statistics (instance statistics)

The statistics resource contains information regarding how AMPS monitors its own statistics.

Element	Description	Type
disk_per_sample	Current size in bytes of the main statistics database file divided by total_samples. This is not the growth since the preceding sample.	snapshot
file_name	Location where statistics are stored. Default: :memory: which stores the statistics database in system memory.	fixed
file_size	Size on disk of the statistics database.	snapshot
interval	Time in milliseconds between statistics database updates.	fixed
memory_used	Size in bytes of the system memory consumption of the statistics database.	snapshot
queries	Number of requests handled by the basic monitoring interface since this instance started, including requests that fail to acquire the statistics database lock. This counts monitoring requests, rather than individual SQL statements.	cumulative
time_per_sample	Average time in seconds spent collecting and committing statistics per recorded sample: total_time / total_samples.	running average
total_commit_time	Accumulated time in seconds spent committing statistics information to the database. Recovered from the most recent persisted sample on restart.	recovered cumulative
total_samples	Accumulated number of statistics samples, including the count recovered from the most recent persisted sample on restart.	recovered cumulative
total_time	Accumulated time in seconds spent collecting and publishing statistics, including commit time. Recovered from the most recent persisted sample on restart.	recovered cumulative

The total_time and total_commit_time values include completed collection intervals through the preceding sample. The current collection cannot include its own full duration in the record being written. Consequently, time_per_sample uses timing totals that lag the sample count by one interval.

When AMPS restarts with a persisted statistics database, total_time, total_commit_time, and total_samples continue from the latest stored values. The averages calculated from these totals can therefore span multiple instance lifetimes. With a new statistics database, these totals start from zero.

The time_per_sample and disk_per_sample metrics are calculated when queried through the administrative interface; they are not stored columns in the statistics database. For the SQL expressions and source table, see Computed Metrics.

Statistics Database Tables: ISTATISTICS_STATIC, ISTATISTICS_DYNAMIC

Admin Path: /amps/instance/statistics/<metric>