BPE 2011: ERA and the move to the cloud

This week, I’m spending a little time with my parents in Ohio and at the 2011 Best Practices Exchange (BPE) in Lexington, Kentucky. The BPE, which brings together state government, academic, and other archivists and librarians and other people seeking to preserve state government enduring information of enduring value, is my favorite archival conference. The Society of American Archivists annual meeting is always first-rate, but it’s gotten a little overwhelming, and I love the Mid-Atlantic Regional Archives Conference (MARAC), but nothing else has the small size, tight focus on state government records, informality, and openness that characterize the BPE.

Before I start detailing today’s highlights, I should say a few things about the content of these posts. For the past few years, those of us who have attended the BPE have tried to adhere to the principle that “what happens at BPE, stays at BPE.” This doesn’t mean that we don’t share what we’ve learned at the BPE (hey, I’m blogging about it!), but it does mean that we’re sensitive to the fact that candor is both essential and risky. The BPE encourages people to speak honestly about how and why projects or programs went wrong and what they learned from the experience. Openness of this sort is encouraging; all too often, we think that we’re alone in making mistakes. It's also helpful: pointing out hidden shallows and lurking icebergs helps other people avoid them. However, sometimes lack of senior manager commitment, conflicts with IT personnel, and other internal problems contribute to failure, and colleagues and supervisors occasionally regard discussion of internal problems as a betrayal. As a result, BPE attendees should exercise some discretion, and those of us who blog about the BPE should be particularly careful; our posts are a single Web search away. As a result, in a few instances I may write about the insights and observations that attendees have shared but obscure identifying details.

Moving on to this year's BPE itself, I'm going to devote the rest of this post to the insights and predictions offered up by U.S. National Archives and Records Administration (NARA) Chief Information Officer Mike Wash, who spoke this morning about the Electronic Records Archives (ERA), NARA’s complex, ambitious, and at times troubled electronic records system, and some changes that are on the horizon.

At present, ERA sort of works: staff use it to take in, process and store electronic records. At present, ERA holds approximately 130 TB of data. The Office of Management and Budget wants NARA to take in 10 TB of data per quarter, and NARA is working with agencies to meet this benchmark. However, ERA lacks an integrated access mechanism, and it contains multiple modules. The Base module handles executive agency data, the EOP module handles presidential records (and includes some internal access mechanisms), the Classified module holds classified records, and several other modules were built to deal with specific problems.

Building ERA taught NARA several lessons:

• Solution architecture is critical. ERA’s multiple modules are a sign of a failed system architecture. Anyone building such a system must carefully consider the business and technical architecture carefully during the planning stage and must manage the architecture carefully over time.
• The governance process must be clear and should start with business stakeholders. What do they really need the system to do, and how do you ensure that everyone stays on the same page throughout the process? Information technology invariably challenges control and authority, but if you set up your governance process properly, you should be able to retain control over system development.
• Over communicate. Funders and other powerful groups need frequent updates; failure to keep feeding information to them can be profoundly damaging.
• You must manage the project. The federal government tends to hire contractors to develop IT systems, and contractor relationships tend to deteriorate about six months after the contract is awarded. Most federal agencies cede authority to contractors because they are loath to be seen as responsible in the event that a project fails, but staying in control of the project increases your chances that you'll get the system you want.
• Watch costs closely. Cost-escalating provisions have a way of sneaking into contracts.
• Be mindful of intellectual property issues. The federal government typically reserves the right to all intellectual property created as a result of contracts, but this doesn’t always happen, and the vendor that built the first iteration of ERA has asserted that it controls some of the technology that now makes the system work; NARA will be much more assertive in working with future ERA vendors.

Wash also made some intriguing observations about some of the challenges that NARA and other archives are confronting:

• At present, our ability to acquire data is limited by bandwidth limitations. It takes more than three days to convey 20 TB of data over a 1 gbps data line and at least a month to convey it via the Internet. NARA recently took custody of 330 TB of 2010 Census data, and it did so by accepting a truckload of hardware; at present, there are no alternatives to this approach.
• The rate of data creation continues to accelerate. The administration of George W. Bush created 80 TB of records over the course of 8 years, but the Obama administration likely created more than 80 TB of data during its first year.

Wash indicated that NARA is starting to think that federal records should be created and maintained in a cloud computing environment and that transfer of custody from the creating agency to NARA should be effected by changing some of the metadata associated with the records being transferred.

Wash noted that the move to cloud computing will bring to the fore new preservation and authentication concerns. It also struck me that the transition that Wash envisions assumes the existence of a single federal government cloud that has adequate storage, security, and access controls and that, at least at this time, many states aren’t yet thinking of constructing such environments. Individual state agencies may be thinking of moving to the cloud, but most states don't seem to be preparing to move to a single, statewide cloud environment. Moreover, owing to its sheer size, the federal government is better able to negotiate favorable contract terms than state or local governments; the terms of service agreements that the feds hammered out with various social media providers are an excellent example. I have the uneasy feeling that some governments will accept, out of lack of knowledge, desperate financial straits, or inability to negotiate optimal terms, public cloud service contracts that prove problematic or outright disastrous.

Its nonetheless apparent that government computing will move into the cloud, that this transition offers both new challenges and new opportunities for managing and preserving records, and that archivists and records managers are going to have come to grips with these changes. The next decade promises to be most interesting.

The Lexington Laundry Company building on West Main Street, Lexington, Kentucky, 20 October 2011. This little gem was built ca. 1929, is an outstanding example of Art Deco architecture in the city, and is part of Lexington's protected Downtown Commercial District. It now houses an art gallery.