Skip to content

Real Time Analysis: LunarMax

feiben edited this page Mar 24, 2016 · 18 revisions

Real time computation is becoming imperative in this big data era. For analyzing big throughput data stream, people want a quick result, not a tedious weekly waiting as what out-dated database systems do in old ways.

The computer hardware development empowers the ability of real time computation. Memory price keeps dropping, and CPU cores are growing. Now in a small size data center, server has 32GB+ memory, multiple cores and TBs HDD is very common, which gives us a big room to design efficient in-memory data structure to perform real-time analysis.

LunarMax is the real time module integrated in LunarBase that provides extremely fast calculation for any of your data. Any hot part of properties specified like rt_analysable = string:name, int:payment, int:age, string:product, are kept a compressed copy in an internal in-memory file system, so that any computation on them has no disk latency. You will load any data on demand. Since LunarBase is capable of managing 64TB data, and in most of the cases, not all dimensions are needed in real-time, so LunarBase persists all but loads in memory only the hot part of what you specify.

We talk about it in the following sectors, see how LunarMax greatly accelerate your business and simplify your IT environment:

Configuration

In-Memory File System: Memory estimation

Use SSD as a cheaper memory

Data Safety

High End Server VS. Cluster

High Availability

Clone this wiki locally