分布式时序计算框架vortex|分布式时序计算框架vortex metrics使用介绍

vortex metrics是一款用Java写的,轻量级的分布式时序计算框架。
vortex metrics依赖分布式流式计算框架vortex, 它是vortex关于时序计算场景的一个具体应用,vortex更像是一套API, 运用这套API可以演绎和解决各种业务场景,而vortex metrics则是根据这套API提出了对时序计算业务场景一种解决方案。
介绍之前,先比较另外一个相似的时序计算工具OpenTSDB, OpenTSDB的存储依赖HBase, 它更像一个时序数据库。OpenTSDB通过TCollector和内置的collectors接收数据,并提供了界面供用户查询。
相比之下,vortex metrics更偏向实时计算,将延迟尽可能地保持在秒级,但由于是内存计算,计算结果是存储在内存中的,有限的存储空间让它并不适用于存储大量指标的业务场景。但同时,对于一个指标的窗口数据需要保留的时间跨度较长时,vortex metrics提供了定制接口,可以保存到外部存储,比如数据库或缓存,但这需要做一些额外的开发工作。此外,vortex metrics也提供了Web查询界面,并且在持续改进中。
实际上,vortex metrics的功能是很单一的,因为它只做一件事,那就是根据客户端源源不断的上报数据实时地计算并输出单位时间窗口内的统计数据,记住这点很重要。
环境配置 Maven:

com.github.paganini2008.atlantis vortex-spring-boot-starter 1.0-RC1

vortex metrics它首先是一个基于SpringBoot的应用程序,前面说过,它依赖vortex框架, 而vortex是通过微服务分布式协调框架tridenter实现集群模式的,并支持动态水平扩展,默认端口为6150
参考配置
# 集群配置 spring.application.name=vortex-metrics spring.application.cluster.name=vortex-metrics-cluster# vortex配置 atlantis.framework.vortex.bufferzone.collectionName=metric# 缓冲区集合名称 atlantis.framework.vortex.bufferzone.pullSize=1000# 缓冲区每次拉取的消息数量 atlantis.framework.vortex.processor.threads=200# 数据消费者线程数

输入地址:http://localhost:6150/metric/ 可以看到这个界面
分布式时序计算框架vortex|分布式时序计算框架vortex metrics使用介绍
文章图片

vortex metrics是将server数据接收端和监控界面做在同一个应用中的,高并发场景下,可能会出现数据计算完成并已同步,但界面会滞后更新的情况。
HTTP API说明 vortex metrics目前只提供了2个接口:
  1. 上报数据接口:
    POSThttp://localhost:6150/metrics/sequence/{dataType}
    请求体示例:
    { "name": "car", "metric": "speed", "value": 120, "timestamp": 1613200609281 }

    参数说明:
    参数名 类型 描述 举例
    dateType 字符串 数据类型,取值范围:bigint,numeric,bool bigint
    name 字符串 应用名称 car
    metric 字符串 上报指标名称 speed
    value 数值 上报数值 120
    timestamp 长整型 上报时间戳 1613200609281
  2. 查看数据接口:
    GEThttp://localhost:6150/metrics/sequence/{dataType}/{name}/{metric}
    响应体示例:
    { "dataType": "bigint", "name": "car", "metric": "speed", "data": { "00:03:00": { "speed": { "middleValue": 5052, "count": 155796, "highestValue": 9999, "lowestValue": 100, "timestamp": 1623513794876 } }, "00:04:00": { "speed": { "middleValue": 5051, "count": 79754, "highestValue": 9999, "lowestValue": 100, "timestamp": 1623513840003 } }, "00:05:00": { "speed": { "middleValue": 5039, "count": 61014, "highestValue": 9999, "lowestValue": 100, "timestamp": 1623513933165 } }, "00:06:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623513960876 } }, "00:07:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514020876 } }, "00:08:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514080876 } }, "00:09:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514140876 } }, "00:10:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514200876 } }, "00:11:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514260876 } }, "00:12:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514320876 } }, "00:13:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514380876 } }, "00:14:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514440876 } }, "00:15:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514500876 } }, "00:16:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514560876 } }, "00:17:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514620876 } }, "00:18:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514680876 } }, "00:19:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514740876 } }, "00:20:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514800876 } }, "00:21:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514860876 } }, "00:22:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514920876 } }, "00:23:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623514980876 } }, "00:24:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515040876 } }, "00:25:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515100876 } }, "00:26:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515160876 } }, "00:27:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515220876 } }, "00:28:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515280876 } }, "00:29:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515340876 } }, "00:30:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515400876 } }, "00:31:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515460876 } }, "00:32:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515520876 } }, "00:33:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515580876 } }, "00:34:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515640876 } }, "00:35:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515700876 } }, "00:36:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515760876 } }, "00:37:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515820876 } }, "00:38:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515880876 } }, "00:39:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623515940876 } }, "00:40:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516000876 } }, "00:41:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516060876 } }, "00:42:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516120876 } }, "00:43:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516180876 } }, "00:44:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516240876 } }, "00:45:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516300876 } }, "00:46:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516360876 } }, "00:47:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516420876 } }, "00:48:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516480876 } }, "00:49:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516540876 } }, "00:50:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516600876 } }, "00:51:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516660876 } }, "00:52:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516720876 } }, "00:53:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516780876 } }, "00:54:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516840876 } }, "00:55:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516900876 } }, "00:56:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623516960876 } }, "00:57:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623517020876 } }, "00:58:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623517080876 } }, "00:59:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623517140876 } }, "01:00:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623517200876 } }, "01:01:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623517260876 } }, "01:02:00": { "speed": { "middleValue": 0, "count": 0, "highestValue": 0, "lowestValue": 0, "timestamp": 1623517320876 } } } }

