1) The document discusses challenges faced in implementing Scrum for a public sector project at GRNET to develop a cloud computing platform.
2) Key constraints included a waterfall-oriented contracting process and inexperience with Scrum by all parties. Corrective measures like training and splitting stories helped address estimation issues.
3) Risks discussed include potential burnout from ceremonies and continuous effort, ensuring self-organization by the team, and whether the project can finish without compromises given engagement levels and estimation challenges faced early on.
Call Girls in Gomti Nagar - 7388211116 - With room Service
Swimming against the waterfall@GRNET
1. 12 Nov 2014
Swimming against the waterfall
@GRNET
Running Scrum in a conservative, multi-constrained setting
Challenges & Risks from the PO perspective
Byron Georgantopoulos, GRNET, e-Infrastructures
byron@grnet.gr, linkedin.com/in/ibyron, @digibyron
11th Agile Meetup, Athens
2. 12 Nov 2014
Outline
• Introduction
• Constraints at the starting line
• Embed Scrum into RFP and subsequent contract
• Running Scrum project
• Evaluation
• Challenges and Risks
2
3. 12 Nov 2014
The Company
GRNET SA (Greek Research and
Technology Network)
Founded in 1998
Provision of network and computing services
for Greek Academic & Research Community
National and European R&D Projects
GRNET Network Management
3
~okeanos Public IaaS cloud
Started in 2011
Virtualized Compute, Network, Storage resources
Build on top of proven OSS (e.g. Google Ganeti)
OpenStack API-compatible
currently >7000 active VMs
okeanos.grnet.gr
4. 12 Nov 2014
The "eScience" Project:
IaaS => PaaS => AaaS
The first project in public sector to be implemented using Scrum
Main pillars:
(1) Hadoop operations over ~okeanos cloud
(2) Virtual Research Environment
(3) Reproducible Research
cloud-enabled data-intensive science for
A&R community
4
5. 12 Nov 2014
Why are we here today?
Discuss Scrum application and lessons learnt, under the following
constraints:
• The contracting authority is a public body
• The contractor is a private sector supplier
• The original RFP and the ΕΣΠΑ framework are waterfall-oriented
• It is the first Scrum experience for everyone involved (almost)
5
7. 12 Nov 2014
Why Scrum?
• Avoid common public project pitfalls:
• Transparency and control - ensure early & sustained visibility
• Business risk minimization - able to modify the PB
• Test the Scrum waters
How:
• Short iterations
• Potentially shippable increment at the end of every sprint
• Prioritized PB items
• Avoid upfront design, flexibiity to change scope
• Frequent interaction and feedback
7
8. 12 Nov 2014
RFP preparation
Very detailed initial specifications (“The system shall…”)
Flat structure of specifications - not hierarchally organized
Scrum explicitly stated as the implementation framework
Scrum as a factor for scoring candidates (10%)
8
9. 12 Nov 2014
Contract
• Scrum explicitly stated on contract
• Mid-to-Large duration (14m), although reduced from original
• Reporting on a monthly basis
• Payments based on reports (checkpoints) & features tested
• Fully estimated Product Backlog plus Definition-of-Done
defined at the end of Pre-Game sprint
• Not finished SBIs re-inserted into next sprint
• Grooming to update and refine Product Backlog
9
10. 12 Nov 2014
Initial Challenges
• Winning proposal close to original RFP
• The team has unknown technological and agile skills
• Expected defensive attitude towards the unknown new
framework
• Open source everything enforces full transparency, may
conflict with isolated dev environments
regarding the Team
10
11. 12 Nov 2014
The Dev Team
• Private sector and a university research lab
• Partially co-located
• Limited familiarity with technology, agile, co-development
• 3 persons at the beginning, additional developers in
following sprints
11
12. 12 Nov 2014
Scrum Master
• Also the contractor’s PM
• Previously participated in Scrum-flavored projects
• Unclear boundaries when wearing both (PM & SM) hats
12
13. 12 Nov 2014
Product Owner
• Learned his lessons from past waterfall projects
• Received feedback from the company's partnership in
other Scrum projects
• Surprised when realised level of engagement (how close &
how often needed to work with the team)
13
14. 12 Nov 2014
Scrum adoption challenges
• Under-estimation
• Expect the managers to give orders
• 99% "Done" (activity vs. result-based)
• Silos of code
• Back-door waterfall attempts (e.g. request fully-fledged
design)
14
15. 12 Nov 2014
Glad :)
• Homogeneous team that 'gels' and works well together
• Minimal interpersonal issues
• Close collaboration with PO (full collab 1d/week, 40-50%
time devoted to the project so far)
• All ceremonies conducted and timeboxed
• Acceptance of Scrum, resistance less than expected
15
16. 12 Nov 2014
Glad wrt. “Why Scrum” choices
What the Project has gained from Scrum:
• Demonstratable software
• Deployable software
• External stakeholders involvement
• Continuous feedback
• Avoid wasted work and upfront design
• Scope changes allowed
16
17. 12 Nov 2014
Sad :(
• Long-lasting tasks (frequently exceeding 2 days)
• 1st story always finishes late >> not smooth burndown
• Unfinished sprint stories >> unpredictable velocity
• Building technology skills vs. business value: 1-0
• “Working software” mentality lagging
• Increasing technical debt (coding standards, test coverage)
• Context switch & not full-time dedication
17
18. 12 Nov 2014
Mad ~:(
• Delays and impediments surface towards the end of sprint,
ignoring the 'elephant in the room'
• ‘Hero’ attitude: work overtime and finish everything at the end
• Status meetings masked into daily scrums
• Definition-of-Done not followed
• More transparency needed (frequent commits, meaningful
comments)
18
19. 12 Nov 2014
Sprint retrospectives
• Have emerged as a major tool for inspecting & adapting
• Gradually people express their views more openly
• So far focused on processes and tools (not people or
relationships)
• Scrum Master and Product Owner still in 'protected' zone
• Need fresh ideas on how are executed and how to fully
engage all team members
19
20. 12 Nov 2014
Corrective measures taken
• Scrum training >> better understanding of process and estimation
• PO close to the team >> direct feedback, tech guidance, team spirit
• Split stories >> better estimates
• Retrospective outcomes embedded into next sprint
• Improvise process when needed, outside the Scrum textbook: (e.g.
live daily scrum at the end of collab day instead of teleconf)
• Pull back, empower dev members >> transfer decision-making to
the team in order to promote self-management
20
21. 12 Nov 2014
Risks and hard questions
• Burn-out:
Estimation & ceremonies
Continuous effort, without breaks
• Self-organizing, managing and owning
• What if the team will react to failures by overestimating
required effort (generally: abusing the rules)
• Engagement level of PO may discourage Scrum adoption
• Will it finish without compromises?
21
22. 12 Nov 2014
Towards Scrum-friendlier RFPs
• Stricter requirements for Scrum team composition:
• Scrum experience / certification
• Seniority
• Sustained effort
• Higher-level, more business-value oriented specifications
• More types of delivery checkpoints
22
23. 12 Nov 2014
The Scrum Coach
• Agile Meetup community >> knowledge exchange forum
• Met coach within Meetups
• A great boost for PO and team: Scrum experience and
guidance, 'external observer', servant leadership
• A public ‘thank you’
23
24. 12 Nov 2014
Conclusion: Scrum increases business value in public
sector projects, certainly worthy to promote and expand
Scrum will pass the Turing test:
• If the Dev team / SM choose to run their next project using Scrum
• If critical Scrum adoption know-how is created and transferred to new
teams inside GRNET and the broader public sector
The Turing Test
24