A label in a metric is a description of this metric, including the key of this description (what the description is), and the value for this description.
List metric descriptors that match the query. If the query is not set, then all of the metric descriptors will be returned. Large responses will be paginated, use the nextPageToken returned in the response to request subsequent pages of results by setting the pageToken query parameter to the value of the nextPageToken.
Distribution data point value type. When writing distribution points, try to be consistent with the boundaries of your buckets. If you must modify the bucket boundaries, then do so by merging, partitioning, or appending rather than skewing them.
The histogram’s bucket. Buckets that form the histogram of a distribution value. If the upper bound of a bucket, say U1, does not equal the lower bound of the next bucket, say L2, this means that there is no event in [U1, L2).
The monitoring data is organized as metrics and stored as data points that are recorded over time. Each data point represents information like the CPU utilization of your virtual machine. A historical record of these data points is called a time series.
List the descriptors of the time series that match the metric and labels values and that have data points in the interval. Large responses are paginated; use the nextPageToken returned in the response to request subsequent pages of results by setting the pageToken query parameter to the value of the nextPageToken.
When writing time series, TimeseriesPoint should be used instead of Timeseries, to enforce single point for each time series in the timeseries.write request.
List the data points of the time series that match the metric and labels values and that have data points in the interval. Large responses are paginated; use the nextPageToken returned in the response to request subsequent pages of results by setting the pageToken query parameter to the value of the nextPageToken.
Put data points to one or more time series for one or more metrics. If a time series does not exist, a new time series will be created. It is not allowed to write a time series point that is older than the existing youngest point of that time series. Points that are older than the existing youngest point of that time series will be discarded silently. Therefore, users should make sure that points of a time series are written sequentially in the order of their end time.