SQL injection attacks aren't going away; in fact, they remain one of the top threats on the Web. Locking the doors to the database is a key part of enterprise security, and it's also where database security vendor Secerno is hoping its approach will separate it from the competition.
Secerno is far from being alone in this market. Among others, it competes in the DAM (database activity monitoring) market with established vendors such as Guardium, Tizor Systems and Imperva. What is different about Secerno is the way it goes about the business of securing the database.
Secerno's DataWall products are powered by the SynoptiQ Engine. The technology works by leveraging a model of how users and applications typically access the database in order to help organizations set granular policies. Once those policies are established, the DataWall appliance uses them as a basis to detect and block suspicious traffic.
SynoptiQ enables DataWall to analyze each database query statement in full and understanding the intent of the whole interaction, rather than just checking for keywords irrespective of their context. With this approach, the SynoptiQ Engine makes a predefined decision based on the intent and therefore does not have to check through all the variations and pattern matching alternatives in a regular expression-based signature approach.
"Pattern detection relies on recognizing known malicious exploits-it is vulnerable to new custom-crafted attacks," said Paul Davie, chief operating officer at Secerno. "The Secerno approach flips this over. It measures what is normal, allowing the owner to detect anything else, whether it has been seen before or not."
"The accuracy with which SynoptiQ matches each query to the defined policy means that it can be trusted inline, as it is by more than half of Secerno's customers," he continued. "From here it can actually stop malicious queries. DB [database] monitoring merely reports on malicious activity, after the event, without actually stopping it."
According to Davie, Secerno's product needs to see one query of each semantic structure in order to build its behavior models.
"In some cases, you will see every type of transaction within an hour," he said. "In others, the normal period is a few days. The main point is that the system is deterministic rather than probabilistic-it is not based on statistical methods. ... Typically, a quality baseline can be built within a day, and we are able to estimate completeness metrics."
Blocking database traffic is often tricky territory. Traditionally, enforcement falls into basic areas-blocking based on access polices and using signatures to thwart known database exploits. While the company's claims of zero false positives were met with skepticism from analysts interviewed by eWEEK, those same people credited Secerno for its approach.
"The data that Secerno collects is not unusual; how they analyze it is," said Adrian Lane, an analyst with Securosis. "By examining the structure and usage of the database queries, they are in essence looking at a type of fingerprint of the SQL language itself. They are unique in this regard, and I believe the approach to be effective with many types of misuse."