[HBASE-9778] Add hint to ExplicitColumnTracker to avoid seeking

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.96.2, 0.98.1, 0.99.0, 0.94.18

Release Note

Introduces a new scan attribute to allow a scan operation with explicit columns (Scan.addColumn) to opportunistically look ahead a few KeyValues (columns/versions) before scheduling a seek operation to seek between columns.

A seek is efficient when it can seek past 5-10 KeyValue (columns) or 512-1024 bytes. With small rows and few versions look ahead is typically more efficient.

API:
{code}
    Scan s = new Scan(...);
    s.addColumn(...);
    // instructs the RegionServer to attempt two iterations of next before scheduling a seek
    s.setAttribute(Scan.HINT_LOOKAHEAD, Bytes.toBytes(2));
    table.getScanner(s);
{code}

• Assignee: Lars Hofhansl
• Reporter: Lars Hofhansl
• Created: 16/Oct/13 04:50
• Updated: 04/Mar/15 22:25
• Resolved: 11/Mar/14 18:35

Description

The issue of slow seeking in ExplicitColumnTracker was brought up by vrodionov on the dev list.

My idea here is to avoid the seeking if we know that there aren't many versions to skip.
How do we know? We'll use the column family's VERSIONS setting as a hint. If VERSIONS is set to 1 (or maybe some value < 10) we'll avoid the seek and call SKIP repeatedly.

HBASE-9769 has some initial number for this approach:
Interestingly it depends on which column(s) is (are) selected.

Some numbers: 4m rows, 5 cols each, 1 cf, 10 bytes values, VERSIONS=1, everything filtered at the server with a ValueFilter. Everything measured in seconds.

Without patch:

Wildcard	Col 1	Col 2	Col 4	Col 5	Col 2+4
6.4	8.5	14.3	14.6	11.1	20.3

With patch:

Wildcard	Col 1	Col 2	Col 4	Col 5	Col 2+4
6.4	8.4	8.9	9.9	6.4	10.0

Variation here was +- 0.2s.

So with this patch scanning is 2x faster than without in some cases, and never slower. No special hint needed, beyond declaring VERSIONS correctly.

Issue Links

• is related to: HBASE-4433 avoid extra next (potentially a seek) if done with column/row
• is related to: HBASE-9000 Linear reseek in Memstore
• relates to: HBASE-13109 Make better SEEK vs SKIP decisions during scanning