Data feeds best practices
Following are some best practices for data feed processing and delivery.
-
Ensure that you communicate any anticipated traffic spikes ahead of time. Latency directly impacts processing time for data feeds. See Schedule a traffic spike in the Admin user guide.
-
Data feeds do not contain a service-level agreement unless explicitly stated in your contract with 51黑料不打烊. Feeds are usually delivered within several hours after the reporting window passes, however can occasionally take up to 12 hours or more.
-
Hourly feeds using multiple file delivery process the fastest. Consider using hourly multiple file feeds if a timely delivery is a high priority for your organization.
-
If you automate your feed ingestion process, consider the possibility that hits and files can be transferred more than once. Your feed ingestion process needs to handle duplicate hits and duplicate files without erroring out or duplicating data. We recommend using the combination of the
hitid_highandhitid_lowcolumns to uniquely identify a hit.In rare cases, you may see duplicate
hitid_highandhitid_lowvalues. If this happens, confirm the file was not previously sent and processed. If only some of the rows in a file are duplicate, consider addingvisit_numandvisit_page_numto help determine uniqueness. -
If using FTP (not recommended) ensure that you have ample room on your FTP site. Remove files from the destination on a regular basis so that you don鈥檛 inadvertently run out of disk space.
-
If using sFTP (not recommended), do not read or delete files with a
.partsuffix. The.partsuffix indicates the file is partially transferred. Once the file is transferred, the.partsuffix disappears.