summaryrefslogtreecommitdiffstats
path: root/docs/guides/configure/performance.md
diff options
context:
space:
mode:
Diffstat (limited to 'docs/guides/configure/performance.md')
-rw-r--r--docs/guides/configure/performance.md269
1 files changed, 108 insertions, 161 deletions
diff --git a/docs/guides/configure/performance.md b/docs/guides/configure/performance.md
index 256d6e854..2e5e105fe 100644
--- a/docs/guides/configure/performance.md
+++ b/docs/guides/configure/performance.md
@@ -1,110 +1,101 @@
-<!--
-title: How to optimize the Netdata Agent's performance
-description: "While the Netdata Agent is designed to monitor a system with only 1% CPU, you can optimize its performance for low-resource systems."
-image: /img/seo/guides/configure/performance.png
-custom_edit_url: https://github.com/netdata/netdata/edit/master/docs/guides/configure/performance.md
--->
-
# How to optimize the Netdata Agent's performance
We designed the Netdata Agent to be incredibly lightweight, even when it's collecting a few thousand dimensions every
-second and visualizing that data into hundreds of charts. When properly configured for a production node, the Agent
-itself should never use more than 1% of a single CPU core, roughly 50-100 MiB of RAM, and minimal disk I/O to collect,
-store, and visualize all this data.
-
-We take this scalability seriously. We have one user [running
-Netdata](https://github.com/netdata/netdata/issues/1323#issuecomment-266427841) on a system with 144 cores and 288
-threads. Despite collecting 100,000 metrics every second, the Agent still only uses 9% CPU utilization on a
-single core.
-
-But not everyone has such powerful systems at their disposal. For example, you might run the Agent on a cloud VM with
-only 512 MiB of RAM, or an IoT device like a [Raspberry Pi](https://github.com/netdata/netdata/blob/master/docs/guides/monitor/pi-hole-raspberry-pi.md). In these
-cases, reducing Netdata's footprint beyond its already diminutive size can pay big dividends, giving your services more
-horsepower while still monitoring the health and the performance of the node, OS, hardware, and applications.
+second and visualizing that data into hundreds of charts. However, the default settings of the Netdata Agent are not
+optimized for performance, but for a simple, standalone setup. We want the first install to give you something you can
+run without any configuration. Most of the settings and options are enabled, since we want you to experience the full thing.
-The default settings of the Netdata Agent are not optimized for performance, but for a simple standalone setup. We want
-the first install to give you something you can run without any configuration. Most of the settings and options are
-enabled, since we want you to experience the full thing.
+By default, Netdata will automatically detect applications running on the node it is installed to start collecting metrics in
+real-time, has health monitoring enabled to evaluate alerts and trains Machine Learning (ML) models for each metric, to detect anomalies.
+This document describes the resources required for the various default capabilities and the strategies to optimize Netdata for production use.
-## Prerequisites
+## Summary of performance optimizations
-- A node running the Netdata Agent.
-- Familiarity with configuring the Netdata Agent with `edit-config`.
+The following table summarizes the effect of each optimization on the CPU, RAM and Disk IO utilization in production.
-If you're not familiar with how to configure the Netdata Agent, read our [node configuration
-doc](https://github.com/netdata/netdata/blob/master/docs/configure/nodes.md) before continuing with this guide. This guide assumes familiarity with the Netdata config
-directory, using `edit-config`, and the process of uncommenting/editing various settings in `netdata.conf` and other
-configuration files.
