What is HFileLink and Reference in HBase?
In HBase there are two things that seems similar but are very different.
Links and References. Names sound similar but they are not.
Links - HFileLink is the implementation name for FileLink
HFileLink describes a link to hfile. It is more like a URL or path to the file. It means a hfile will be served regardless of its location being region directory or archive directory By default it will always look into the original location(region dir) defined by link name. If it doesn’t find the file there it will fall back to archive directory. So this is helpful for snapshots. As snapshots is nothing but list of hfile links to original hfiles of table for which snapshot it taken.