开源分布式数据库InfluxDB介绍

jieforest · 发表于 2014-5-18 09:17

InfluxDB 是一个开源，分布式，时间序列，事件，可度量和无外部依赖的数据库。

InfluxDB有三大特性：
1. Time Series （时间序列）：你可以使用与时间有关的相关函数（如最大，最小，求和等）
2. Metrics（度量）：你可以实时对大量数据进行计算
3. Eevents（事件）：它支持任意的事件数据

HTTP API：

influxdb.writePoints("user_events", [{email: "paul@influxdb.org", state: "NY", type: "click"}]);

复制代码

jieforest · 发表于 2014-5-18 09:17

查询语句：

series = influxdb.query(
"select percentile(value, 90) from response_times " +
"group by time(1h) where time > now() - 2d;");
series = influxdb.query(
"select count(email) from events " +
"group by time(10m) " +
"where email =~ /.*gmail\.com/ and time > now() - 1d");

复制代码

内置浏览器：你可以通过自带的浏览工具查看你的数据。
适用于个性化分析：它适合于各种个性化分析。

jieforest · 发表于 2014-5-18 09:18

Overview
InfluxDB is a time series, events, and metrics database. It's written in Go and has no external dependencies. That means once you install it there's nothing else to manage (like Redis, HBase, or whatever). It's designed to be distributed and scale horizontally, but be useful even if you're only running it on a single box. There are three of us ( Paul Dix, Todd Persen, and John Shahid ) working full time on the core of the database. It's the result of a "pivot" from Errplane (YC W13).

If you're looking to just start writing some code to play around with the API, head over to the getting started section of the documentation.

Design Goals
Here are some goals we're targeting while building InfluxDB.

Stores metrics data (like response times and cpu load. e.g. what you'd put into Graphite).
Stores events data (like exceptions, user analytics, or business analytics).
HTTP(S) interface for reading and writing data. Shouldn't require additional server code to be useful directly from the browser.
Security model that will enable user facing analytics dashboards connecting directly to the HTTP API.
Horizontally scalable.
On disk and in memory. It shouldn't require a cluster of machines keeping everything in memory since most analytics data is cold most of the time.
Simple to install and manage. Shouldn't require setting up external dependencies like Zookeeper and Hadoop.
Ability to compute percentiles and other functions on the fly.
Ability to automatically downsample data on different windows of time.
Can efficiently and automatically clear out raw data daily to free up space. However, some time series will keep data forever, while others will keep only a rolling window (like the last 30 days)
Should be able to add on custom real-time and batch analytics processing.
Pubsub interface
Ability to automatically expand storage by adding nodes to a cluster. Should make node replacement quick and easy.
Highly elastic to expand computing capacity (compute nodes should scale up and down quickly)

jieforest · 发表于 2014-5-18 09:18

Why?
When we built Errplane, we wanted the data model to be flexible enough to store events like exceptions along with more traditional metrics like response times and server stats. At the same time we noticed that other companies were also building custom time series APIs on top of different databases for analytics and metrics. Depending on the requirements these APIs would be built on top of a regular SQL database, Redis, HBase, or Cassandra.

We thought the community might benefit from the work we'd already done with our scalable backend. We wanted something that had the HTTP API built in that would scale out to billions of metrics or events. We also wanted something that would make it simple to query for downsampled data, percentiles, and other aggregates at scale. Our hope is that once there's a standard API, the community will be able to build useful tooling around it for data collection, visualization, and analysis.