I'm not as familiar with extended statistics - can you help me understand why it would be STS_PROVIDER, STS_PROGRAM_NO (from table "B") that you mention? If I look at the plan, I see the following column references from table "B": The ones you mention: access("B"."STS_PROGRAM_NO"="C"."STP_PROGRAM_NO" AND "B"."STS_PROVIDER"="C"."STP_PROVIDER") PLUS: access("B"."STS_PERF_SOURCE"='CB') filter("B"."STS_PERF_SOURCE"='CB') Would I add STS_PERF_SOURCE to the extended statistics information? I'm confused on how you identified those particular columns I guess as the ones that should have the extended statistics? I see those are access columns and doesn't include filter columns is why I ask. (So, ultimately the question is: "How do you identify the most appropriate columns for extended statistics?") Thanks, Chris On Tue, Jan 27, 2015 at 9:12 AM, Sayan Sergeevich Malakshinov < malakshinovss@xxxxxxxxx> wrote: > > Also, it completed successfully for period 20142 but the explain > > plan for that one said it would need 573 TB of TEMP space so now I > > realize I have zero way to "guess" how much temp space it really > > needs but that 150 GB isn't enough. > > > > The stats are accurate for 100% estimation so I'm not sure why the > > estimate on the temp space is so high. > > How different are values of "A-Rows" from "E-rows"? > It seems that you have to gather extended statistics on > (STS_PROVIDER,STS_PROGRAM_NO) and on > (STS_PERF_PERIOD,STS_DPS_TYPE,STG_GROUP_NO,STG_GROUP_SEQ_NO) column sets. > > -- > Best regards, > Sayan Malakshinov > http://orasql.org