33
AppSec USA 2014 Denver, Colorado AppSec Survey 2.0: Fine-Tuning an AppSec Training Program Based on Data John B. Dickson, CISSP @johnbdickson September 18, 2014

AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Embed Size (px)

DESCRIPTION

Measuring the effectiveness of any security activity is widely discussed – security leaders debate the topic with a religious fervor rivaling that of any other hot button issue. Virtually every organization has some sort of application security training effort, but data on training effectiveness remains scarce. Last year our research team delivered the first-ever survey that captured developer awareness of secure coding concepts and the impact of formal application security training on a developer’s ability to write secure code. We learned that most software developer were aware of certain application security concepts, yet when asked how to write more secure code, they faired poorly. This year’s 600-developer survey provides more quantitative data on what software developers understand about application security, both concepts and practices. It dives most deeply into awareness of defensive coding practices, which most developers largely did not grasp in the 2013 survey. It also is separates respondents by roles, so we can better understand how architects, developers, and QA staff grasp key application security concepts and put them to work. It better captures how software developers learn in general, so one can tailor any security training effort to how software developers, in practice, actually learn. This information will provide data to application security managers responsible for corporate security training that should allow them them to make more fact-based decisions about security training.

Citation preview

Page 1: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

AppSec USA 2014 Denver, Colorado

AppSec Survey 2.0: Fine-Tuning an AppSec Training Program Based on

Data

John B. Dickson, CISSP @johnbdickson

September 18, 2014

Page 2: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

John  B.  Dickson,  CISSP  

•  Application Security Enthusiast •  Ex-AF Guy & ISSA Distinguished Fellow

•  Serial Entrepreneur & MBA Type

•  Dad

Introduction

Page 3: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

When Not Thinking about AppSec…

I am Snake Hunting on a Ranch in South Texas  

Page 4: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Snake Hunting Essentials

Cooler  Hat   Cool  Hat  

Snake  Guards   Common  Gardening  Tools   Machete  

Guy  who  has  a  machete  and  who  is  actually  good  at  “catching”  snakes  

OWASP  AppSec  2011  t-­‐shirt  

© Copyright 2014 Denim Group - All Rights Reserved

Page 5: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

•  Background  •  Premise  •  AppSec  Study  1.0  Results  –  What  We  Learned  •  Approach  and  Survey  ParKcipants  •  Key  Results  •  What  We  Can  Put  To  Work  •  Conclusions  and  QuesKons  &  Answers  

Overview

Page 6: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

•  Things  we  Knew  Last  Year  

•  Key  Findings  of  Last  Year’s  Study  

•  AddiKonal  Stuff  We  Learned  Along  the  Way  

•  Development  training  is  hard  

•  Results  are  rarely  measured  for  ROI  

•  Training  is  typically  part  of  any  AppSec  program  

AppSec Study 1.0 Results

Page 7: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

•  Things  we  Knew  Last  Year  

•  Key  Findings  of  Last  Year’s  Study  

•  AddiKonal  Stuff  We  Learned  Long  the  Way  

•  25%  retenKon  aXer  training  

•  QA  did  worse  than  architects  and  soXware  developers  

•  Respondents  answered  basic  awareness  quesKons  but  not  coding  pracKces  

AppSec Study 1.0 Results

Page 8: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

•  Things  we  Knew  Last  Year  

•  Key  Findings  of  Last  Year’s  Study  

•  AddiConal  Stuff  We  Learned  Long  the  Way  

•  SoXware  developers  learn  differently  than  companies  teach  

•  IncenKves  ma[er  •  Surveys  are  hard!    

AppSec Study 1.0 Results

Page 9: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Overview of 2014 “2.0” Study

•  600  respondents  •  Represents  mulKple  industries  •  Asked  the  same  applicaKon  security  quesKons  as  

2013  survey  •  Expanded  to  include  training  method  quesKons    •  No  “before”  and  “aXer”  analysis  •  No  classroom  training  opportuniKes  •  Used  more  social  media  •  Data  collecKon  ongoing  

Page 10: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Approach and Survey Participants

Sample  QuesCons      

QuesKons  that  tested  basic  knowledge  of  applicaKon  security:    

•  ApplicaKon  security  is  best  defined  as…  •  Threat  Modeling  is…  •  Input  ValidaKon  is…  

Page 11: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Approach and Survey Participants