    响应体json中的data为时序数据,vortex metrics默认的统计窗口为1分钟,滚动保留前60条记录,也就是用户可以通过界面看到前60分钟的统计数据。其中,highestValue,lowestValue,middleValue分别为最大值,最小值,平均值,count是 这1分钟内的数据条数,timestamp是时间戳,准确点讲,是这1分钟内最后一条数据的时间戳。
    vortex metrics默认的统计窗口,保留前多少条记录,可以通过配置文件设置的,但系统需要重启才能生效。
如何监控指标 上图中的location, 输入查询数据接口地址,比如举个上面的例子,统计小客车的平均速度,接口应该是,http://localhost:6150/metrics/sequence/bigint/car/speed
将这个ur复制到location输入框,点击search, 可以看到:
分布式时序计算框架vortex|分布式时序计算框架vortex metrics使用介绍
文章图片

此时是没数据的
vortex metrics内置了3个测试接口, 都是GET方式
  1. http://localhost:6150/metrics/test/bigint/{name}/{metric}
  2. http://localhost:6150/metrics/test/numeric/{name}/{metric}
  3. http://localhost:6150/metrics/test/bool/{name}/{metric}
    这里为了测试vortex metrics集群性能,我起了两台服务,端口分别为6150和6151
然后,打开你的压测工具,这里以JMeter为例
设置线程组
分布式时序计算框架vortex|分布式时序计算框架vortex metrics使用介绍
文章图片

设置HTTP请求
分布式时序计算框架vortex|分布式时序计算框架vortex metrics使用介绍
文章图片

观察压测结果:
分布式时序计算框架vortex|分布式时序计算框架vortex metrics使用介绍
文章图片

分布式时序计算框架vortex|分布式时序计算框架vortex metrics使用介绍
文章图片

可以看到,实时统计1分钟内的最大值,最小值,平均值和个数,延迟在秒级
前面说过,vortex metrics是内存计算,计算结果驻留在内存中的,如果存在大量指标数据,可能会出现延时,所以它不适用于存储大量指标的运算场景(不过你可以部署多个集群来解决这个问题),目前只测算过指标在1000个以内的场景,高并发下延时保持在1~3分钟内。
如何保证数据一致性 vortex metrics集群的节点通过内部网络通讯实现相互复制,保证数据最终一致性,即每个节点的数据最终都是一样的,全量的。
也就是说:
http://localhost:6150/metrics/sequence/bigint/car/speed
http://localhost:6151/metrics/sequence/bigint/car/speed
两个接口返回的数据是最终一致的
如何保存历史数据 前面说到,vortex metrics默认的统计窗口为1分钟,滚动保留前60条记录,那如果你想保留之前的历史记录,需要另外做开发。
首先要实现接口:
public interface MetricEvictionHandler> {void onEldestMetricRemoval(I identifier, String metric, T metricUnit); }

具体可参考:LoggingMetricEvictionHandler类(默认实现)
你还要实现接口:
public interface MetricSequencerFactory {GenericUserMetricSequencer getBigIntMetricSequencer(); GenericUserMetricSequencer getNumericMetricSequencer(); GenericUserMetricSequencer getBoolMetricSequencer(); }

【分布式时序计算框架vortex|分布式时序计算框架vortex metrics使用介绍】具体可参考:DefaultMetricSequencerFactory类(默认实现)

    推荐阅读