TSDB: Time Series Database. It is a data management system featuring efficient time series data writing and reading and statistical analysis.
Time Series Data: a series of metric data that is generated continuously at a set interval, for example, the data points collected per second during the air monitoring.
Metric: a device measurement indicator, such as wind speed or temperature.
Although metric specifies which indicator to monitor, it does not sepcifcy which object the indicator belongs to. To sepecify the object, use the subcategory, tag.
A tag composes of a TagKey and a TagValue, for example, “city (TagKey) = Hangzhou (TagValue)”. More examples: data center = A, IP = 22.214.171.124.
Note: Two tags can be same only when both their TagKeys and TagValues are identical; if the TagKeys are same, but the TagValues are different, then they are two different tags.
Specifing that the metric is “Temperature”, and the tag “city = Hangzhou” means to monitor the temperature of Hangzhou.
Tagk: TagKey, the category of the monitored object specified by the metric (the specific object of the category that is located by a corresponding TagValue), for example, country, province, city, data center, or IP.
Tagv: TagValue, the corresponding value of TagKey, for example. when TagKey is “country”, TagValue can be “China”.
Value: the value of a metric, for example, Level 15 (wind) and 20℃ (temperature).
Timestamp: the time point when data (metrics) are generated.
Data Point: is the every metric value collected at a particular interval (continuous timestamps) for a particualr indicator (defined by metrics and tags) of the monitored object. A data point is defined as “1 Metric + N Tags (N >= 1) + 1 Timestamp + 1 Value”.
Time Series: a particualr indicator (defined by metrics and tags) of the monitored object. A time series is defined as “1 Metric + N Tags or key-value pairs (N >= 1)”. The accumulation of values in a time series does not result in the increase of time series. The time series diagram is as follows:
Timeline: same as Time series.
Time Precision: the precision of the writing time of the time series data, such as millisecond, second, minute, hour. For example, one temperature data point is collected every one second or one load CPU usage is collected every five minutes.
Data Group: used to divide data into groups by tag value, in order to compare data that is of the same metric but from differenct devices. For example, group the temperature data (metric) by city (tag value). It is similar to the SQL: select temperature from xxx group by city where city in (Shanghai, Hangzhou).
Space Aggregation: a calculation process in which data of multiple dimensions in a space is aggregated into one timeline when multiple timelines (multiple data collecting devices) are generated for one metric query. For example, in order to show the pollution index of a city distric, the average value of pollution indicators at all monitoring points is calculated, and the calculation is called space aggregation.
Downsampling: when the queried time range is wide and the raw data’s time precision is high, in order to meet certain business demand and improve the query efficiency, you need to reduce the dispaly precision of the queried data. This is called downsampling. For example, when you query a year’s data collected at a time interval of one second, after downsampling, the data is displayed by day.
Data’s Validity Period: the validity period specified for the data. The data that exceeds the validity period is expired and automatically released.