+ - 0:00:00
Notes for current slide

Presenter notes contain extra information which might be useful if you intend to use these slides for teaching.

Press P again to switch presenter notes off

Press C to create a new window where the same presentation will be displayed. This window is linked to the main window. Changing slides on one will cause the slide to change on the other.

Useful when presenting.

Notes for next slide



Galaxy Administrator Time Burden and Technology Usage



last_modification Updated:   purlPURL: gxy.io/GTN:S00122

text-document Plain-text slides |

Tip: press P to view the presenter notes | arrow-keys Use arrow keys to move between slides
1 / 37

Presenter notes contain extra information which might be useful if you intend to use these slides for teaching.

Press P again to switch presenter notes off

Press C to create a new window where the same presentation will be displayed. This window is linked to the main window. Changing slides on one will cause the slide to change on the other.

Useful when presenting.

Context

2 / 37

Context

3 / 37

Overview of Participants' Galaxy Instances'

4 / 37

Active Users

Histogram of active users, showing a median of 20 and mean of 21, ranging from as few as 1-5 users and as many as 50

  • 3 categories : <10, circa 20 & circa 50
  • Active vs signed up: no problem having a lot more signed up than active users, no need to delete inactive users (in case they come back)
5 / 37

Computing Method

Pie chart. Computing Method. Showing 11% Pulsar, 33% Local, 44% Batch scheduler & 11% Other

  • Admins generally use a batch scheduler (often HTCondor)
  • Some admins use Pulsar to send jobs to remote machines
  • And a few use the local scheduler which will not survive a machine restart.
6 / 37

Object Store Backend

Pie chart. Object store backend. Showing 11% Local, 44% NAS & 44% S3

  • Local storage (SSD/HDDs attached to the machine) and NASs are popular options for data storage
  • S3 is a less common option amongst small scale admins
  • You can use whatever storage you have available
7 / 37

Ansible Usage

Pie chart. Ansible usage. Showing 78% positive.

Is a highly-used (about 80%) tool that, according to the other sections, greatly simplifies admin.

8 / 37

Gravity Usage

  • Since Galaxy 23.0, Gravity is used to manage Galaxy processes
  • If you use Ansible to setup Galaxy, this process is mostly transparent, and you may not know you're using Gravity
9 / 37

User Support & Training

10 / 37

End-User Support Burden

Histogram of the average monthly hours per month dedicated to end-user support. Showing a median of 5 and mean of 6. Large variability, ranging from 0 to 20.

  • 1 hour per week average end-user support
  • Lots of training (though that’s part of the next question)
  • Lots of developing &/or debugging users’ tools/workflows for them.
11 / 37

User Training Burden

Histogram of the average monthly hours per month dedicated to user training. Showing a median of 5 and mean of 6. Large variability, ranging from 0 to 10.

  • 1 hour per week average user training
  • Usually makes use of existing Galaxy tutorials, sometimes in-person
  • The tutorials aren’t always specific enough, requiring some custom Q&A
12 / 37

Tool & Workflow Dev & Maintenance

13 / 37

User Tool Dev & Planemo Usage

Pie chart. Do users develop tools themselves? Showing 33% positive. Pie chart. If users develop tools themselves, do they use Planemo? Showing, once filtered, 33% positive.)

  • Users developing their own tools is rare, and if they do, they rarely use Planemo.
  • Tool creation is actually mostly done through copy/pasting XMLs
  • Planemo shines when it comes to tool testing, linting and publishing
14 / 37

Admin Workflow Dev Burden

Histogram of the average monthly hours per month dedicated to developing workflows, by the admin. Showing a median of 4 and mean of 3. Large variability, ranging from 0 to 8.

  • The administrative burden of running Galaxy is highly variable
  • On average admins spend 4 hours a month managing Galaxy
15 / 37

Admin Tool Dev Planemo

Pie chart. Do admins use Planemo? Showing 56% positive.

  • Many small scale Galaxy admins will find themselves responsible for developing tools
  • Amongst tool-developing admins, 5/7 use Planemo
  • Planemo is a useful tool for tool testing, linting and publishing tools for your local Galaxy
16 / 37

Admin Tool Dev Creation Burden

Histogram of the average monthly hours per month dedicated to creating tools, by the admin. Showing a median of 4 and mean of 6. Centered around 5, with an outlier at 20.

Small (1 hour per week avg.) amount of tool dev per month.

17 / 37

Admin Tool Dev Updating Burden

Histogram of the average monthly hours per month dedicated to updating tools, by the admin. Showing a median of 1.5 and mean of 4. Mostly 1, several others lower than 5, and an outlier at 16.

  • Excluding the extreme values, about an hour a month
  • I wondered if tools became invalid because of non-backwards-compatible format changes, but that does not seem to be the case
18 / 37

Tool Storage & Packaging

19 / 37

Admin Tool Dev Storage

Histogram of amount of admins that use certain tool storage types (multiple-choice question). Showing 7 local filesystem users, 3 public Mercurial toolshed users, and 1 blank answer

  • Most tools are files on the same machine as Galaxy
  • Some tools are used from public toolsheds, none from private ones
  • Currently, admins write the tool (managed in git), test & lint it with Planemo, then publish it locally, sometimes versioned (ex: tool_v1.1)
20 / 37

Tool Dev Packaging Method

Histogram of amount of admins that use certain tool packaging methods (multiple-choice question). Showing 7 conda users, 2 container users, and 2 blank answers

Lots of Conda, a few containers.

