You use the direct data ingestion mechanism to send data directly to the Wavefront service instead of a proxy. While the Wavefront proxy has many benefits, direct data ingestion can be the best approach for some use cases.
Wavefront currently expects that your host, application, or custom code send metrics to a Wavefront proxy installed in your environment. The proxy forwards metrics to the Wavefront service. In production environments, two proxies behind a load balancer guarantee availability and prevent data loss.
Because some customers told us they’d like to send data directly to the Wavefront service, we’re now making this functionality available.
The following examples illustrate how to send data to Wavefront.
- An API token is required. Referred to as
<TOKEN>in the examples.
- You must know your Wavefront domain name. These examples use
- Currently, we support only Wavefront Data Format, which is named
wavefront. If you don’t specify
f=wavefront, we still use that format.
Multiple Data Points
wavefront.txt contains 1 or more lines in the Wavefront data format. You can send it to Wavefront like this:
cat wavefront.txt | curl -H "Authorization: Bearer <TOKEN>" -F file=@- https://mydomain.wavefront.com/report
Single Data Point
You can send a single data point like this:
echo "hello.world 1 source=<myhost>" | curl -H "Authorization: Bearer <TOKEN>" --data @- https://mydomain.wavefront.com/report
You can perform direct ingestion of histogram distributions in histogram distribution format. You cannot perform direct ingestion of histogram data in Wavefront data format.
Here’s a simple example:
echo '!M #20 30 #10 5 request.latency source=appServer1 region=us-west' | curl -H "Authorization: Bearer <TOKEN>" --data @- https://mydomain.wavefront.com/report?f=histogram
- Enclose the distribution in single quotes for compatibility with the
- Specify the distribution using the histogram distribution format.
f=histogramat the end of the
--dataargument to ensure the input is treated as a histogram distribution.
Trace Data (Spans)
You can perform direct ingestion of trace data in Wavefront span format.
Here’s a simple example:
echo "getAllUsers source=localhost traceId=7b3bf470-9456-11e8-9eb6-529269fb1459 spanId=0313bafe-9457-11e8-9eb6-529269fb1459 parent=2f64e538-9457-11e8-9eb6-529269fb1459 application=Wavefront service=istio cluster=none shard=none http.method=GET <start_milliseconds> <duration_milliseconds>" | curl -H "Authorization: Bearer <TOKEN>" --data @- https://mydomain.wavefront.com/report?f=trace
- Specify the span using Wavefront span format.
f=traceat the end of the
--dataargument to ensure the input is treated as trace data.
shard=nonefor a span that does not have values for the
<duration_milliseconds>with numbers indicating the span’s start time and duration.
Comparing Proxy and Direct Ingestion
Direct ingestion has some benefits, but also some limitations:
- No disk queueing
- No rate limiting
- No blacklist or whitelist
- No preprocessor
- Only Wavefront data format is currently supported. No support for OpenTSB, JSON, and Pickle
- No support for log ingestion
Some limitations, for example supported data formats, might be resolved in the future, but other limitations are benefits of the Wavefront proxy.