Sex offender tagging system crashes
Database screw-up takes 16,000 off the grid
Some 16,000 assorted sex offenders, parolees and other convicts across the US were left unmonitored for 12 hours last week, because the monitoring firm's database exceeded a 2.1 billion record threshold.
The tagging system was run by BI Incorporated of Boulder, Colorado, and is called the BI Total Access Server. What should happen is that tagged offender's movements are detected by a GPS system, or other monitoring technology, and notifications sent automatically to BI's customers. On reaching the database limit, this system shut down - although BI still received the offenders' moment data. It took 12 hours for the database, to be restored, and it now has a limit of 1 trillion records.
When the database was back up the delayed alerts were sent out to BI's customers.
BI Inc. spokeswoman Monica Hook said in a statement: "The BI TotalAccess Server is used for a variety of offender risk classifications, including low-, medium- and high-risk defendants and offenders." BI staff told its customers what had happened and kept them updated with progress reports as the resolution took place.
Hook's statement said: "BI's primary concern is public safety. BI is — and has been for 25 years — a quality-driven organisation. BI is completely committed to public safety and the trust our customers place in us. The monitoring and systems team responded with speed and professionalism." Hook did not apologise for this 12-hour lapse and quality failure due to a seemingly obvious problem; databases can run out of record space.
Her statement included this it's-not-our-fault statement: "Len Silverston, a nationally recognised database expert, said the issue that caused the server problem was very unusual and not a reflection of any memory storage problems. 'In my 28 years as a database management expert, running out of values in a column in a table is very rare and very difficult to reach,' he said. 'To see this coming with what the designers had put in place would be very difficult.'"
Hook did say: "We are working with Microsoft to develop a warning system on database thresholds so we can anticipate these issues in the future.”
The implication here is that there was no threshold-about-to-be-exceeded warning before the event. BI was monitoring its offenders better than it was monitoring its own database. ®
Sponsored: Hyper-scale data management