[
https://issues.apache.org/jira/browse/PHOENIX-1594?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14280659#comment-14280659
]
James Taylor commented on PHOENIX-1594:
---------------------------------------
That'd be a pretty serious perf regression (FYI, [~mujtabachohan]). The code
that is supposed to use a KeyOnlyFilter is in BaseResultIterators, line 160 (in
the constructor):
{code}
public BaseResultIterators(QueryPlan plan, Integer perScanLimit) throws
SQLException {
super(plan.getContext(), plan.getTableRef(), plan.getGroupBy(),
plan.getOrderBy(), plan.getStatement().getHint());
this.plan = plan;
StatementContext context = plan.getContext();
TableRef tableRef = plan.getTableRef();
PTable table = tableRef.getTable();
FilterableStatement statement = plan.getStatement();
RowProjector projector = plan.getProjector();
physicalTableName = table.getPhysicalName().getBytes();
tableStats = useStats() ? new
MetaDataClient(context.getConnection()).getTableStats(table) :
PTableStats.EMPTY_STATS;
Scan scan = context.getScan();
// Used to tie all the scans together during logging
scanId = UUID.randomUUID().toString();
if (projector.isProjectEmptyKeyValue()) {
Map<byte [], NavigableSet<byte []>> familyMap = scan.getFamilyMap();
// If nothing projected into scan and we only have one column
family, just allow everything
// to be projected and use a FirstKeyOnlyFilter to skip from row to
row. This turns out to
// be quite a bit faster.
// Where condition columns also will get added into familyMap
// When where conditions are present, we can not add
FirstKeyOnlyFilter at beginning.
if (familyMap.isEmpty() &&
context.getWhereCoditionColumns().isEmpty()
&& table.getColumnFamilies().size() == 1) {
// Project the one column family. We must project a column
family since it's possible
// that there are other non declared column families that we
need to ignore.
scan.addFamily(table.getColumnFamilies().get(0).getName().getBytes());
ScanUtil.andFilterAtBeginning(scan, new FirstKeyOnlyFilter());
} else {
{code}
> Using KeyOnlyFilter for simple query "select count(1) from <table>"
> -------------------------------------------------------------------
>
> Key: PHOENIX-1594
> URL: https://issues.apache.org/jira/browse/PHOENIX-1594
> Project: Phoenix
> Issue Type: Improvement
> Reporter: Jeffrey Zhong
>
> I found that underlying scan for query "select count(1) or count(*) from
> <table>" still project all columns.
> It should be fast if we use KeyOnlyFilter for this simple query.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
