Extensions to Dynamic Server

The IBM Informix TimeSeries DataBlade module allows you to create an intuitively organized data model by grouping all the timestamped data for an entity into a single row in a database table, using the TimeSeries data type. The TimeSeries data type is a type constructor that creates a collection of elements that are row types, as illustrated in Figure 5.

Figure 5. Time Series Architecture

You create the row type to fit your data; the first field, or column, in the row must be a timestamp of type DATETIME YEAR TO FRACTION 5, but the rest of the columns can be any data type supported in row types. Since the time series data is in a row type instead of an opaque type, you can retrieve individual columns within an element.

The collection elements are indexed according to their timestamp, making retrieval of chronologically contiguous elements very fast. Once the number of elements exceeds the user-supplied threshold, the IBM Informix TimeSeries DataBlade module moves all the elements to a container, which exists in a user-defined dbspace. Containers are necessary because time series data typically becomes too large to fit in a database table. Containers also allow you to retrieve only the information you need, instead of the whole time series, as would happen if you used a smart large object to store your data.

For both regular and irregular time series, the Calendar and CalendarPattern data types allow you to specify an arbitrarily complex pattern of when entries are accepted. For regular time series, the calendar additionally creates a vector from which to calculate an element's position, so that an element's timestamp does not have to be stored, or even specified.

The IBM Informix TimeSeries DataBlade module provides a wide variety of routines to manage and manipulate time series data. The routines allow you to manipulate columns in an element, an element itself, a portion of a time series, a whole time series, or many time series. In addition, there are routines to manage calendars, containers, and metadata. Most routines are implemented in both an SQL interface and an API library.

The IBM Informix TimeSeries DataBlade module also includes system tables to record information about calendars, time series, and containers.