Agile Adoption without Exec Level Support · 2017. 8. 4. · Support: An inside-out approach to...
Transcript of Agile Adoption without Exec Level Support · 2017. 8. 4. · Support: An inside-out approach to...
7/26/17
1
© 2017 Lean-Agile Partners Inc. All rights reserved.
Agile Adoption without Executive-Level Support:
An inside-out approach to team based Agile adoption
Nancy Van Schooenderwoert @vanschoo
1
Nancy Van Schooenderwoert n Originally an electronics and software designer n 15 years safety-critical embedded systems development experience n Since 2002: Agile coaching of teams and managers in regulated industries n Industries: Aerospace (Flight simulation), Medical Devices, Sonar
Weaponry, Scientific Instruments, Industrial Controls, Financial Services n BSCE (Computer Engineering) from Rochester Institute of Technology n Active in Agile New England & Agile Alliance n Speaker at conferences worldwide
2
7/26/17
2
Agenda
3
n What we achieved as a team n New things our data told us n Back to the “prequel” – team’s origins
n Interaction part + 4 rules you can use n Benefits we never expected
Opportunity to try for zero-defects code
n First customer – agricultural
n Based on new science, new CPU, new OS port, new NIR h/w sensor, new algorithm…
n Agile team delivered 1st field units in 6 months
n Surprise! New customer wants to use it for blood analysis - It’s now a medical device
Grain Monitor System (GMS)
4
7/26/17
3
First we did these things as a team n Reviewed the s/w design I had begun n Chose a coding standard, tool use practices n Decided that we wanted to know 4 things throughout
our work together n Quality of our code n Schedule performance n Labor used n Volume of work done
5
Labor Distribution, 1st year
6
0
5
10
15
20
25
30
35
40
Detaile
d_de
sign
Code_
CSU_test
CSC_integ
_test
Proce
ss_d
evel
s/w_m
anag
emen
t
Tool_s
uppo
rt
Techn
ical_l
ead
Admin_
supp
ort
CSCI_tes
t
Sys_in
tegrat
ion
Suppo
rt
Requir
emen
ts
Per
cent
of T
eam
Lab
or
Source: “Embedded Agile Project by the Numbers with Newbies”, available at http://www.leanagilepartners.com/publications.html
7/26/17
4
Labor Distribution, 2 Years
7
0
5
10
15
20
25
Detaile
d_de
sign
Code_
CSU_test
CSC_integ
_test
Proce
ss_d
evel
s/w_m
anag
emen
t
Tool_s
uppo
rt
Techn
ical_l
ead
Admin_
supp
ort
CSCI_tes
t
Sys_in
tegrat
ion
Suppo
rt
Requir
emen
ts
Per
cent
of T
eam
Lab
or
0
5
10
15
20
25
30
35
40
Detaile
d_de
sign
Code_
CSU_test
CSC_integ
_test
Proce
ss_d
evel
s/w_m
anag
emen
t
Tool_s
uppo
rt
Techn
ical_l
ead
Admin_
supp
ort
CSCI_tes
t
Sys_in
tegrat
ion
Suppo
rt
Requir
emen
tsP
erce
nt o
f Tea
m L
abor
0
5
10
15
20
25
30
35
Detailed_design
Code_CSU_test
CSC_integ_test
Process_devel
s/w_management
Tool_support
Technical_lead
Admin_support
CSCI_test
Sys_integration
Support
Requirements
Per
cent
of T
eam
Lab
or
Year 1 Year 2
First Iteration
Hardware evolving…
8
1 2 3 4
5 6 7
7/26/17
5
Results n 29,500 lines of tested, working realtime embedded code
delivered (in C, non-comment lines), GMS app n 5,000 lines Windows utility code (VB) n 2,000 lines Perl code (Calibration Table builder utility) + other
utility code n 51 bugs total for project, active list <2 bugs n 1.5% bug rate would have given 442 bugs n Stable software throughout development
9
Defects: 3 years Development
10
n Defect injection rate of 0.22 defects/Function point n Productivity 3x industry norm
Bugs - sorted by Severity level and time found
CriticalModerateCosmetic
Totals
51
13
20
18
S/W Releases
Year 1 Year 2 Year 3
“Production Ready”
Field trials worldwide
7/26/17
6
Quality level among the highest
11
Team Defects/Function Point
Follett Software* 0.0128 agile BMC Software* 0.048 agile GMS 0.22 agile Industry Best** 2.0 traditional Industry average** 4.5 traditional
* Computed from data reported in Cutter IT Journal, Vol. 9, No. 9 (Sept 2008), page 10 ** Capers Jones presentation for Boston SPIN, Oct., 2002
Audited for GAMP compliance
3 day audit of software practices n Done for pharmaceutical customer (use spectrometer
for blood and tissue analysis) n Auditor impressed by thoroughness and consistency of
our work n A few minor findings, easily fixed
n E.g. make internal status flags readable without using debug tool
GAMP = Good Automated Manufacturing Practice
12
7/26/17
7
Agenda
13
n What we achieved as a team n New things our data told us n Back to the “prequel” – team’s origins
n Interaction part + 4 rules you can use n Benefits we never expected
Code base and defects over time
Code base grows from zero for clean-start project. Expect same shape for open defects?
