Frequently Asked Questions (FAQ)
Find answers to frequently-asked questions below.
When initially connecting to a Livestream endpoint, how old is the data that starts streaming?
Livestream starts streaming data collected at the time of the client's connection. In the case of a client disconnect/reconnect, data is streamed from the point of disconnection. This brief backfill period only occurs if the disconnection is shorter than several minutes.
How do I customize the reconnection backfill logic?
You can use the reset query string to request old or new data. For example, ?reset=largest requests only the newest data from the stream and ignores data missed during reconnection. The ?reset=smallest query string starts streaming the oldest data available and attempts to catch up to the present.
What is the latency between when a hit is collected by Adobe and when it appears in Livestream?
Latency for Livestream data typically ranges between 20 seconds and 5 minutes. This time range is not a guarantee or a service-level agreement. Several factors can increase this time range beyond 5 minutes, including:
- Report suites configured for A4T increase Livestream latency by about 10 minutes. Livestream data must wait for the A4T collection pipeline to finish.
- Instances where collection or processing is delayed upstream. See Data availability and latency in Adobe Analytics for more information.
Can I request uncompressed Livestream data?
No. Livestream requires clients to support Gzip compression by default.
What transfer encoding is used?
Livestream uses Chunked transfer encoding. There can be more than one record per chunk, and each record is separated by a carriage return/line feed (CRLF). Many http client libraries handle chunked transfer encoding transparently.
Can I create multiple connections to the same stream?
Yes. Use the maxConnections GET query parameter. If multiple connections are created, data is distributed across each connection. Data is grouped by visitor IDs, but is out of order. The timestamp field can be used to sort the hits. A best effort is made to evenly distribute hits. Because data is grouped by visitor ID, a visitor that produces a large volume of data can create differences in volumes for each client. A maximum of 8 connections is allowed.
If multiple instances of the same stream are required, Adobe recommends that you create infrastructure to replicate that data.
How do I avoid receiving duplicate records?
The likelihood of receiving duplicate records increases during reconnect or when new clients connect to an existing stream. The hitIdHigh and hitIdLow columns can be used to deduplicate hits.
What do I do with empty records?
Empty records are sometimes returned in the stream. These can be ignored.
Where does Livestream occur in the data processing order?
Livestream data is only partially processed to mitigate latency. See Processing order in the Analytics technotes guide for more information.
Livestream includes basic processing, such as Processing rules, VISTA rules, and geolocation lookups. It does not include persistence, such as eVars persisting data across hits within a visit. It also does not include visit-based or visitor-based data like visits, visit number, unique visitors, or customer loyalty.
