. Tabell 5.57 Härledning av anslagsnivån 2020–2022 för 2:14 Linköpings universitet: Forskning och utbildning på forskarnivå Tusental kronor

2020 2021 2022

Anvisat 2019 1 897 931 897 931 897 931

Förändring till följd av:

Pris- och löneomräkning 2 17 136 32 409 47 383

Beslut 107 013 107 822 109 558

Varav BP20 3 107 013 107 013 107 013

Varav 3

Höjda basanslag 13 616 13 616 13 616

Förstärkning SFO ICT 72 000 72 000 72 000

Basanslag 2020 21 397 21 397 21 397

Överföring till/från andra anslag

Övrigt

Förslag/beräknat anslag 1 022 080 1 038 162 1 054 872 1 Statens budget enligt riksdagens beslut i december 2018 (bet. 2018/19:FiU10). Beloppet är således exklusive beslut om ändringar i statens budget. 2 Pris- och löneomräkningen baseras på anvisade medel 2019. Övriga föränd-ringskomponenter redovisas i löpande priser och inkluderar därmed en pris- och löneomräkning. Pris- och löneomräkningen för 2021–2022 är preliminär. 3 Exklusive pris- och löneomräkning.

5.7.15 2:15 Karolinska institutet: Utbildning på grundnivå och avancerad nivå

Tabell 5.58 Anslagsutveckling 2:15 Karolinska institutet: Utbildning på grundnivå och avancerad nivå Tusental kronor

2018

Utfall

690 576

Anslags- sparande

34 192

2019

Anslag

731 153

1 Utgifts- prognos

722 862

2020 Förslag 747 524

2021 Beräknat 761 150 2

2022 Beräknat 773 505 3 1 Inklusive beslut om ändringar i statens budget 2019 och förslag till ändringar i samband med denna proposition. 2 Motsvarar 748 654 tkr i 2020 års prisnivå. 3 Motsvarar 748 755 tkr i 2020 års prisnivå.

Ändamål

Anslaget får användas för ersättning till universi-tetet för högskoleutbildning på grundnivå och avancerad nivå samt behörighetsgivande och högskoleintroducerande utbildning.

Budget för avgiftsbelagd verksamhet

Tabell 5.59 Uppdragsverksamhet Tusental kronor

Uppdragsverksamhet Intäkter Kostnader Resultat (intäkt - kostnad)

Utfall 2018 350 030 346 134 3 896

(varav tjänsteexport) 116 431 116 431 0

Prognos 2019 355 000 360 000 -5 000

(varav tjänsteexport) 118 100 118 100 0

Budget 2020 360 000 365 000 -5 000

(varav tjänsteexport) 119 700 119 700 0

Regeringens överväganden

För att genomföra utbyggnaden av läkarutbild-ningen som inleddes 2018 beräknas anslaget öka med 2 290 000 kronor fr.o.m. 2020 och med ytterligare 1 145 000 kronor fr.o.m. 2021.

Anslaget beräknas öka med 998 000 kronor fr.o.m. 2020 med anledning av den utbyggnad av högskolan som inleddes 2015.

Page 3: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word

Mjukvara finns överallt“About half of Sweden’s exported products are critically dependent of software. A majority of the costs associated with R&D are already related to software development in many companies.”

Page 4: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word

Var finns (inte) kod?

• kommunikation• produktion• transport• byggsektorn• statsförvaltning• finanssektorn• media• sjukvård

• övervakning• integritet• upphovsrätt• miljö & energi• sociala

relationer• utbildning

Page 5: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word

Vad är speciellt med mjukvara?• Complexity• Conformity• Changeability• Invisibility

“No Silver Bullet – Essence and Accidents of Software Engineering”F. P. Brooks, IEEE Computer, April 1987

Page 6: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word

23INDUSTRIELL EKONOMI | SPECIALISERINGAR

Programvaruintensiva system (ps)SOFTWARE INTENSIVE SYSTEMS

Års

kurs

4

Läsperiod 1 Läsperiod 2 Läsperiod 3 Läsperiod 4

ETSN052Programvaruutveck-ling för stora system

7,5 hp (A)