+Optimization | CPU | RAM | Disk IO
+-- | -- | -- |--
+[Use streaming and replication](#use-streaming-and-replication) | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark:
+[Disable unneeded plugins or collectors](#disable-unneeded-plugins-or-collectors) | :heavy_check_mark: | :heavy_check_mark: | :heavy_check_mark:
+[Reduce data collection frequency](#reduce-collection-frequency) | :heavy_check_mark: | | :heavy_check_mark:
+[Change how long Netdata stores metrics](https://github.com/netdata/netdata/blob/master/docs/store/change-metrics-storage.md) | | :heavy_check_mark: | :heavy_check_mark:
+[Use a different metric storage database](https://github.com/netdata/netdata/blob/master/database/README.md) | | :heavy_check_mark: | :heavy_check_mark:
+[Disable machine learning](#disable-machine-learning) | :heavy_check_mark: | |
+[Use a reverse proxy](#run-netdata-behind-a-proxy) | :heavy_check_mark: | |
+[Disable/lower gzip compression for the agent dashboard](#disablelower-gzip-compression-for-the-dashboard) | :heavy_check_mark: | |
-## What affects Netdata's performance?
+## Resources required by a default Netdata installation
Netdata's performance is primarily affected by **data collection/retention** and **clients accessing data**.
-You can configure almost all aspects of data collection/retention, and certain aspects of clients accessing data. For
-example, you can't control how many users might be viewing a local Agent dashboard, [viewing an
-infrastructure](https://github.com/netdata/netdata/blob/master/docs/visualize/overview-infrastructure.md) in real-time with Netdata Cloud, or running [Metric
-Correlations](https://github.com/netdata/netdata/blob/master/docs/cloud/insights/metric-correlations.md).
+You can configure almost all aspects of data collection/retention, and certain aspects of clients accessing data.
+
+### CPU consumption
+
+Expect about:
+ - 1-3% of a single core for the netdata core
+ - 1-3% of a single core for the various collectors (e.g. go.d.plugin, apps.plugin)
+ - 5-10% of a single core, when ML training runs
-The Netdata Agent runs with the lowest possible [process scheduling
-policy](https://github.com/netdata/netdata/blob/master/daemon/README.md#netdata-process-scheduling-policy), which is `nice 19`, and uses the `idle` process scheduler.
+Your experience may vary depending on the number of metrics collected, the collectors enabled and the specific environment they
+run on, i.e. the work they have to do to collect these metrics.
+
+As a general rule, for modern hardware and VMs, the total CPU consumption of a standalone Netdata installation, including all its components,
+should be below 5 - 15% of a single core. For example, on 8 core server it will use only 0.6% - 1.8% of a total CPU capacity, depending on
+the CPU characteristics.
+
+The Netdata Agent runs with the lowest possible [process scheduling policy](https://github.com/netdata/netdata/blob/master/daemon/README.md#netdata-process-scheduling-policy), which is `nice 19`, and uses the `idle` process scheduler.
Together, these settings ensure that the Agent only gets CPU resources when the node has CPU resources to space. If the
node reaches 100% CPU utilization, the Agent is stopped first to ensure your applications get any available resources.
-In addition, under heavy load, collectors that require disk I/O may stop and show gaps in charts.
-Let's walk through the best ways to improve the Netdata Agent's performance.
+To reduce CPU usage you can [disable machine learning](#disable-machine-learning),
+[use streaming and replication](#use-streaming-and-replication),
+[reduce the data collection frequency](#reduce-collection-frequency), [disable unneeded plugins or collectors](#disable-unneeded-plugins-or-collectors), [use a reverse proxy](#run-netdata-behind-a-proxy), and [disable/lower gzip compression for the agent dashboard](#disablelower-gzip-compression-for-the-dashboard).
-## Reduce collection frequency
+### Memory consumption
-The fastest way to improve the Agent's resource utilization is to reduce how often it collects metrics.
+The memory footprint of Netdata is mainly influenced by the number of metrics concurrently being collected. Expect about 150MB of RAM for a typical 64-bit server collecting about 2000 to 3000 metrics.
-### Global
+To estimate and control memory consumption, you can [disable unneeded plugins or collectors](#disable-unneeded-plugins-or-collectors), [change how long Netdata stores metrics](https://github.com/netdata/netdata/blob/master/docs/store/change-metrics-storage.md), or [use a different metric storage database](https://github.com/netdata/netdata/blob/master/database/README.md).
