November 22, 2017

Evergreen: Your Homegrown ILS

An in-house team successfully competes with commercial vendors in the library automation sphere

About three years ago, the Georgia Public Library Service (GPLS) looked for a new integrated library system (ILS) to serve its large consortial group of libraries across the state and found its needs frustrated by the commercial ILS market. This September 5, it debuted a new library system and catalog. Evergreen was developed by a small in-house team using open source technologies, at significantly lower cost than the commercial options that were available. This strategy has proven dramatically more flexible in meeting the needs of GPLS, and the new system has been welcomed by librarians and patrons alike.

Not your father’s library system

You might imagine a homegrown library system as clunky and unattractive. Not Evergreen. Staff developed an interface based on the engine of the popular open source web browser Firefox, designing it from the ground up to meet the needs of users. It’s full of modern and convenient features, including tabs for juggling multiple screens of information.

The public catalog is where Evergreen really shines. In an era when many have proclaimed ‘My OPAC sucks,’ the developers of Evergreen set out to counter the shortcomings of the catalog compared with modern web applications – modeling interfaces users are comfortable with like those of Google and Amazon.

Among Evergreen’s characteristics is spell-checking of search terms with suggested alternates, much like Google’s suggestions when you misspell a word. Evergreen features faceted browsing by subjects, authors, and series to find related books easily from the title you’re currently viewing, as well as conveniently grouping various editions and formats of the same title together (in the vein of the FRBR model; see Linda Gonzalez’s ‘What Is FRBR?’ LJ netConnect, Spring 2005, p. 12). Evergreen enables tight integration of enriching content such as book covers, tables of contents, and reviews, which are bolted onto many catalogs as an afterthought, appearing in a different window and clashing with the catalog’s look and feel.

Evergreen’s catalog also improves patrons’ ability to manage their accounts. They can see which books are checked out, renew books and place holds, and create ‘bookbags,’ reading lists that can be shared with others.

Beginning the journey

The road to Evergreen started back in 2004, with a vision of a better ILS for Georgia’s libraries. GPLS, Georgia’s state library agency, administers PINES (Public Information Network for Electronic Services), which provides library automation services for public libraries statewide at no cost to the member libraries. PINES serves 44 distinct library systems with over 250 locations – two-thirds of all the libraries in Georgia – and has a shared collection of over eight million items.

Given this organizational structure, PINES needs a system that allows local control of users, policies, and other administrative functions, while it maintains a centralized system to facilitate the statewide borrowing network. The sheer size of PINES, with hundreds of staff and 1.6 million cardholders, demands a robust and fast system.

When PINES librarians looked at commercial systems to replace the aging ILS, they weren’t convinced that any of them could meet these requirements. Brad LaJeunesse, PINES system administrator, explains, ‘The ILS is like a train. In the traditional ILS [for a single library or library system], all of the users and entities are thrown into one big boxcar. What we wanted was 250 separate passenger cabs for our libraries. We also didn’t want 250 separate trains.’

The PINES group approached GPLS with the idea to create an ILS from scratch to meet these needs and was initially met with skepticism. Eventually, it got a green light for a trial year to begin to develop the system.

Everything old is new again

Library systems and other software developed in-house were once common, and some contemporary ILS’s have their roots in systems originally developed at an institution – for example, NOTIS from Northwestern University led to Endeavor. Today, however, homegrown software has largely been replaced with commercial offerings. Most libraries find it easier to have a vendor develop and maintain software systems rather than recruiting, retaining, and managing internal staff to do so.

Software development has changed greatly in the last decade, largely owing to the open source movement. Rather than building software components from the ground up, in-house teams often rely on a large array of highly refined components for various functions, including the Apache web server for scalable applications. Leveraging existing, freely available source code means developers can build on the experience of many others who have provided solutions to common problems, freeing them up to concentrate on their own unique issues. This can also make it much easier to get outside expertise or bring a new developer up to speed, since the system is based on well-known components.

This is the environment that made Evergreen possible with a small development team and an aggressive time line.

People and processes

PINES already had LaJeunesse and Jason Etheridge, an interface designer, on staff. In addition, Bill Erickson, systems developer, and Mike Rylander, database developer, were hired for the project. The PINES staff also included members with many years of public library experience who contributed greatly to the design of the system.

PINES policymaking is governed by an executive committee, elected from library directors around the state. It is advised by a number of subcommittees composed of librarians focusing on functional areas, such as cataloging. The development team solicited input from these groups in building the software and convened focus groups around the state to hear what librarians and staff needed. They did not seek public input in this phase.

In the past, a number of policies had been dictated by the limitations of the ILS. ‘The focus groups forced staff to think through the procedures they had been using and to look more globally at what we wanted to accomplish,’ says Kathryn Ames, director of the Athens Regional Library System. The PINES team wanted to free these policies from the limitations of previous software, requiring a change in mindset from accustomed operating procedures. ‘At first,’ says LaJeunesse, ‘people would suggest workarounds of the old system as features for the software. Jason told them, ‘pretend it’s magic.’ That seemed to help.’

