guileen/node-rts

Redis time series data store

rts

Use redis as time series data store.

• Data for the line chart of a period time, can be use to stat sum/avg/max/min/count of a behavior.

e.g. Every 5 minutes, sum of user click of the website.

e.g. Every Day, average consume money of the website.

• Aggregate data of DayOfWeek or HoursOfDay, to analytisic behavior.

e.g. The sum of click from Monday to SunDay.

Init

var rts = require('rts')(options);

var rts = require('rts')({
        redis: redis,
        gran: '5m, 1h, 1d, 1w',
        points: 500,
        prefix: ''
});

Options:

• redis the redis client
• gran granularity of recored time. Format is '{number}{unit}, {number}{unit}...'. The unit can be s second, m minute, h hour, d day, w week, M month, y year. e.g. 5m, 1h, store data for 5 mintues and 1 hour.
• points how many data will be keep. We care about recently n points of data for each granularity. Think about 1s data, if store 1 day data, it is 86400 rows data, we should reduce it to 5m data, it is only 288 rows. We just care small granularity data for recently, for the history, we just care large granualrity, like revenue of 2010-11, not revenue of 2010-11-05 12 o'clock.
• prefix the redis key prefix.

Record

record(behavior, [behaviorValue], [statistics], [aggregations])

• behavior
• behaviorValue
• statistics: Array of statistic type, avariable member of sum, count, avg, max, min. default [sum].
• aggragations
  • hm hour of day, splite data every month
  • hq hour of day, splite data every season
  • hy hour of day, splite data every year
  • dm day of week, splite data every month.
  • dq day of week, splite data every season
  • dy day of week, splite data every year

ts.record('click') // recore a click
ts.record('consume', 5, null, ['hm', 'dq']) // recored behavior with value, and aggragation. 
ts.record('delay', 100, ['avg','max','min'])

Query

ts.getStat(statistic, behavior, granularity, fromDate, toDate, callback)

ts[statistic](behavior, granularity, fromDate, toDate, callback)

ts.aggr{statistic}(behavior, aggragation, date, callback)

• statistic sum, avg, count, max, min. Must be a member when statistics when record!
• behavior
• granularity The time unit, see Options, Must be a part of options.gran
• fromDate
• toDate

ts.sum('click', '5m', fromDate, \[toDate\], callback)
ts.count('delay', '5m', fromDate, \[toDate\], callback)
ts.avg('delay', '5m', fromDate, \[toDate\], callback)
ts.max('delay', '5m', fromDate, \[toDate\], callback)
ts.min('delay', '5m', fromDate, \[toDate\], callback)

ts.aggravg('click', 'hm', date, callback)

Others

ts.setValue(name, stat, value, gran, timestamp, callback)

ts.getTimePeriord(gran, points)