wiki:Burndown

Introduction

After a somewhat hesitant start, we record the amount of work completed in each sprint. Each task is given an estimate - a number that measures its (estimated) size relative to other tasks. (It is emphasised that these numbers - "estimate points" - are relative numbers, they are not an estimate of hours or days.) The figure logged for the sprint is the sum of the estimate points for the completed tasks

By logging the amount of work completed (as measured by estimate points) in each sprint, we get some idea of the team's "velocity" - the number of estimate points the team can get through in each sprint. Owing to team meetings, some sprints are shorter than ten working days: to to allow for this the velocity is calculated in "estimate points per day". Also, as the team size can be different in different sprints (i.e. the average team size is smaller during holiday periods), a smoothed value taken over the last three sprints for which data are available is calculated, and this is the figure used for planning.

Notes:

  • A task is complete only when the following have been completed:
    • When the work - including documentation - is done.
    • A review has been carried out and the reviewer has agreed that the work is completed.
    • Code has been merged back into the master branch.
    • The task has been closed in Trac.
  • A task estimate can be modified by the person doing the work if they decide that the original estimate was incorrect. (Doing this should increase the accuracy of historical information and so improve the estimate of velocity.) However, this should be a rare event: if a task has been significantly under-estimated, the task should be split into sub-tasks.

A burn down chart is a way of displaying progress in a project and shows the remaining work in the sprint backlog. At present, no such chart is maintained, instead we rely on the list of remaining tasks in Trac.

Year 3/4 Sprints

Sprint Duration
(working days)
Tasks
Closed
Points Points/Day Average
Points/Day
(Last 3 Sprints)
Notes
2011-04-05 7 3 8 Some of the team were attending IETF-80, so
this sprint is not included in calculations of team velocity.
2011-04-19 10 20 79 7.9
2011-05-03 10 14 79 7.9
2011-05-17 10 26 66 6.6 7.5 Some of the team were at RIPE-62
2011-05-31 10 19 78 7.8 7.4
2011-06-14 10 17 60 6.0 6.8
2011-06-28 10 33 77 7.7 7.2 Some of team unavailable for some of the time
2011-07-12 10 26 81 8.1 7.3
2011-08-02 10 16 56 5.6 7.1 15-day sprint. However, as a number of the team were on holiday or
at IETF-81 for a week or so, for the purposes of statistics the length
of the sprint has been taken as 10 working days.
2011-08-16 10 17 85 8.5 7.4
2011-08-30 10 12 43 4.3 6.1 15-day sprint. However, as this encompassed a one-week face-to-face
meeting, for the purposes of statistics the length of the sprint has been
taken as 10 working days.
2011-09-27 10 7 26 2.6 5.1 15-day sprint. However, as this encompassed the one-week ISC "all-
hands" meeting, for the purposes of statistics the length of the sprint
has been taken as 10 working days. (in retrospect, as the numbers also reflect, it turned out to be significantly less than that)
2011-10-11 10 27 86 8.6 5.2
2011-10-25 10 15 45 4.5 5.2 See SprintPlanning20111025 for some potential explanations of why we are low this time.
2011-11-08 10 17 86 8.6 7.2
2011-11-22 10 27 93 9.3 7.5
2011-12-06 10 12 37 3.7 7.2 A number of holidays, also 11 tickets in review at the end of the sprint
2011-12-20 10 23 62 6.2 6.4
2012-01-10 15 13 37 2.4 4.1 Lots of people gone around Christmas and new year's (we should really be counting available developer days instead of calendar days btw)
2012-01-24 5 13 27 5.4 4.7 Short sprint after F2F
2012-02-07 10 14 54 5.4 4.4
2012-02-21 10 18 73 7.3 6.0
2012-03-06 10 21 85 8.5 7.0
2012-03-20 10 30 90 9.0 8.3
2012-04-03 10 25 79 7.9 8.5
2012-04-17 10 21 65 6.5 7.8 2 holidays during this sprint, also Stephen has other priorities right now. (repeat comment about available developer days)
2012-05-01 5 8 26 5.2 6.5 (partial sprint due to F@F)
2012-05-15 10 29 97 9.7 7.1
2012-05-29 10 11 50 5.0 6.6 (consensus seems to be that holidays and underestimation led to a lower productivity in terms of estimated points here)
2012-06-12 10 21 86 8.6 7.7 (technically we did more work; 5 tickets were added that were not estimated, 4 of which have been done)
2012-06-19 5 10 47 9.4 7.3 This was a "hardening" sprint
2012-07-03 10 9 37 3.7 6.8 Less points planned; with Jelte on vacation and Jinmei moving to resolver research
2012-07-17 10 10 39 3.9 5.7 Jinmei also on resolver research; team members away 3 working days
2012-07-31 10 17 59 5.9 4.5
2012-08-07 5 12 43 8.6 6.3
2012-08-21 11 6 21 2.1 5.5 a LOT of tickets in review (59 points worth)
2012-09-04 9 17 77 8.6 6.4
2012-09-18 10 8 35 3.5 4.7 Two devs on vacation (one part, one full), some tickets took way longer than anticipated (#2108 and #2109)
2012-09-25 5 4 20 4.0
2012-10-09 10 21 79 7.9 5.1
2012-10-23 10 16 73 7.3 6.4
2012-11-06 10 13 50 5.0 6.7 One ticket got added without estimate (a sensitive ticket), also there was an alpha release
2012-12-04 10 10 32 3.2 5.2 Number of tickets *almost* done, also a few tickets with long turnaround time (still waiting) and a big bottleneck ticket, with a lot of dependants
2012-12-18 10 13 55 5.5 4.6 A few tickets were added and hence had no estimate
2013-01-08 15 14 61 4.0 4.3 3 weeks due to holidays & PTO
2013-01-22 10 18 58 5.8 5.1
2013-02-05 10 21 70 7.0 5.6
2013-02-19 10 17 57 5.7 6.2
2013-03-05 10 14 50.5 5.1 5.9 16 points of research done but tickets still open
2013-03-19 10 13 58.5 5.9 5.6 16 points of research done but tickets still open
2013-04-02 10 10 35 3.5 4.8 Many developers had time off, a developer left, etc.
2013-04-29 15 11 45 3.0 4.1 F2F meeting, plus we took an extra unplanned week into the sprint
2013-05-14 10 7 30 3.0 3.2
2013-05-28 10 11 43 4.3 3.4
2013-06-11 10 7 30 3.0 3.4
2013-06-25 10 10 42 4.2 3.8
2013-07-09 10 10 45 4.5 3.9
2013-07-23 10 4 4 0.4 3.0
2013-08-06 10 13 38 3.8 2.9
2013-08-20 10 10 21 2.1 2.1
2013-09-03 10 3 10 1.0 2.3
2013-09-17 10 4 12 1.2 1.4
2013-10-01 10 5 10 1.0 1.0


