From 0f8a78b8c83b675366b0def7db98f4ca6ee587bc Mon Sep 17 00:00:00 2001
From: Scott Anderson <scott@influxdata.com>
Date: Mon, 1 Jul 2019 17:09:20 -0600
Subject: [PATCH] optimize writes to influxdb

---
 .../best-practices/optimize-writes.md         | 53 +++++++++++++++++++
 1 file changed, 53 insertions(+)
 create mode 100644 content/v2.0/write-data/best-practices/optimize-writes.md

diff --git a/content/v2.0/write-data/best-practices/optimize-writes.md b/content/v2.0/write-data/best-practices/optimize-writes.md
new file mode 100644
index 000000000..71ca9390a
--- /dev/null
+++ b/content/v2.0/write-data/best-practices/optimize-writes.md
@@ -0,0 +1,53 @@
+---
+title: Optimize writes
+seotitle: Optimize writes to InfluxDB
+description: >
+  Optimize performance and system overhead when writing data to InfluxDB using some simple tips.
+weight: 202
+menu:
+  v2_0:
+    parent: write-best-practices
+v2.0/tags: [best practices, write]
+---
+
+Optimize performance and system overhead when writing data to InfluxDB using some simple tips.
+
+_[Telegraf](/v2.0/write-data/use-telegraf/), [InfluxDB scrapers](/v2.0/write-data/scrape-data/),
+and many [InfluxDB client libraries](/v2.0/reference/client-libraries/) employ these optimizations by default._
+
+---
+
+## Batch writes
+Minimize network overhead when writing data to InfluxDB by writing data in batches.
+
+{{% note %}}
+The recommend batch size is 5000 lines of line protocol.
+{{% /note %}}
+
+## Sort tags by key
+Sort tags by key before writing data points to InfluxDB.
+_The sort should match results from the [Go `bytes.Compare` function](http://golang.org/pkg/bytes/#Compare)
+(lexicographic order)._
+
+```sh
+# Line protocol example with unsorted tags
+measurement,tagC=therefore,tagE=am,tagA=i,tagD=i,tagB=think fieldKey=fieldValue 1562020262
+
+# Optimized line protocol example with tags sorted by key
+measurement,tagA=i,tagB=think,tagC=therefore,tagD=i,tagE=am fieldKey=fieldValue 1562020262
+```
+
+## Use the coarsest time precision possible
+InfluxDB lets you write data with up to nanosecond precision.
+If data isn't collected at that precision, there is no need to write data at that precision.
+Use the coarsest precision possible for timestamps.
+This can result in significant compression improvements.
+
+_Specify timestamp precision when [writing to InfluxDB](/v2.0/write-data/#precision)._
+
+## Synchronize hosts with NTP
+Use the Network Time Protocol (NTP) to synchronize time between hosts.
+InfluxDB uses a host's local time in UTC to assign timestamps to data if no
+timestamp is included in the line protocol.
+If hosts' clocks aren't synchronized with NTP, the timestamps on data written
+to InfluxDB can be inaccurate.