-If you don't need per-second metrics, or if the Netdata Agent uses a lot of CPU even when no one is viewing that node's
-dashboard, configure the Agent to collect metrics less often.
-Open `netdata.conf` and edit the `update every` setting. The default is `1`, meaning that the Agent collects metrics
-every second.
+### Disk footprint and I/O
-If you change this to `2`, Netdata enforces a minimum `update every` setting of 2 seconds, and collects metrics every
-other second, which will effectively halve CPU utilization. Set this to `5` or `10` to collect metrics every 5 or 10
-seconds, respectively.
+By default, Netdata should not use more than 1GB of disk space, most of which is dedicated for storing metric data and metadata. For typical installations collecting 2000 - 3000 metrics, this storage should provide a few days of high-resolution retention (per second), about a month of mid-resolution retention (per minute) and more than a year of low-resolution retention (per hour).
-```conf
-[global]
- update every = 5
-```
+Netdata spreads I/O operations across time. For typical standalone installations there should be a few write operations every 5-10 seconds of a few kilobytes each, occasionally up to 1MB. In addition, under heavy load, collectors that require disk I/O may stop and show gaps in charts.
-### Specific plugin or collector
+To configure retention, you can [change how long Netdata stores metrics](https://github.com/netdata/netdata/blob/master/docs/store/change-metrics-storage.md).
+To control disk I/O [use a different metric storage database](https://github.com/netdata/netdata/blob/master/database/README.md), avoid querying the
+production system [using streaming and replication](#use-streaming-and-replication), [reduce the data collection frequency](#reduce-collection-frequency), and [disable unneeded plugins or collectors](#disable-unneeded-plugins-or-collectors).
-Every collector and plugin has its own `update every` setting, which you can also change in the `go.d.conf`,
-`python.d.conf`, or `charts.d.conf` files, or in individual collector configuration files. If the `update
-every` for an individual collector is less than the global, the Netdata Agent uses the global setting. See the [enable
-or configure a collector](https://github.com/netdata/netdata/blob/master/docs/collect/enable-configure.md) doc for details.
+## Use streaming and replication
-To reduce the frequency of an [internal
-plugin/collector](https://github.com/netdata/netdata/blob/master/docs/collect/how-collectors-work.md#collector-architecture-and-terminology), open `netdata.conf` and
-find the appropriate section. For example, to reduce the frequency of the `apps` plugin, which collects and visualizes
-metrics on application resource utilization:
+For all production environments, parent Netdata nodes outside the production infrastructure should be receiving all
+collected data from children Netdata nodes running on the production infrastructure, using [streaming and replication](https://github.com/netdata/netdata/blob/master/docs/metrics-storage-management/enable-streaming.md).
-```conf
-[plugin:apps]
- update every = 5
-```
+### Disable health checks on the child nodes
-To [configure an individual collector](https://github.com/netdata/netdata/blob/master/docs/collect/enable-configure.md), open its specific configuration file with
-`edit-config` and look for the `update_every` setting. For example, to reduce the frequency of the `nginx` collector,
-run `sudo ./edit-config go.d/nginx.conf`:
+When you set up streaming, we recommend you run your health checks on the parent. This saves resources on the children
+and makes it easier to configure or disable alerts and agent notifications.
+
+The parents by default run health checks for each child, as long as the child is connected (the details are in `stream.conf`).
+On the child nodes you should add to `netdata.conf` the following:
```conf
-# [ GLOBAL ]
-update_every: 10
+[health]
+ enabled = no
```
+### Use memory mode ram or save for the child nodes
+
+See [using a different metric storage database](https://github.com/netdata/netdata/blob/master/database/README.md).
+
## Disable unneeded plugins or collectors
If you know that you don't need an [entire plugin or a specific
-collector](https://github.com/netdata/netdata/blob/master/docs/collect/how-collectors-work.md#collector-architecture-and-terminology), you can disable any of them.
+collector](https://github.com/netdata/netdata/blob/master/collectors/README.md#collector-architecture-and-terminology), you can disable any of them.
