[sweerie_banana]

Real-time or reactive computing is a conviction that outlines any computing system that must answer to change in the environment in consonance with the exact time constraints, usually on the order of milliseconds. Lacking to meet those time constraints would make the system impractical or ineffectual.

Real-time computing can be categorized into two:
1. The number of information is enormous and the results cant be computed earlier, while user response has to be in real-time. Its main uses are for specific data analysis and processing. It can process part of the information in advance and combining it with the real-time computing results to ameliorate processing coherence.

2. The data source is real-time and ongoing is called streaming data. A data stream is a cluster of a sequence of data records that are boundless in time distribution and amount. A data record is the tiniest unit of data streams. Real-time data computing and analysis can examine and number information dynamically and in real-time, this has major pragmatic importance on system monitoring, organizing, and management.

The real-time computing process can be divided into three processes:

1.The real-time data collection.
It must make sure the cluster of all the data and must provide real-time data for real-time applications. Response time must be real-time and little lag. Configuration should be simple and deployment should be easy. The system is required to be stable and reliable. Big data acquisition tools are like Facebook's open-source scribe, Linkedin's open-source Kafka, and Hadoops Chukwa.

2.Real time data computing.
They in general adopt the given out architecture of MPP, data storage is then assigned to large-scale nodes, to meet the real-time requirements. They use large scale distributed file systems like NoSQL databases.

3.Real time query service.
Its implementation can be categorized into three ways namely:
a}.Full memory which provides data read services directly and dumps onto disks or database for backups regularly.
b}Semi memory which uses Redis, Memcache, and other databases to provide real-time query services and leave backup operations to the systems.
c}Full disk which uses NoSQL database such as Hbase.

Google's Dremel system is the most well known.

[ptrikha21]

Well described. This could have worked as an article.
From your article, I infer that NoSql databases like Redis have some big differences with NoSql dbs like Hbase.
And where do MongoDb and Cassandra fit in?
And is Dremel some sort of a web service?