Rootconf 2017

This talk will focus on TICK stack and comparision between other open source monitoring tools. In order to build the monitoring architecture of such large well growing and stable infrastructure we should build or choose monitoring tool/stack which suits perfect in the terms of cost which is one of obvious reason and performance. So TICK stack is one of the best stack for monitoring, build by influxData’s guys. This talk will cover all baics understanding of internal components and other features like it provides lots of inbuilt functionality to integrate with common tools and flexible to add new integration based on business requirement with lots of additional functionality like helps on scaling, fast to build full monitoring system, full support of community.

Outline

• Basic introduction of system and application monitoring

  • What kind of data should we collect so that it can be easily help us on analyse long term trends,
    alerting, building dashboard, ad-hoc retrospective analysis(production issue debugging).

  • Categories metrics into 2 part:

    • Work Metrics(mainly application metrics)
    • Resource Metrics(mainly system metrics)

  • Key points when capturing monitoring data: well understood, Granular, tagged by scope, Long-lived.

  • Monitoring Architecture

    • Pull Based: A central collector periodically requests metrics from each monitored system.
    • Push Based: Metrics are periodically sent by each monitored system to a central collector.
    • Explain both the architectures with example: Prometheus For pull based and TICK for Push based.
      ** Will Explaining the pros and cos based on following parameters:
      1. Discovery
      2. Scalability
      3. Security
      4. Operation Complexity
      5. Latency
      6. Flexibility
      7. Use Cases

  • Internal Component of TICK

    • Telegraf: * How it works ?
      * Data collection, process, and aggregate works ?
      * which protocal used to send data and why ?
      * Output plugin support and internal component.
    • InfluxDb: * Key Concepts of influxdb.
      * benchmark with other data storage.
      * Schema Design, Comparison to SQL
      * Design Insights and Tradeoffs in InfluxDB
      * Advantages of using influxDb.
    • Chronograf: * Intro of chronograf and grafana.
    • Kapacitor: * Introduction of tickscript.
      * writing alerts integrations
      * External integration plugins available.
    • Monitor all layers of your stack together, so that you can see what is happening everywhere, at the same time, with no gaps.

Speaker bio

Praveen is a System Engineer at gojek, who love exploring/learning things specially on monitoring and centralized logging.

Slides

https://speakerdeck.com/achilles/tick-monitoring-1