<\/p>\n
A) Good old way to query partial data<\/strong><\/p>\n SELECT * FROM SAMTAB WHERE rand() < 0.01 order by 1 returning ~1% of overall dataset. You can always check data demography (ID values starts at 100 000, step 1 = last record 109 999)<\/p>\n B) Mr.Bernoulli? Come and play with us!<\/strong><\/p>\n You can find further info about this sampling here<\/a>. Usage is very simple:<\/p>\n select * from samtab tablesample bernoulli(0.8) order by 1 returning ~0,8% of overall dataset.<\/p>\n As Bernoulli is pure row-level sample method, every record must be read = can be very demanding and usage needs your consideration.<\/p>\n C) Let\u00b4s turn to next page<\/strong><\/p>\n Very efficient sampling method (using pages, not rows)<\/p>\n select * from samtab tablesample system(0.8) order by 1<\/span><\/p>\n returning ~0,8% of overall dataset, using same pages = demography is completely different very often. In this case, there are sets of following records.<\/p>\n -a-<\/p>\n","protected":false},"excerpt":{"rendered":" Sometimes it would be very helpful to have a chance gather sample values. What if rownum limitation affects quality of records? What if the main table logic is residental in the middle of huge table? Welcome to data sampling world … Continue reading
\n<\/span><\/p>\n<\/p>\n
\n<\/span><\/p>\n<\/p>\n
<\/p>\n