Keep in mind that if a plugin/collector has nothing to do, it simply shuts down and does not consume system resources.
You will only improve the Agent's performance by disabling plugins/collectors that are actively collecting metrics.
@@ -137,42 +128,60 @@ modules:
fail2ban: no
```
-## Lower memory usage for metrics retention
+## Reduce collection frequency
+
+The fastest way to improve the Agent's resource utilization is to reduce how often it collects metrics.
+
+### Global
+
+If you don't need per-second metrics, or if the Netdata Agent uses a lot of CPU even when no one is viewing that node's
+dashboard, [configure the Agent](https://github.com/netdata/netdata/blob/master/docs/configure/nodes.md) to collect metrics less often.
+
+Open `netdata.conf` and edit the `update every` setting. The default is `1`, meaning that the Agent collects metrics
+every second.
+
+If you change this to `2`, Netdata enforces a minimum `update every` setting of 2 seconds, and collects metrics every
+other second, which will effectively halve CPU utilization. Set this to `5` or `10` to collect metrics every 5 or 10
+seconds, respectively.
-Reduce the disk space that the [database engine](https://github.com/netdata/netdata/blob/master/database/engine/README.md) uses to retain metrics by editing
-the `dbengine multihost disk space` option in `netdata.conf`. The default value is `256`, but can be set to a minimum of
-`64`. By reducing the disk space allocation, Netdata also needs to store less metadata in the node's memory.
+```conf
+[global]
+ update every = 5
+```
-The `page cache size` option also directly impacts Netdata's memory usage, but has a minimum value of `32`.
+### Specific plugin or collector
-Reducing the value of `dbengine multihost disk space` does slim down Netdata's resource usage, but it also reduces how
-long Netdata retains metrics. Find the right balance of performance and metrics retention by using the [dbengine
-calculator](https://github.com/netdata/netdata/blob/master/docs/store/change-metrics-storage.md#calculate-the-system-resources-ram-disk-space-needed-to-store-metrics).
+Every collector and plugin has its own `update every` setting, which you can also change in the `go.d.conf`,
+`python.d.conf`, or `charts.d.conf` files, or in individual collector configuration files. If the `update
+every` for an individual collector is less than the global, the Netdata Agent uses the global setting. See the [collectors configuration reference](https://github.com/netdata/netdata/blob/master/collectors/REFERENCE.md) for details.
-All the settings are found in the `[global]` section of `netdata.conf`:
+To reduce the frequency of an [internal
+plugin/collector](https://github.com/netdata/netdata/blob/master/collectors/README.md#collector-architecture-and-terminology), open `netdata.conf` and
+find the appropriate section. For example, to reduce the frequency of the `apps` plugin, which collects and visualizes
+metrics on application resource utilization:
```conf
-[db]
- memory mode = dbengine
- page cache size = 32
- dbengine multihost disk space = 256
+[plugin:apps]
+ update every = 5
```
-To save even more memory, you can disable the dbengine and reduce retention to just 30 minutes, as shown below:
+To [configure an individual collector](https://github.com/netdata/netdata/blob/master/collectors/REFERENCE.md#configure-a-collector), open its specific configuration file with
+`edit-config` and look for the `update_every` setting. For example, to reduce the frequency of the `nginx` collector,
+run `sudo ./edit-config go.d/nginx.conf`:
```conf
-[db]
- storage tiers = 1
- mode = alloc
- retention = 1800
+# [ GLOBAL ]
+update_every: 10
```
-Metric retention is not important in certain use cases, such as:
- - Data collection nodes stream collected metrics collected to a centralization point.
- - Data collection nodes export their metrics to another time series DB, or are scraped by Prometheus
- - Netdata installed only during incidents, to get richer information.
-In such cases, you may not want to use the dbengine at all and instead opt for memory mode
-`memory mode = alloc` or `memory mode = none`.
+## Lower memory usage for metrics retention
+
+See how to [change how long Netdata stores metrics](https://github.com/netdata/netdata/blob/master/docs/store/change-metrics-storage.md).
