McObject Debuts Database Monitoring Display

Issaquah, Wash.-based McObject, a provider of embedded in-memory databases,  is developing a graphical user interface for monitoring activity in its eXtremeDB embedded database, as well as an API for accessing the captured statistics, which can help users configure the database to enhance application performance.

The vendor plans to release a GUI this quarter for capturing database statistics, including disk input/output wait time and throughput, CPU load and the host processor’s memory usage, as well as memory usage by the database or cache, to support the development of ticker plants, low-latency trading platforms and matching engines. “The idea is to give a developer or database administrator visibility into what’s going on within the database, so they can tune parameters within eXtremeDB to minimize jitter and latency,” says McObject co-founder and chief executive Steve Graves.

Clients can configure how frequently eXtremeDB takes a snapshot of the database—for example, a user could take daily snapshots at the end of a trading day to build a historical database, or intra-day to trigger a flush of the database’s transaction log cache during normal operations, and only storing the snapshot, to avoid filling the entire disk’s memory, Graves says. The new GUI would also enable customers to track how often the cache is emptied and how long it takes.

The same statistics will be available via an API, to enable developers to display them within other applications or to program applications to adjust themselves dynamically in response to the statistics, which in particular serves clients who embed eXtremeDB into their own branded applications, Graves says.

McObject is also changing the database’s runtime to optimize it for financial services clients—whereas its traditional clients have included embedded systems providers, such as suppliers of set-top boxes for televisions. For example, while eXtremeDB was originally built to accommodate the highly variable data in embedded systems—such as the different lengths of program descriptions in the TV guide of a set-top box—there is much less variation to datasets such as tick data, so the vendor has optimized the storage and performance of the database for this kind of data.

Footprint
In addition, after releasing the Cluster edition of eXtremeDB last year—which distributes database processing over multiple hardware nodes to increase processing power and scalability to support growing data volumes—the vendor is now enabling data nodes to store local tables so data is only distributed if required by an application.

“In embedded systems, the watchword is having a small footprint and keeping the code size small, but in the capital markets, with the growing volume of data that needs to be processed in real time, we’re providing the ability to efficiently and cost-effectively expand with the Cluster edition,” says Ted Kenney, director of marketing at McObject.