Hi, I have a row key like this:
<id>:<time> There are millions of records per id throughout the day. When querying for id=xyz, I want one row (either the first or the last depending) for every 5 minute time bucket in a specific time range. I can do this efficiently with a single scan and custom filter that uses the SEEK_NEXT_USING_HINT feature to skip from the beginning of one five minute bucket to the next. However, it only works if I want the first row in every time bucket. If I want the last row in every time bucket I have to reverse my whole scan. Is there a way to seek in the opposite direction of the scan? I've tried but such seek hints seem to be silently ignored. If not, is there some other efficient, single-scan way to get the last row for every time bucket while scanning in ascending order? Thanks, Whitney