The team moved quickly, building functions and interfaces and testing them with users, then redesigning until they were a good fit. As the team began showing prototype systems in early months, skepticism about the project began to dissolve, and librarians grew more confident. Dusty Gres, director of the Ohoopee Regional Library System in Vidalia, admits to being leery of the project at first but says, ‘The one thing that changed my mind, almost midstream, was that I saw a demonstration of an early release, saw an immediate problem with one part of it, brought it up right then, and the next time I saw a demo, it was fixed.’

When the trial period ended in July 2005, PINES had an ‘alpha’ version of the software with basic functionality. By that point, there really wasn’t much question of whether the project should continue. Results had been overwhelmingly positive, and the executive committee gave the go-ahead with a unanimous vote.

Flipping the switch

Labor Day weekend 2006 was chosen as the launch date for Evergreen back in 2005, and the date stuck. PINES staff wanted to have at least the same functionality in Evergreen as in the old ILS, so circulation, cataloging, and an OPAC needed to make it into the 1.0 release. (PINES never included acquisitions or serials; the structure of the organization, with budgets distributed over many independent organizations statewide, simply made those too difficult. They are slated for version 2.0 of Evergreen, however, tentatively scheduled for summer 2007.)

PINES continued in anticipation of the launch, refining Evergreen’s features and interfaces. The staff also went to work planning a smooth transition from the old ILS to Evergreen, including migrating data with patron information, bibliographic records, bills, checkouts, and holds. They performed several test runs of the data migration and had users spot-check information to identify any problems before the system went into production.

Meanwhile, the rest of the PINES staff spent most of August training workers at library systems across Georgia. PINES central staff split into teams for circulation and cataloging and trained hundreds of local administrators, who in turn trained staff at each library. Many libraries created training materials that were shared with others using a wiki for documentation.

Then, on the Friday evening before Labor Day, the old ILS was shut down. The team worked over the weekend and had the new system up and running in the afternoon on Monday (Labor Day), in preparation for libraries to reopen Tuesday morning.

Evergreen’s first day was bumpy. The PINES developers, in their quest to have everything perfect for launch, hadn’t slept much all weekend, and this diminished their ability to respond to issues that cropped up throughout the day. The PINES Help Desk manager position had also recently become vacant. Additionally, the new web catalog drew a great deal of attention from patrons and others on its first day, and usage was suddenly much greater than previous levels, leading to slow performance and some downtime through the day.

In all, however, the migration went well. Migrating an ILS is a notoriously punishing experience. PINES staff had collective wisdom from several large ILS moves in the past, and everyone knew the first day would be tough. On Tuesday night, the team regrouped and made a number of fixes and changes, and the second day went more smoothly.

The PINES team continues to work with staff to iron out issues and bugs. ‘There continues to be some tweaking – in fact, that’s probably the best thing about developing this product,’ Ames says. ‘As problems crop up, and they inevitably do, the software developers have acted quickly to respond with a fix.’ Anyone can submit bug reports, which are then assigned to the appropriate development team member and prioritized using an automated tracking system.

Real-world results

Now that Evergreen is working, many doubts about the ability of an in-house team to deliver a robust and feature-rich library system have been laid to rest. ‘The strong point of Evergreen is that it is intuitive and forgiving,’ Gres says. ‘For my system – rural small libraries, many of them staffed by one person, relying solely on my small professional staff and me for training, computer expertise, and backup – I have had less trouble with this system in the two months we have been in operation than I did in the first two weeks when we automated on a commercial system.’ And since Evergreen is developed in-house, the PINES developers are accessible to users and ready to respond. ‘With the vendor [SirsiDynix] for our old system, my concerns about weaknesses and needs for certain areas to be more user-friendly were not met with any real understanding,’ says Gres. ‘I guess that’s been the best thing about this: the response of the development team. They listened.’ Ames adds that training also continues to be important, as staff members become more comfortable with new processes.

Cashing in

Developing Evergreen was significantly cheaper than a commercial system. The estimated upgrade and migration of the old system to a newer commercial ILS was quoted at nearly $2 million and additionally required a new server for $1.5 million. PINES paid the vendor $200,000 annually in support and maintenance charges and paid a hosting company another $200,000 to house the server.

The Linux server cluster and storage system used for Evergreen cost only about $250,000 – much less than the large server required by the vendor system because it’s built from many smaller and less-expensive servers working together. This equipment is hosted in a collocation facility for less than half the cost of hosting the old systems. Two staff members were added to PINES to develop Evergreen, and some money also went toward consultants and other outside services, but the vendor migration costs and annual support fees were completely avoided.

