JDB


JDB is a powerful database and storage solution.


Key features


Performance

Database lookup is faster than 1 million lookups per thread in a multithreaded setup, or more than 5 million lookups per second in a single-threaded setup.
Inserts depend on how many different containers you are inserting to.
If inserting to the same container, inserts are faster than 1 million inserts per second single-thread. For multi-thread and same container, insert speed will be be divided between threads.
Deletions are about the same speed as insert based on the same container.
If you insert into multiple containers with multiple threads, performance is increased multiple times.

Check real life performance test here


Storage

JDB has its own filesystem and do not require traditional filesystems like EXT4 or XFS (traditional filesystems are also supported). JDB can utilise raw storage devices directly preventing overhead and extra fragmentation. This is particularly useful when running virtual machines that already have disks allocated from an underlying physical storage medium.
When JDB runs out of available space, it can automatically request more storage to the JVM virtualization system through API and extend its capacity automatically.


Clustering

All nodes in a JDB cluster are active, meaning JDB is a true active-active solution where all nodes are equal.
Sharding, horizontal scaling, map-reduce or similar can easily be achieved by client software due to the asynchronous and zero-copy technology JDB provides.


Monitoring

The system support the real-time monitoring solution Loke