ETS170Kravhantering

7,5 hp (A)

ETS200Programvarutestning

7,5 hp (A)

ETS061Simulering7,5 hp (A)

FRTN10Flervariabelreglering

7,5 hp (A)

EDAN10Konfigurations

hantering7,5 hp (A)

ETSN01Avancerad

telekommunikation7,5 hp (A)

FRTN01Realtidssystem

10 hp (A)

Års

kurs

5

Läsperiod 1 Läsperiod 2 Läsperiod 3 Läsperiod 4

MION25Teknologistrategier

7,5 hp (A)

ETN35 Projekt i Elektro- & informationsteknik

7,5 hp (A)Examensarbete

30 hp (A)

EDAF60Objektorienterad

modellering och design 4,5 hp (G2)

ETSN15 ETSN20 EITN95ETSN20Programvarutestning

7,5 hp (A)

ETSN15Kravhantering

7,5 hp (A)

ETSN10Nätverksarkitektur och

prestanda7,5 hp (A)

EDAF75Databasteknik7,105 hp (G2)

Page 7: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word

UtvecklingsprojektOrganisation:

2 projektledare3 systemansvariga

8 utvecklare4 testare

Totalt: 17 personer

UtvecklingsmodellDefinierar faser, aktiviteter och dokument

Kvalitetsuppföljning:•granskningar•mätdata

Utvecklingsmiljö:- Språk- Verktyg- OS

Uppdrag:Lägg till funktioner till ett IoT-system

Prep by: Softhouse ­­­ Doc resp: Softhouse ­­­ Date: 15­08­24 ­­­ Doc. no: 1.0.2  ­­­ Rev: PA1  

Overview  The System for the LTH PUSS course consists of one or more Frontends, each one running on a MVD device, connected to a Backend server running in Linux. The System is used to control and get data from BLE (Bluetooth Low Energy) devices. This is done by calling the Backend API which can get data from the database or send commands to the Frontend. The BLE devices used are a Wiced Sense kit and a Nexturn light bulb.   The MVD device used in this system is a regular Android smartphone, with these Frontend requirements. 

Diagram

 

Use cases

The functional requirements from a user of the system are to:  1. Turn on/off the light bulb 2. Change color of the light bulb 3. Activate/deactivate that the sensor sends data. 4. Get sensor data for different sensor types (humidity, temperature…) within chosen time 

intervals. 

System scope

The System as delivered and deployed at course start contains only the Frontend and Backend. Any Android app connecting to REST API or Web interface are not included. A specification for developing such an app is found in System specification.  No BLE devices are included in delivery of System. 

3/26 

Page 8: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word

Forskning

ELLIIT is a strategic research environment funded by the Swedish government in 2010, as part of its initiative to support strong research in information technology and mobile communications

…is Sweden’s largestindividual research program ever, a major national initiativefor strategically motivatedbasic research, education and faculty recruitment. The vision of WASP is Excellent research and competence in artificialintelligence, autonomoussystems and software for the benefit of Swedish industry.

Page 9: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word

Example research – WASP

• Continuous Experimentation Scenarios: A Case Study in e-Commerce

system, search engine, ads display algorithms, etc. Thesesoftware components are packaged as a single product thatshares behaviour data. The Apptus software product providesweb shops with intelligence and can be considered “the brain”of these shops.

Apptus has a business-to-business relationship with theircustomers who operate e-commerce web shops. These cus-tomers usually have a business-to-customer relationship withthe consumers. This study focuses primarily on Apptus’ use ofexperimentation. The consumers are the source of hypothesesfor these experiments. For Apptus, these hypotheses mainlyimpact the algorithms used in their software product, and areelicited through exploratory data analysis and experiments bydevelopers and data scientists. Even though Apptus has nodirect relationship with the consumers they have access toconsumer data for their customers’ web shops.

