The eXtremeDB database system accelerates high performance desktop, workstation and server-based applications

Designed from scratch as an in-memory database system (IMDS), eXtremeDB delivers breakthrough responsiveness along with high scalability, reliability and a proven track record.

These characteristics drive eXtremeDB’s growth as a real-time database for high performance software, whether residing on individual PCs/workstations for local processes, or on servers.

Why are developers of high performance software ranging from algorithmic trading systems to Web-based e-commerce turning to eXtremeDB?

Performance

As an IMDS, eXtremeDB eliminates I/O, cache management, data transfer and other sources of overhead inherent in disk storage. In contrast, these performance-draining processes are hard-wired into most other database systems.

eXtremeDB’s in-process architecture accelerates performance, compared to the client/server design of relational database management systems (RDBMSs) and many IMDSs. eXtremeDB supports standard SQL, but developers can use a faster native C/C++ API. Information is stored as C/C++ data types (including structures, vectors and arrays), eliminating the overhead of conversion to SQL data types.

Even among IMDSs, eXtremeDB's execution path is exceptionally short. A code size as small as 150K heralds an efficiency obtained by eliminating extraneous processing and paying attention to even small sources of latency.

Scalability

Although eXtremeDB itself imposes a small code footprint, it excels in handling very large databases. Features that help maximize throughput and fully leverage multi-core CPUs include the following:

64-bit support (eXtremeDB-64) enables in-memory databases to scale to terabyte-plus sizes

With eXtremeDB Fusion, selected record types can be stored to disk (or flash chip), with on-disk database size limited only by available file system space

An optional multi-version concurrency control (MVCC) transaction manager eliminates "pessimistic" locking to accelerate multi-threaded applications running on multi-core hardware

eXtremeDB Cluster manages databases across multiple hardware nodes. This can dramatically increases available net processing power while maximizing uptime and reducing system expansion cost (via the use of "commodity" servers)

How fast is an in-memory database at scale? In the first benchmark of an IMDS exceeding one terabyte in RAM, eXtremeDB-64 demonstrated near linear performance gains while scaling from one core up to 160 cores, and achieved more than 87 million query transactions per second.

Reliability

From its development tools to support for distributed architectures, eXtremeDB meets the reliability imperative of mission critical high performance systems.

eXtremeDB transactions safeguard data integrity with support for the ACID (atomic, consistent, isolated and durable) properties. A type-safe application programming interface (API) and other advanced debugging features help produce bulletproof code.

To mitigate volatility, eXtremeDB Fusion provides highly selective persistent data storage, while eXtremeDB Transaction Logging delivers transaction durability via roll-forward recovery. Both editions enable precise durability/performance tradeoffs.

Both eXtremeDB High Availability (based on master-slave replication) and eXtremeDB Cluster (in which every database instance is "master") provide the basis for distributed architectures in which the database can survives hardware and/or software failures.

Proven Track Record

Perhaps eXtremeDB's most confidence-inspiring "feature" is its track record. Used in production systems for close to a decade, eXtremeDB powers millions of real-time applications in fields as diverse as capital markets, on-line social networks, power grid management and telecom services.

To determine how eXtremeDB fits in your high performance application development project, browse the eXtremeDB product family, download evaluation software - or contact McObject to discuss your needs.