Relational vs. Time-Series

Relational databases - databases that maintain a set of separate, related files that are combined when required - are fine for most data needs. However, time-series data has several unique requirements that make a relational database inadequate. First, it is usually collected in massive quantities. Second, the data is often input at an extremely rapid pace (sometimes in less-than one-second intervals). Because relational databases are not designed for time-series data, they require tons of disk space to organize and store it, and frequently become jammed, costing organizations time and money.

Time-series data, therefore, requires a specialized database to meet its rigorous demands. In response to this need, the founders of LIM developed Historis - one of the fastest time-series databases in the world.

What is the difference between a relational database and the Historis time-series database? Relational databases are extremely flexible and allow users to store and retrieve virtually any type of information. This one-size-fits-all solution, however, comes at a price when it comes to time-series data. Since there is no standard organization or table structure for these systems, it is up to the user to understand where and how the data is stored. This results in tremendous inefficiency in terms of both input and retrieval.

Historis, on the other hand, is designed specifically to meet the intense demands of time-series data. By accepting only time-series information, the system can focus on optimizing its storage, manipulation, and retrieval algorithms for this type of data. Historis does not actually store every date associated with values in the database. This allows it to dynamically insert dates into the positions where dates are not stored, thus reducing the I/O requirements on the server as well as the overall disk space required for loaded data. Furthermore, Historis' 64-bit capacity means that data can be input at any speed with no fear of clogging or jamming.

Additionally, collected time-series data must be quickly retrievable. The same problems encountered while inputting time-series data into a relational database reappear during the retrieval process.

To maximize retrieval efficiency, Historis loads each relation's dimensional data onto the disk in a single pre-derived tree structure schema that resides in the server's RAM at all times. Historis then utilizes memory mapping to access the information. This method limits the amount of work the system has to do when retrieving data, resulting in unparalleled speed and flexibility.

The Historis time-series database meets a specific need in a specific manner. Relational databases are excellent for working with a variety of data types in a variety of environments. However, if you are looking to perform serious time-series analysis, no other solution on the planet can match Historis in terms of speed, capacity, and scalability.