Sunsetting
Introductions
Bret Davidson
Associate Head, Digital Library Initiatives
Jason Ronallo
Head of Digital Library Initiatives
NCSU Libraries
Slides with speaker notes:
https://ronallo.com/presentations/sunsetting-dlf
Hi I'm Jason and me and my colleague Bret will be talking about sunsetting projects.
Sunsetting
- Challenges
- Case Studies
- Lessons Learned
We'll start talking about some of the challenges of this type of work, some illustrative case studies, and share some lessons learned along the way.
Definitions
(and Word Choice)
First I want to start off with some definitions. I personally find the funnest, most exciting way to start off any presentation is to go through some dictionary definitions. Really I want to talk a bit about why we use the term "sunsetting".
DECOMMISSIONING
We could talk about decommissioning. Decommissioning captures a lot of what we'll be talking about. It is a process. But it sounds sterile and formal. Are we docking a naval vessel into port and filling it with mothballs?
Shut Down
We could say we're talking about shutting things down, but that's just a technical piece. "Shut down" is about the mechanics involved but misses too much of the human and isn't as easy as flipping a switch.
Winding Down
Winding Down: Gives the impression that projects and services are just wind up toys and then just continue to walk under their own power until they wind down. This one minimizes the ongoing, often hidden, work that happens to maintain something.
Killing Off
We could say that we're talking about "killing off" projects, but that sounds too brutal and bloody.
Pulling the Plug
Pulling the Plug: Too traumatic. Such a difficult choice to make.
Death
But in some ways this is a talk about death. This is a talk about the death of projects and services. And we want to recognize that this can be uncomfortable and emotional work.
Socio-Technical
Above all this is about the social work involved as well as the technical aspects.
Sunset
We don't want to get too dark about it even in this season. So we prefer the term "sunset". Sunsets are beautiful while we know dark and night are coming. We recognize there is sometimes a need to use imprecise and softening language for difficult and emotional situations. I feel like we need a word like sunset to help us cope with this kind of work.
Sunsetting Technical Projects
We'll be talking about projects here, and for us that mostly will mean technology projects. Lots of web applications and other software we've developed or adopted. That's where we work, but I think it could apply equally well in other areas. Other services, instruction.
Project Lifecycle
- Project acquisition
- Initiation
- Launch
- Iterative development
- Stability
We proposed this talk because we'd like to talk with you all more about how sunsetting fits into the project lifecycle. We'll often talk about project acquisition and new project launches and iterative development. Successful projects. All the clearly positive stuff.
Failure
We're beginning to talk about failure in some venues like Fail4Lib at Code4Lib.
Sunsetting Isn't Failure
But sunsetting isn't a kind of failure. Sometimes what you sunset are some of your greatest successes.
Sunset
We think it is time to start talking about sunsetting more in its own right. The term is helpful because it covers not just shutting something down but also that time when a project is still shining a light.
Twilight
Sometimes there's a twilight period for a project.
Victims of Our Own Success
We've been victims of our own success with the number of successful projects we've developed. So it has been long past time for us to begin talking about sunsetting and working on it. So this is our first attempt at sharing some of what we've learned.
Why Sunset?
- Mental space
- Opportunity costs
- Portfolio balancing
- Outdated tech
- Not useful or used anymore
So what are some reasons why you might want to think more about sunsetting projects? There are lots of reasons, but one I want to mention now that's often overlooked. For software projects that seem like there's not a lot of activity to others, still often requires ongoing maintenance work. Even if maintenance is minimal each project still takes up some mental space.
Sunset to Clear Space for New Work
The main reason we've made such an effort in trying to sunset projects is so that we can clear space for new work. We don't want to get to the point where we're not moving forward because of a maintenance burden.
Lighten the Load
For folks to feel like they have the space to take on new initiatives and to have the time to think to generate new ideas, we need to lighten their load.
Portfolio Management and Balancing
- Periodically review every project
- By person or by functional area
- Ask the tough questions
- Honestly assess
OK. So how do we actually go about this in practice? What does the process look like? For us it starts with examining portfolios. [READ SLIDE] The idea is that you hold up each project and ask questions about its value.
Seeing Past Success
Sometimes it take seeing past your past successes when evaluating a project to see current realities.
Candidate for Sunsetting?
- What are the maintenance costs?
- Takes more time to fix than it is worth?
- Different calculation of value now than before?
- Still have required in-house expertise?
So these are some of the questions you can ask even of projects which have had some success.
Options
- Sunset
- Deep maintenance mode
- Downgrade
- Hand off
- Migrate
- Double down
Once we've evaluated the value we can start talking about the options. There are gradations of options to consider. We like to put all options on the table. Sunsetting is the most extreme possibility but having it available means that it can still help move the needle towards decreasing effort.
Default to Sunset?
Consider whether you can start this discussion with a default of sunsetting. This can help push to make sure the evaluation makes the strongest current case for not sunsetting.
Who Cares?
Who still cares about the project? And this now gets to the most difficult part. These are the people you'll need to work with for making a decision.
Who Sunsets?
OK, you've made the decision to sunset, who does the work? One reason to get stakeholders on board is that it takes both technical and non-technical folks to get the job done. This isn't just about flipping a technical switch.
Roles
In the same way it helps to set clear roles for a new project just kicking off, you'll want to set roles for your sunsetting team.
Someone Else's Job
One thing to note is that there are some different roles and it is not necessarily the person who created something who needs to have responsibility for sunsetting it. That can sometimes be someone else's job. This can have an immediate benefit of clearing more time for the current maintainer to move on to other work.
