Summary

We publish LunarBase as a doc-oriented NoSQL database, not merely a key-value store or a cache solution. Since a big cache system is self-contained, LunarBase has a low deployment dependency and cost, you don't have to deploy another k-v cache for you hot data anymore.

It is:

with a persistent DB storage;
with a self-contained big cache system, enabled concurrency, high throughput capability;
Free scalable, document-oriented without predefined schema, as all NoSQL database designed for;stem
Benefit from the principle of Google Big table, LunarBase has extremely fast retrieval speed;
for multiple purpose. One DB for both data storage and search, and even full text engin with simple word parsing plugins;

A single node supports:

single DB stores 2 billion records, each has a size limitation of 32K bytes. Hence the total size of one DB is 64TB；

One server runs several LunarBase Instances, each manages one DB. The number of instances depends on the hardware capability；

Concurrency support depends on the number of cores of CPU. There is a one-one mapping from internal threads and CPU cores；

Memory Consumption: LunarBase has an internal MMU, will not produce memory fragments after a long time running. This is an very important feature for any server-end middle ware;

Query speed: on a 4-core X86 server, mechanical HDD, 8G mem, LunarBase responses a query in 10 ms. If the big cache is opened and all hot data stores in it, the query speed of hot data are extremely fast.