Describe the issue
Currently the table has a rowkey of the form objectId_time. While searching along the objectId is fast (prefix search), search along the time axis remains quite slow (substring or regex search). To overcome this problem, one should use the fact that HBase is schemaless, and append new rows at the end of the current structure whose rowkey will be on the form t_time_objectId. In that way, one can search along both directions (at the cost to add more rows - but each is small as it does not contain columns, just the rowkey).
JulienPeloton
commented
3 years ago
Describe the issue Currently the table has a rowkey of the form
objectId_time. While searching along theobjectIdis fast (prefix search), search along thetimeaxis remains quite slow (substringorregexsearch). To overcome this problem, one should use the fact that HBase is schemaless, and append new rows at the end of the current structure whose rowkey will be on the formt_time_objectId. In that way, one can search along both directions (at the cost to add more rows - but each is small as it does not contain columns, just the rowkey).