Timelines and Time Estimation
We know time estimation for creating new software is regularly off by 2x, we're finding that time estimation for sunsetting can be off by way more.
Moving Parts
One reason for this is that sunsetting often has more moving parts. When you initially create something you have fewer staff involved. Something new has zero users.
Digging
To sunset often means doing some digging. Institutional memory fades. Past documentation practices may not have been up to current standards.
Carving Out Time
It is never someones top priority to sunset a project, so you need to figure out how to prioritize the work. Again sometimes making it someone else's job or setting deadlines can help.
There are a slew of activites involved in sunsetting projects, some technical, some non-technical.One activty is what our associate director Steve Morris has called "pickling". This is the process of internal knowledge capture and records retention. This is one way that we use our intranet.
It is important to celebrate and memorialize past accomplishments. One kind of memorial is a tombstone. For example, this can often be the page where a redirect lands. These memorials help gain acceptance of sunsetting and with maintaining institutional memory about the value of a service.
Project Pages
Project pages are another way to memorialize projects by documenting for external audiences why a project existed, who worked on it, and why it was successful.
Misc Technical Tasks
- remove application code and files
- remove databases
- clean out file system
- retain only needed log files
- add redirects
- change sitemaps
- update public github repositories
There are of course a variety of technical tasks related to sunsetting as well. In many ways, deletes are harder than updates. Change sitemap URLs to HTTPS was the prompt for changing sitemaps. Update public repos about not using any longer.
Sunsetting
Now going to talk about a variety of projects that we've either sunset, handed off, or migrated in some way. All of these were successful projects for us and represent areas where people made their careers. It's sometimes easy to take these projects for granted and its a testament to their quality that they lived on this long. As Jason already said, one of the common challenges among all of these projects is getting past the emotional and professional attachment to these great projects.
(Were)Wolfwalk
WolfWalk is a photographic guide to the history of North Carolina State University optimized for mobile devices. NCSU Libraries' first iOS app, testbed for GIS and mobile. Lost in-house expertise in iOS development, have web version of the service, campus interest in taking over iOS work.
(Were)Wolfwalk
Example of sunsetting expensive parts of the project while preserving the core.
(Were)WolfWikis
It often takes some research to figure out who the current users actually are. Since starting the service lots of commercial wikis came along. Took a long time to sunset, especially due to the large amount of of communication with faculty.
User Migration Paths
- Alternatives
- Data exports
- Data transformations
- Hand holding
Before you communicate with your users you'll need to do some research. Development of migration plans--which ended up not being needed by users but were a safety net to avoid isssues. How do we still direct our users to services that will meet their needs? Example of Eric's work exploring alternatives, exporting data and hand-holding of users as we shut the service down.
Groupfinder
An application for helping students meet up in groups at the library. Project that was sunset due to changing conditions on the ground. Students now use social media or SMS to coordinate use of spaces. Example of value proposition changing.
Mobile Catalog
Time before responsive design, mobile catalog lifecycle has been a devolution away from stand alone app. This application was handed off to Library IT. Also closely related to other services managed by IT.
Red, White & Black
Static sites are one way of accomplishing this. RWB highlights African American history at NC State in combination with an annual tour led by NC State faculty.
Red, White & Black
Important historical content that doesn't change much, was a good candidate for a Jekyll static site. Example of Migration and reducing mental burden.
Student Leadership Initiative
Another example of mature, important project that is a good candidate for Jekyll. Video oral history site that shows how technology has changed, for example search is now possible client side.
Computer Simulation Archive
And another. Comment how now we start the discussion about new projects with static site before considering dynamic. Static sites don't necessarily mean loss of functionality.
Computer Availability
Migrated from expensive frameworkless code, making it easier to extend and maintain, and also greatly improved the core components of the system by including new reports for debugging patron machine logins.
Historical State Search
Historical State Search exposes special collections materials related to the history of NC State. This project was recently migrated to use our open source "quicksearch" platform. Allows us to utilize a core platform for a second service and demonstrates an additional use case of quick search for the community. A combination of doubling down and migrating that reduces mental burden and lowers opportunity costs. See our colleague Nushrat Khan's talk tomorrow at 3:00 pm in the Discovery session for more info.
Lentil
Highly successful project that played a big role in eary Hunt publicity, jumpstarted part of our social media archiving program, and provided a test bed for large scale displays in Hunt. Maintain several instances for different activities, including the NC Lit Fest and Makerspace. Sunset the lit fest instance and met with stakeholders to scale back the Makerspace instance into "display only" content.
Instgram API troubles. In many ways, Instagram is making the decision about how this lives on.
Even after several migrations and the threat of API deprecation, lentil lives on and is still valued by library admin. Example of doubling down while sunsetting components that aren't used anymore.
How can libraries be better at sunsetting services in order to build capacity for the future?
How do we overcome our predisposition toward preservation?
Not just an issue with digital projects or software
Future Work
Sunsetting is a constant part of the project lifecycle. We hope to make this a more visible part of our ongoing project management practices, especially in terms of managing macro-level issues around portfolio size and trajectory. We also want to make this more visible in the library community.
What can you do?
Examine how you could reshape your current portfolio.
Engage your organization by calling out opportunities to create space.
Send us your case studies!
Autumn is a natural time for reflection and celebration of the past year. It's also a time of anticipation and excitement about the new year. Sunsetting projects is a natural part of the lifecycle of work and is as much about starting new work as it is about ending past projects.
Bonus
YKK If we can anticipate any questions add a slide about those here.