21 / 37

Tool Dev Packaging Burden

Histogram of the average monthly hours per month dedicated to tool packaging. 100% answered 1.

  • Admins generally spend very little time managing conda environments in Galaxy.
  • Galaxy can automatically create and manage conda environments for every installed tool
22 / 37
23 / 37

DB Tech

Pie chart. Database Technology. Showing 100% postgresql

  • It is strongly recommended to use PostgreSQL in production
24 / 37

DB Schema Update Frequency

Histogram of the average number of months between upgrades of their Galaxy instance's database's schema. Showing a median of 12 and mean of 11. Large variability, ranging from 4 to 24.

  • Happens on average once a year.
  • If Galaxy is managed with Ansible, this happens automatically as part of updating Galaxy.
25 / 37

DB Schema Update Duration

Histogram of the average duration, in hours, a database schema migration takes. Showing a median of 1 and mean of 0.6. All answers were 0 or 1.

Is very quick and transparent (assuming Ansible usage)

26 / 37

DB Backup Frequency

Histogram of the average number of weeks between database back-ups. Showing a median of 1 and mean of 11. Nearly everyone answered 1, however there were 2 outliers at 12 and 52.

27 / 37

Galaxy Version & Upgrades

28 / 37

Current Version & Upgrade Frequency

Histogram of the current Galaxy version. Showing a median of 22 and mean of 22. 2 using 21, 3 using 22 & 4 using 23. Histogram of the average number of months between upgrades of their Galaxy instance's version. Showing a median of 12 and mean of 12. Large variability, ranging from 4 to 24.

  • Half the respondents are able to continue using a version that is 2-3 years old
  • This is not recommended as versions older than a year do not receive security updates
  • But most update about once a year
29 / 37

Galaxy Upgrades Duration (excluding DB schema migration duration)

Histogram of the average duration, in hours, an upgrade takes. Showing a median of 3 and mean of 4. Mostly under 3, but a few values around 7.

  • Avg 3 hours, big variance, which doesn’t seem proportional to the amount of users, but rather other factors
  • Some updates are very simple, especially if you use Ansible
  • Some non-backwards-compatible changes like uWSGI to Gunicorn take longer, but these changes are not a regular occurrence in Galaxy.
  • Some admins take longer because of custom plugins (that need to be updated) or test deployment environments that need to be maintained.
30 / 37

Miscellaneous: Restarts, Crashes & Other

31 / 37

Intentional Restarts Frequency

Histogram of the number of intentional Galaxy restarts per year. Showing a median of 12 and a mean of 21. Lots of variability between 4 and 20. An outlier at 100.

  • If you configure Galaxy to use a job scheduler (SLURM, HTCondor, etc), restarts will not interrupt existing jobs.
  • Generally admins restart about once a month, in order to apply system updates or reconfigure Galaxy or its tools
  • Some Galaxy / tools reconfiguration is possible without restarting
32 / 37

Crashes Frequency

Histogram of the number of crashes per year. Showing a median of 4 and a mean of 3. Lots of variability between 0 and 8.

  • Avg. once every 3-4 months
  • Causes: Storage space full, or access rights problems, or very rarely TUS (but a solution was found for this since the poll (see the full report))
33 / 37

Other Recurrent Tasks Duration

Histogram of the amount of hours spent per month, on other administrative tasks not yet covered. Showing a median of 4, a mean of 14. Nearly all around 2, but a few outliers at 10, 40 and 50.

  • 3h per month avg
  • Cleaning "paused" jobs
  • Adjusting user quotas & notifying users
  • Networking with other admins
  • Troubleshooting tools
  • Testing new Galaxy versions
  • Custom welcome-pages
34 / 37

Take Home

35 / 37

Total Burden, Non-Dev Admin Tasks

Table showing how the total amount of non-dev admin tasks, in hours per month, was calculated. By showing the initial frequency (eg weekly, monthly, annually), the hours taken, and then multiplying the hours by the ratio to bring it to the monthly equivalent. Results: end-user support : 5, user training: 5, DB migration: 0, Galaxy upgrade: .3, DB back-up: 0, intentional restart: 1, crashes: .6, other: 3, total: 15.

  • Per month: circa 15 hours
  • Per week: half a working day
  • Reasonable amount
  • Actually slightly less because there is some overlap in user training between the "End-user support" and "User training" questions
36 / 37

Thank You!

This material is the result of a collaborative work. Thanks to the Galaxy Training Network and all the contributors!

page logo

Tutorial Content is licensed under Creative Commons Attribution 4.0 International License.

37 / 37

Context

2 / 37
Paused

Help

Keyboard shortcuts

, , Pg Up, k Go to previous slide
, , Pg Dn, Space, j Go to next slide
Home Go to first slide
End Go to last slide
Number + Return Go to specific slide
b / m / f Toggle blackout / mirrored / fullscreen mode
c Clone slideshow
p Toggle presenter mode
t Restart the presentation timer
?, h Toggle this help
Esc Back to slideshow