Sample  QuesCons      

QuesKons  that  tested  understanding  of  defensive  coding:    

•  Marking  a  cookie  as  “secure”  will…  •  Which  of  the  following  will  help  protect  against  XSS…  

•  Which  of  the  following  is  NOT  an  example  of  good  session  policy…  

Page 12: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Delivery  Means    •  Direct  Delivery  of  Customized  Links  via  E-­‐mail  

•  Survey  Monkey  paid  •  Social  Media  

– Facebook  – Linkedin  

Targets  •  SoXware  Developers  •  Architects  •  Quality  Assurance  

Approach and Survey Participants

Page 13: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Demographic Questions Asked

•  What  is  your  primary  job  funcKon?    

•  What  is  your  company's  size?    

•  How  many  years  of  soXware  development  experience  do  you  have?    

•  How  much  previous  applicaKon  security  training  have  you  received?  

 

Page 14: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

2014 Study Demographics

Less  than  a  Year  18%  

1-­‐2  Years  9%  

2-­‐4  Years  10%  

4-­‐7  Years  13%  

7-­‐12  Years  16%  

More  than  12  Years  34%  

How  many  years  of  soMware  development  experience  do  you  

have?  

Page 15: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

2014 Study Demographics

Other  35%  

SoXware  Developer  

53%  

Quality  Assurance  

6%  

Architect  6%  

What  is  your  primary  job  funcCon?  

Page 16: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

2014 Study Demographics

8%  8%  

29%  

8%  10%  

37%  

What  is  your  company  size?  

1-­‐24  Employees  

25-­‐99  Employees  

100-­‐499  Employees  

500-­‐2499  Employees  

2500-­‐9999  Employees  

10,000  or  more  Employees  

Page 17: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

2014 Study Demographics

None  31%  

Less  than  a  Day  19%  

At  least  1  day,  but  less  than  2  days  

17%  

At  least  2  days,  but  less  than  3  days  

8%  

More  than  3  days  25%  

How  much  previous  applicaCon  security  training  experience  have  

you  received?  

Page 18: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Key Survey Results

•  Data  shows  soXware  developers  posiKvely  answer  quesKons  about  applicaKon  security  56%  of  the  Kme  

•  2013  Denim  Group  study  results:          58%  •  2014  Aspect  Study:          60%  

Page 19: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Change Implementation

Yes    33%  

No    25%  

I  don't  know  42%  

Did  your  organizaCon  implement  any  SDLC  or  process  improvement  steps  to  formalize  concepts  learned  in  training?  

Page 20: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Types of Training Received

0   50   100   150   200   250  

Instructor-­‐Led  PresentaKons  

e-­‐Learning,  CBT  

Social  Media  

Social  Learning  Plaqorms    

Developer  E-­‐mail  Lists  or  RSS  feeds  

Crowdsourcing  Sites  

Websites  

Webinars  or  Videos  

1-­‐on-­‐1  Coaching  

Wri[en  Materials  

Other  

Types  of  Training  Received  

Page 21: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

E-Learning & Instructor-Led Training

0   100   200   300  

Instructor-­‐Led  PresentaKons  

e-­‐Learning,  CBT  

Social  Media  

Social  Learning  Plaqorms    

Developer  E-­‐mail  Lists  or  RSS  feeds  

Crowdsourcing  Sites  

Websites  

Webinars  or  Videos  

1-­‐on-­‐1  Coaching  

Wri[en  Materials  

Other  

Types  of  Training  Received  

E-­‐Learning  &  Instructor-­‐led  Training  are  SKll  the  Primary  ApplicaKon  Security  Training  Approach  

Page 22: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Perceived Effectiveness of Training

0   50   100   150   200   250   300   350   400   450   500  

Instructor-­‐Led  PresentaKons  

e-­‐Learning,  CBT  

Social  Media  

Social  Learning  Plaqorms    

Developer  E-­‐mail  Lists  or  RSS  feeds  

Crowdsourcing  Sites  

Websites  

Webinars  or  Videos  

1-­‐on-­‐1  Coaching  

Wri[en  Materials  

1:  Not  EffecKve  

2:  Somewhat  EffecKve  

3:  Very  EffecKve  

Page 23: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Question Types

41%  

59%  

0%   10%   20%   30%   40%   50%   60%   70%  

PrescripKve  QuesKons  

Awareness  QuesKons  

%  of  QuesKons  Answered  Correctly  

Respondents  Fared  Far  Worse  on  QuesKons  Involving  Secure  Coding  PracKces  versus  ApplicaKon  Security  Awareness  QuesKons    