Apptus has around 40 employees of which around 20 aresoftware engineers split over five cross-functional teams. Rolesand their involvement in the experiment process are similarto those defined in the RIGHT model [3]. Product ownersdefine experiments and set goals. The experiments and theconnected metrics are designed by data scientists, while soft-ware developers and quality assurance implement and verifythe functionality. This functionality is put into productionby the DevOps and release engineers, that also execute andmonitor the experiments. However, the case company does nothave personnel with a dedicated quality assurance or releaseengineer role, instead they share these roles among themselves.

We used snowball sampling for selecting interviewees andinterviewed five people, namely two data scientists, one soft-ware developer, one DevOps engineer, and one product owner.The interviews were conducted in Swedish for about one houreach during in Spring and Summer of 2017. The first authortranscribed the audio recorded interviews word-for-word (40pages of text) and a summary was sent to each intervieweefor validation. The transcripts were analysed using thematiccoding in two steps, first using explorative descriptive codes(by first author) that were then grouped into themes (by bothauthors).

III. RESULTS: EXPERIMENTATION SCENARIOS

We identified four main scenarios of experimentation,namely (a) manual and (b) automated optimisation, and (c)external and (d) internal validation of a new feature or product,see Fig. 1. The scenarios were gathered from the interviewmaterial regarding experiments conducted and the tools devel-oped at the case company. We now outline the aim of and rolesinvolved in each scenario. For each scenario, we also presentsteps that the case company have taken to support automationfor these, with the exception of external validation (scenarioc) where no automation was mentioned by our interviewees.The differences between the scenarios was commented on byan interviewed data scientist: “Optimisation is done becauseyou have a lot of different components that work in differentsettings, and you do not have the time and ability to do thatfor all sites. [. . . ] Use A/B tests when you have a hypothesis

Case company Customer

Marketer

Consumers

A B A B

(a) Manual optimisation. Marketer sets up an experiment to optimisethe web shop using functionality in the case company’s product. Noinvolvement from the case company.

A

Consumers

B

CustomersCustomersCustomersCase company

NE C D A B

(b) Automated optimisation. The software product is automaticallyoptimised for different customers. No human involvement is necessaryonce the software product is deployed.

Competitor

ACase company

DevOps

Customer

IT

Consumers

N A B

(c) External validation. A customer’s IT engineer initiates an experimentto validate the case company’s offering against another offer. Thesoftware of the web shop is changed. The case company’s DevOps areinvolved as observers.

CustomersCustomersCustomers

IT

ConsumersCase company

DeveloperNA B

(d) Internal validation. A Developer from the case company initiatean experiment to validate a changed feature using several customers’web shops. The case company’s software product changes. The affectedcustomers’ IT departments monitor the experiment.

Figure 1: Experimentation scenarios. The big rectangles showthe companies involved, arrows indicate the initiator of theexperiments, the small boxes represent software components,with multiples of them—marked A and B—indicate variationsof the software and a grey box with N indicate no change.

and you want to know something about the world. That iscompletely different. Both are really important.”

A. Manual optimisation scenarioWhen manually optimising a web shop for, e.g. increased

product sales, this is performed by marketers responsible forthe web shop, see Fig. 1a. In this scenario, we focus onthe experimentation that uses the case company’s softwareproduct, but experimentation was said to be used elsewheretoo, such as marketing experiments involving variations ofe-mails in campaigns to users. A marketer holds a crucialrole in e-commerce. They organise the day to day busi-ness of a web shop and perform activities such as creatingsales campaigns that push certain consumer products, performsocial media advertising, write blog posts, or also performoptimisation experiments on the web store. As explainedby the interviewees, marketers are not software developers

���

Page 10: PROFESSOR PER RUNESON, DATAVETENSKAP, LTH … · 2019. 10. 15. · each during in Spring and Summer of 2017. The first author transcribed the audio recorded interviews word-for-word