Year 2 Sprints

Year 2 of the project was the first year in which Agile was used for development. As the work naturally divided into two areas - production of the authoritative server and production of the resolver - the developers were divided into two teams: the A-Team (authoritative server and miscellaneous) and the R-Team (Resolver). A discrete part of the A-Team's work was production of statistics code, so although the A-Team was managed as a single entity, the sprint progress for the statistics sub-team was tracked separately.

Regarding the estimates: the A-Team and R-Team estimate points are almost, but not completely equivalent. The estimate is a relative measure of task size, not an absolute one: the estimates are made by different people (each of whom may have a different personal equation), and the teams used different baselines for the estimates. In addition, there were different numbers of people working full-time on the sprints, so each team got through a different amount of work in the same time. However they are close enough for generalised conclusions to be drawn.

The results of the sprints are shown below.

Combined Statistics for All Teams

Sprint Duration
(working days)
Tasks
Closed
Points Points/Day Average Points/Day
(Last 3 Sprints)
Notes
2010-11-02/03 10 Estimates were made in days, so not comparable.
2010-11-16/17 10 Estimates in days/record incomplete.
2010-11-30/12-01 10
2010-12-14/15 10 Contained the Christmas break
2010-12-28/29 8 Last two days were part of the BIND-10 meeting
2011-01-25/26 7 18 45.5 6.5 First three days were part of BIND-10 meeting
2011-02-08/09 10 13 47 4.7 CNNIC developers (50% of the team) were on holiday
from Jan 31
2011-02-22/23 10 21 62.8 6.3 5.8 CNNIC developers (50% of the team) were on holiday
until Feb 14
2011-03-08/09 10 19 67 6.7 5.9
2011-03-16/17 8 21 80.5 10.1 7.5

A-Team Sprints

Non-Statistics Work

Sprint Duration
(working days)
Tasks
Closed
Points Points/Day Average Points/Day
(Last 3 Sprints)
Notes
2010-11-03? Task record incomplete
2010-11-17? Estimates incomplete
2010-12-01? Estimates incomplete
2010-12-15? 10 53 5.3
2010-12-29 7 No record owing to Christmas/absence of team managers,
and last three days were part of the BIND-10 meeting
2011-01-26 8 9 26.5 3.3 First two days were part of BIND-10 meeting
2011-02-09 10 6 23 2.3 3.7 CNNIC developers (50% of the team) were on holiday
from Jan 31
2011-02-23 10 10 29.8 3.0 2.8 CNNIC developers (50% of the team) were on holiday
until Feb 14
2011-03-09 10 11 45 4.5 3.3
2011-03-16 7 8 21 3.0 3.5

Statistics

Sprint Duration
(working days)
Tasks
Closed
Points Points/Day Average Points/Day
(Last 3 Sprints)
Notes
2010-11-03? Task record incomplete
2010-11-17? Estimates incomplete
2010-12-01? Estimates incomplete
2010-12-15? 10 0 0
2010-12-29 7 No record owing to Christmas/absence of team managers,
and last three days were part of the BIND-10 meeting
2011-01-26 8 0 0 0 First two days were part of BIND-10 meeting
2011-02-09 10 1 5 0.5 0.2
2011-02-23 10 0 0 0 0.2
2011-03-09 10 0 0 0 0.2
2011-03-16 7 0 0 0 0

R-Team Sprints

Sprint Duration
(working days)
Tasks
Closed
Points Points/Day Average Points/Day
(Last 3 Sprints)
Notes
2010-11-02 10 Estimates were made in days, so not comparable.
2010-11-16 10 Estimates in days/record incomplete.
2010-11-30 10 25 2.5
2010-12-14 10 2 0.2 Contained the Christmas break
2010-12-28 8 14 1.8 1.5 Last two days were part of the BIND-10 meeting
2011-01-25 7 9 19 2.7 1.4 First three days were part of BIND-10 meeting
2011-02-08 10 6 19 1.9 2.1 CNNIC developers (50% of the team) were on holiday
from Jan 31
2011-02-22 10 11 33 3.3 2.6 CNNIC developers (50% of the team) were on holiday
until Feb 14
2011-03-08 10 8 22 2.2 2.5
2011-03-16 8 13 59.5 7.5 4.1
Last modified 4 years ago Last modified on Oct 1, 2013, 8:06:05 AM

Attachments (1)

Download all attachments as: .zip