Page 24: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Pass Rate by Job Function

Average  Pass  Rate  

0%  

5%  

10%  

15%  

20%  

25%  

Other   SoXware  Developer   Quality  Assurance   Architect  

70%  or  more  quesKons  answered  correctly  

Quality  Assurance  respondents  Fared  50%  worse  than    soXware  developers  and  architects  

Page 25: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Pass Rate by Previous Training

Average  Pass  Rate  

0%  

5%  

10%  

15%  

20%  

25%  

30%  

Less  than  a  Day  or  None   At  least  1  day,  but  less  than  3  days   More  than  3  days  

70%  or  more  correct  

The  Pass  Rate  More  Than  Doubled  for  Respondents  Who  Had  More  Than  Three  Days  ApplicaKon  Security  Training    

Page 26: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Pass Rate by Job Function: Security

Average  Pass  Rate  

0%  

10%  

20%  

30%  

40%  

50%  

60%  

70%  

80%  

90%  

Security-­‐Related   Everyone  Else  

70%  or  more  quesKons  answered  correctly  

Respondents  that  worked  for  security  organizaKons  or  vendors  DID  fare  well  compared  to  other  respondents  

Page 27: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

What we Can Put to Work

•  Refresher  training  is  criCcal  •  Even  with  3+  days  of  appsec  training,  most  

respondents  did  not  have  a  “passing”  grade  of  70%  

•  Like  any  other  training  topic,  leX  unreinforced,  what  learned  will  be  forgo[en  over  Kme  •  ParKcularly  given  the  lack  of  SDLC  changes  

•  Likely  an  area  for  addiKonal  study  for  2015  appsec  training  study  

Page 28: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

What we Can Put to Work

•  Training  without  SDLC  changes  likely  will  produce  the  same  results  •  33%  of  the  respondents  said  their  organizaKon  

implemented  some  security  SDLC  improvements  •  67%  either  answered  “no”  or  “don’t  know”  •  OrganizaKons  cannot  rely  exclusively  on  

developers  retenKon  and  iniKaKve  to  produce  long-­‐term  decline  in  applicaKon  vulnerabiliKes  

Page 29: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

What we Can Put to Work

•  Augment  QA  with  Focused  AppSec  Training  •  QA  has  consistently  responded  poorly  relaKve  to  

developers  and  architects  •  Many  organizaKon  put  their  most  junior  

developers  in  QA  to  start  •  QA  is  where  appsec  “lives”  in  many  

organizaKons  •  OrganizaKons  might  considering  “doubling  

down”  on  appsec  training  for  QA  staff  to  compensate  for  this  fact  

 

Page 30: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

What we Can Put to Work

•  IncenCves  Ma`er  When  Working  with  Developers    

•  We  used  incenKves  throughout  the  study  to  collect  responses  -­‐  #Success!  

•  SoXware  developers  have  infinite  reasons  to  ignore  engagement  by  the  AppSec  team  

•  Rewards  help  nudge  soXware  developers    

Page 31: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

What we Can Put to Work

•  Training  programs  must  be  tailored  to  be  effecCve  

•  Formal  programs  like  classroom  training  and  e-­‐Learning  are  sKll  the  bread  and  bu[er  of  appsec  training  programs  

•  ConsumpKon  rates  of  e-­‐Learning  sKll  abysmal  without  incenKves  or  internal  markeKng  

•  Add  newer  ways  of  learning  to  reinforce  certain  key  points  and  to  serve  AppSec  corner  cases  

•  Leverage  current  events  to  reinforce  other  key  points  

Page 32: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

Conclusions

•  Data  shows  soXware  developers  posiKvely  answer  quesKons  about  applicaKon  security  56%  of  the  Kme  

•  Data-­‐driven  applicaKon  security  programs  will  likely  be  more  successful  and  chart  improvement  

•  SophisKcated  security  managers  use  incenKves  and  tailor  programs  to  improve  appsec  IQ  

Page 33: AppSec Survey 2.0 Fine-Tuning an AppSec Training Program Based on Data

White  Paper?  MenCon  it  on  Twi`er        

John  B.  Dickson,  CISSP  @johnbdickson  #appsecstudy  

Questions and Answers