Cloud and robots: creative data storage solutions

Late November in Chicago: The snow flurries around the imposing gargoyles atop the Harold Washington Library. The Chicago Public Library (CPL) holds 5.7 million volumes in its stacks and millions of digital versions on its servers. Troves of TIFF files, high-res scans of the library’s extensive special collections, are held in dark storage. Outside data centers manage its growing information on patron use — data that is critical, librarians say, for proving their value amid strained public budgets. During the recession, the city slashed public spending through consolidation, creating a central IT department for public institutions called DO IT, the Department of Innovation and Technology, but taking away the CPL’s lone data-storage chief. Michelle Frisque, director of library technology at the Chicago Public Library, uses creative strategies to keep CPL’s data-storage costs down. At the same time, Michelle Frisque, CPL’s director of library technology, says innovative programs like a free 3D-printer workshop and tech incubator have increased patrons’ digital usage. So has the still-deflated economy, with cash-strapped households canceling pricey Internet connections at home. The CPL is preparing for a soft launch next year of Biblio Commons, a Serial Attached SCSI (SAS)-based solution that tries to engage users in the virtual environment through posting reviews and recommendations and improved search. DO IT’s new city-data portal hosts some of the library’s most important stats, such as the most popular titles in each neighborhood branch; how many kids participate in the summer reading program (60,232 in 2012, reading a cumulative 1.5 million books, 400,000 more than the previous year); or the library system’s WiFi usage (up to 400,000 sessions per month from 22,000 a year before). As the libraries’ digital use has grown, Frisque has developed a strategy to keep data-storage costs down. The two-pronged strategy combines moving tens of thousands of archives off the network server — where space comes at a premium because of the super fast processing time and high-res images — and into the cloud, a popular solution among small businesses; and outsourcing applications and user products to a SAS-based system. “We’d been running products in house on our servers, but as the product was sometimes slower than we expected, there was lots of maintenance. We’d have to add more hardware space, but then there was all this processing power that wasn’t being utilized. We needed something that could adapt and change with our changing needs,” Frisque says from her 10th-floor office at Harold Washington. Keeping costs down Will this data storage strategy turn out to be a cost saver? Frisque can’t say. Corporations and major institutions worldwide are exploring alternatives as storage needs outstrip traditional options. “When you’re talking about petabytes of information, that’s millions of dollars in management costs,” Phil Goodman, a principal architect at Cognizant Technologies, said during the Storage Decisions conference in Chicago last June. “You have an inverse relationship between the capacity of the device and its performance.” Goodman advises clients to explore automated tiering — placing a thin layer of solid-state drive (SSD) at the top — and software that detects how often files are being used; moving hot data from inexpensive, cold storage to the high-performing disk and back, once it cools. “If it’s Facebook and people post pictures but don’t look at them very often, you want something that doesn’t cost very much. If I’m buying something on Amazon, I probably want that to happen really quickly,” says Tom Coughlin, an electrical engineer and storage consultant. Library robots Some big academic institutions and media-and-entertainment companies have opted for the priciest solution to archive materials: robotic libraries. They rely on robots to locate, select, load and later shelve cartridges of magnetic tape. The University of Chicago’s Mansueto Library holds 3.5 million volumes in underground storage that can be accessed by robotic cranes. Like the data itself, storage is constantly evolving, with the costs dropping as the capacity rises. Today’s devices will probably be obsolete within a decade. “Organizations have to have a migration plan, and with the amount of content some of them have, it might be a continual migration,” says Couglin.