+
+## Use a different metric storage database
+
+Consider [using a different metric storage database](https://github.com/netdata/netdata/blob/master/database/README.md) when running Netdata on IoT devices,
+and for children in a parent-child set up based on [streaming and replication](https://github.com/netdata/netdata/blob/master/docs/metrics-storage-management/enable-streaming.md).
## Disable machine learning
@@ -185,34 +194,12 @@ with the following:
enabled = no
```
-## Run Netdata behind Nginx
+## Run Netdata behind a proxy
-A dedicated web server like Nginx provides far more robustness than the Agent's internal [web server](https://github.com/netdata/netdata/blob/master/web/README.md).
+A dedicated web server like nginx provides more robustness than the Agent's internal [web server](https://github.com/netdata/netdata/blob/master/web/README.md).
Nginx can handle more concurrent connections, reuse idle connections, and use fast gzip compression to reduce payloads.
-For details on installing Nginx as a proxy for the local Agent dashboard, see our [Nginx
-doc](https://github.com/netdata/netdata/blob/master/docs/Running-behind-nginx.md).
-
-After you complete Nginx setup according to the doc linked above, we recommend setting `keepalive` to `1024`, and using
-gzip compression with the following options in the `location /` block:
-
-```conf
- location / {
- ...
- gzip on;
- gzip_proxied any;
- gzip_types *;
- }
-```
-
-Finally, edit `netdata.conf` with the following settings:
-
-```conf
-[global]
- bind socket to IP = 127.0.0.1
- disconnect idle web clients after seconds = 3600
- enable web responses gzip compression = no
-```
+For details on installing another web server as a proxy for the local Agent dashboard, see [reverse proxies](https://github.com/netdata/netdata/blob/master/docs/category-overview-pages/reverse-proxies.md).
## Disable/lower gzip compression for the dashboard
@@ -235,43 +222,3 @@ Or to lower the default compression level:
gzip compression level = 1
```
-## Disable logs
-
-If you installation is working correctly, and you're not actively auditing Netdata's logs, disable them in
-`netdata.conf`.
-
-```conf
-[logs]
- debug log = none
- error log = none
- access log = none
-```
-
-## Disable health checks
-
-If you are streaming metrics to parent nodes, we recommend you run your health checks on the parent, for all the metrics collected
-by the children nodes. This saves resources on the children and makes it easier to configure or disable alerts and agent notifications.
-
-The parents by default run health checks for each child, as long as it is connected (the details are in `stream.conf`).
-On the child nodes you should add to `netdata.conf` the following:
-
-```conf
-[health]
- enabled = no
-```
-
-## What's next?
-
-We hope this guide helped you better understand how to optimize the performance of the Netdata Agent.
-
-Now that your Agent is running smoothly, we recommend you [secure your nodes](https://github.com/netdata/netdata/blob/master/docs/configure/nodes.md) if you haven't
-already.
-
-Next, dive into some of Netdata's more complex features, such as configuring its health watchdog or exporting metrics to
-an external time-series database.
-
-- [Interact with dashboards and charts](https://github.com/netdata/netdata/blob/master/docs/visualize/interact-dashboards-charts.md)
-- [Configure health alarms](https://github.com/netdata/netdata/blob/master/docs/monitor/configure-alarms.md)
-- [Export metrics to external time-series databases](https://github.com/netdata/netdata/blob/master/docs/export/external-databases.md)
-
-[![analytics](https://www.google-analytics.com/collect?v=1&aip=1&t=pageview&_s=1&ds=github&dr=https%3A%2F%2Fgithub.com%2Fnetdata%2Fnetdata&dl=https%3A%2F%2Fmy-netdata.io%2Fgithub%2Fdocs%2Fguides%2Fconfigure%2Fperformance.md&_u=MAC~&cid=5792dfd7-8dc4-476b-af31-da2fdb9f93d2&tid=UA-64295674-3)](<>)