Some of that savings is being used up by increased costs for shipping holds and overdue notice mailings. The more user-friendly system has increased holds by 30 percent over the old system. ‘When it’s not painful, people use it,’ says LaJeunesse. Overall, libraries report that patrons have been very happy with the new system and its ‘Amazon-like’ features.

Jennifer Wood, of Lincolnton, GA, a patron of the Lincoln County Library, appreciates the new catalog. She homeschools her three children and says she really values the feature that allows her to sort out easily juvenile and adult items in searches. ‘My favorite feature would have to be the ability to order items from other libraries,’ she says. ‘Before, we would have to go into the library to do that.’ She also says the new catalog is faster, and it’s ‘just cooler.’

Nevertheless, Marshall Breeding, director for innovative technologies and research at Vanderbilt University, Nashville (and author of LJ‘s annual Automation Marketplace survey), remains skeptical about the future of homegrown systems. ‘Implementing open source projects such as Evergreen for critical library functions involves a higher level of risk than the commercial alternatives,’ he says. ‘[PINES] swapped out a full-featured system for one early in its development cycle. It’s just beginning.’ Breeding says while open source software solves some problems, libraries also look at things like the track record of a company, capitalization, support, and customer relationship management.

The future of Evergreen

Work on Evergreen continues, with the PINES team gearing up for a second version, which will include acquisitions and serials management and enhanced reporting features.

Evergreen’s source code has also been released under an open source license, making it available for other libraries to use and adapt. Its modern interfaces and fresh thinking about library automation problems are sure to be influential in other developments as well. ‘The new software is generating considerable excitement among patrons and library staff across Georgia, across the country, and even internationally,’ says Julie Walker, PINES program director. ‘We’ve had inquiries from as far away as South Africa.’ It joins other open source library systems such as Koha, increasing the noncommercial options for library systems (see ‘Evergreen Powers LibLime,’ above). PINES folks are showcasing the system at a number of upcoming events and conferences and will strive to build a community around the open source project, since improvements to the software benefit all users.

Implementing Evergreen was certainly hard work, but it has shown that an in-house team can successfully compete with commercial vendors in the library automation sphere. Much of the project’s success can be attributed to the close working relationship of developers and end users – a condition that is difficult to replicate for a software vendor serving many customers and the reason that in-house development may sometimes be a viable and even preferable option for libraries.


Link List
Evergreen Catalog
www.gapines.org
Open ILS Project
www.open-ils.org

 

 

Behind the Scenes

Evergreen makes use of a number of open source technologies. The client application used by staff is based on XUL (XML User Interface Language), a scheme for interface design that emerged as part of the Firefox web browser.

Data for Evergreen is stored in a PostgreSQL database (the same database technology used by DSpace), and the application logic is handled by routines in C and Perl. Apache is used as a web server.

‘Messaging’ is an important part of application frameworks, allowing different processes to exchange information. For Evergreen, it was especially important to have a framework for messaging that was highly robust and scalable. The developers designed a framework (released under an open source license as ‘OpenSRF’) based on Jabber, a decentralized XML technology originally intended for messaging between people, like AOL Instant MessengerSM. Jabber’s scalable design also made it ideal for use as an application messaging framework.

All this software runs on a cluster of about 40 Linux servers in a highly secure and robust data center (also used by Google and other big-name customers), with redundant connections to electrical power, data networks, and other features to keep the servers up and running at all times. The servers pack some serious computing power and have over two terabytes of storage.


Evergreen Powers LibLime

If your library is located outside Georgia, you may still be able to use the Evergreen system via LibLime, an emerging open source support company. Joshua Ferraro, LibLime president, describes LibLime’s role, noting that they provide ‘hosting, data migration, installation, training, and technical support’ to libraries looking to migrate to Evergreen. LibLime is intimately familiar with the Evergreen project, having been retained by PINES during its development to provide quality assurance and National Circulation Interface Protocol (NCIP) support.

Ferraro says, ‘[PINES] librarians are happy with Evergreen. Plus, suggestions are sometimes implemented within hours.’ One benefit of open source projects is the visibility of the bug list, where enhancements and problems can be openly tracked.

LibLime currently has four employees and partners with ten subcontractors; it is in the process of hiring another four employees for development and support. The company has 50 clients, including public libraries in Ohio and libraries in France, New Zealand, Switzerland, and Canada. Several large U.S. public libraries are reportedly in discussion with the company about Evergreen.

The company’s origins are in the Koha project, an open source ILS developed for New Zealand libraries. Ferraro became familiar with Koha as system administrator, Nelsonville Public Library, OH, where he implemented it with then-director Stephen Hedges in 2002. Ferraro became release manager for Koha version 3 this year. Meanwhile, Hedges has become executive director at OPLIN, the Ohio Public Library Information Network. – Jay Datema


Jonathan Weber is an information architect and technical writer based in Pittsburgh. He contributed user documentation for Evergreen while he was a library science student at the University of Pittsburgh

Share