Time
Line
s of
Cod
e O
pen
Def
ects
2
29.5K
14
Open defects turned out to be flat. No correlation to code size!
7/26/17
8
Defect difficulty not inherent n Difficulty of troubleshooting a s/w defect
depends most on length of time it was hidden – not on the type of defect
n Suggests that the strongest way to eliminate defects is to detect them quickly
n Agile excels in doing exactly that!
15
Def
ect N
o.
Defect Inserted Defect FoundR
eqm
ts
Des
ign
Cod
ing
Inte
g. T
est
Pos
t Rel
.In
tern
al
Pos
t Rel
.E
xter
nal
16
17
19
18
21
20
22
23
24
25
26
27
28
Not a Bug (for us) - Partner's code has bug.30
32
29
33
34
36 Not a Bug (for us) - Diagnostic PC code has bug.
35
31 Not a Bug (for us) - Partner's code has bug.
Source: N. Van Schooenderwoert; “Embedded Agile Project by the Numbers with Newbies”; Agile 2006 Conference; Available at www.leanagilepartners.com/Publications
Defect Severity KeyCosmetic Defect - Could leave as is
Moderate Defect - Can work around, but must fix later
Critical Defect - Must fix immediately
Length of time from injection of defect till its symptoms were visible
Mobile spectrometer - results
16
050
100200
400800
BestUS Europe
Malpractice0
500
1000
1500
2000
2500
3000
Defects Delivered to Customer
Total function Points
Defect Rate Data from Capers JonesTraditional teams injected many bugs, then removed most.
Spectrometer team injected fewer & removed most immediately
Spectrometer team performed here
7/26/17
9
Agenda
17
n What we achieved as a team n New things our data told us n Back to the “prequel” – team’s origins
n Interaction part + 4 rules you can use n Benefits we never expected
n (story, and discussion)
18
7/26/17
10
Agenda
19
n What we achieved as a team n New things our data told us n Back to the “prequel” – team’s origins
n Interaction part + 4 rules you can use n Benefits we never expected
My guidelines n Know when to walk away
n “BATNA” – best alternative to a negotiated agreement n Every negotiation has 2 tracks: substance, relationship n Cannot trade substance for relationship & vice versa
20
7/26/17
11
My guidelines n Create a teamwork foundation
n Something the group values… n Could be a problem all want to address n Coding standard, learning together, etc. n For GMS team, learning more about embedded sys coding,
(and getting this project delivered!)
21
My guidelines n Action over debate
n Debate rarely settles anything n Instead do an experiment or try a change till next sprint
22
7/26/17
12
My guidelines n Use principled negotiation
n See “Getting to Yes” by Fisher, Ury, Patton n (role play on positional bargaining)
23
Agenda
24
n What we achieved as a team n New things our data told us n Back to the “prequel” – team’s origins
n Interaction part + 4 rules you can use n Benefits we never expected
7/26/17
13
Agility spread to hardware n Only the s/w team was using Agile practices, but… n Frequent s/w releases created many more
opportunities to improve h/w-s/w interaction n Result was h/w became more Agile “without trying” n Some examples (next slide)
25
S/W team’s services to other teams n Some measurements inconclusive due to voltages out of
range – so added s/w monitoring of h/w key areas
n Field problems that could not be isolated to one area (opto, sensor, electronics) could be investigated thru special s/w releases for troubleshooting
n Hand assembly of field units improved by downloadable collection of s/w drivers with command-line menu
26
7/26/17
14
Whose bug is it? n Hardware is presumed innocent n But…
n We gave EEs code to check h/w n We gave simple tests to exercise s/w n This saved other groups much time
n By end of GMS project, even EEs were suspecting h/w before s/w!
n S/W no longer “guilty till proven innocent”
27
28
Contact Info
Nancy Van Schooenderwoert LeanAgilePartners
@vanschoo
§ Lean-Agile coaching for software and hardware teams
§ Safety-critical, regulated industry coaching/ training is our specialty
§ Lean-Agile coaching for stakeholders and senior managers
Coaching and Training: