Database Vulnerabilities In The Spotlight

 
 
By Lisa Vaas  |  Posted 2003-02-20
 
 
 
Databases and DBAs got a double body-blow over the past week, with the news that hackers grabbed up to 8 million credit card account numbers and, separately, that Oracle Corp. released five advisories to cover six Oracle 9i Release 2 vulnerabilities.

Data Processors International Inc., a credit card transaction processing firm, confirmed Wednesday that its customer databases have been hacked by "an unauthorized outside party" and that it remains unclear whether hackers nabbed useable data, according to a company spokesman. Widely published news accounts quote other credit card companies and the U.S. Secret Service as saying that the responsible party made off with up to 8 million account numbers from Data Processors, which is based in Omaha, Neb.

Scott Jones, the DPI spokesman, declined to say when the intrusion was detected. Jones said that DPI immediately notified credit card companies and law enforcement agencies upon discovering the intrusion. DPI is working with the Secret Service and the FBI. Jones said that, so far, there has been no reported misuse of information related to the event. Personal information including account holder names, addresses, telephone numbers and social security numbers were not obtained through the intrusion, he said.

The Secret Service and the FBI advised DPI against speaking publicly about such issues as which vendors or vendors databases were targeted. Citing customer confidentiality, spokeswomen for the three major database vendors—Microsoft Corp., Oracle Corp. and IBM—said that it was unlikely that they would state whether DPI runs its business on their respective databases. They had not responded to requests to identify DPI as a customer by the time this story went to press.

Nevertheless, the incident points to hackers increasing focus on databases, as opposed to Web servers, their traditional target. Two other recent incidents lead to the same conclusion: Slammer, the worm that struck at Microsoft SQL Server database instances late last month, bringing the Internet to a crawl; and the recent discovery of the six Oracle database flaws.

Oracle, of Redwood Shores, Calif., last week released patches for four critical vulnerabilities and two that were deemed merely serious. The flaws were discovered by Next-Generation Security Software, the British security firm that discovered the bug behind Slammer.

The Oracle 9i vulnerabilities include four critical buffer overflows in various components of Oracles database server software. Overflows occur when applications mishandle memory, thus allowing attackers to insert their own code into the applications execution.

Mary Ann Davidson, Oracles chief security officer, defended what may have seemed like a blizzard of security advisories. "Where things are related and theyre combined in one patch, of course well do one alert," she said. "But in this case, I said, I dont want to play the numbers game. I dont want to say Ooo, lets make the numbers look smaller. We put ourselves in our customers position [and ask,] Whats easiest for them?"

Davidson said that Oracle has recently revised its bug-handling processes, coming up with a formula to determine how to handle security alerts. Oracle plugs conditions into the formula to arrive at a severity threshold, Davidson said. The conditions include factors such as whether the vulnerability is public and if a potential hacker has to have an account or special privileges to exploit the vulnerability. Oracle derives a number from this formula, which it then uses to determine whether to release a patch set or discrete patches.

The idea, Davidson said, is to avoid plaguing customers with multitudes of patches. "Customers in general hate applying large numbers of patches," Davidson said. "You try to find a sweet spot. We run Oracle on Oracle. Our IT people have the same issues. They like nice stable software, and they like things in nice patch sets."

Rocket Fuel