通過shell指令碼生成查詢表資料的sql (r2筆記63天）
阿新 • • 發佈：2022-05-04
在工作中我們需要查詢表的資料條數，一般來說就是使用select count(1)或者select count(*)之類的語句。 當然了對於不同的表來說，應該還是可以做一些細分，能夠最大程度的提高效率，比如表中含有主鍵列，嘗試走索引掃面可能會被全表掃描效率要高。如果表中的資料太多，而且沒有一些相關的約束，可以考慮使用並行來提高等等。 以下就是使用shell指令碼所做的查詢資料條數的一個例子，看起來有些太不值得了，但是如果資料量很大的情況下這些分析就格外有用了。 比如表customer，資料量不是很大，可以直接走索引來做。
> ksh get_tab_cnt.sh prdappo customer ************************************************ CUSTOMER .859375 CUSTOMER select /*+ index_ffs(CUSTOMER,CUSTOMER_pk ) parallel_index(CUSTOMER,CUSTOMER_pk,1) */ 'CUSTOMER,', count(*) from CUSTOMER ;
如果表比較大，可以同時開啟並行。
************************************************
GREEMENT 10.359375
GREEMENT
select /*+ index_ffs(GREEMENT,GREEMENT_pk ) parallel_index(GREEMENT,GREEMENT_pk,4) */ 'GREEMENT,', count(*) from GREEMENT ; 對應的指令碼如下：
#!/bin/ksh #get_tab_cnt_sql DATE=`/bin/date '+%y%m%d_%H%M%S'` constr1=$DB_CONN_STR@$SH_DB_SID constr1=`echo ${constr1} | tr '[:lower:]' '[:upper:]'` timestamp=$3 time_vs="" if [ "${timestamp}" = "" ] then echo " " echo " " else if [ "${timestamp2}" = "" ] then exit; else time_vs=" As of timestamp to_date('${timestamp}','mm/dd/yyyy hh24:mi:ss') " time_vt=" As of timestamp to_date('${timestamp2}','mm/dd/yyyy hh24:mi:ss') " fi fi user1=`echo $constr1 | awk -F "/" '{print $1}'` inst1=`echo $constr1 | awk -F "@" '{print $2}'` if [[ -z ${inst1} ]];then inst1=$ORACLE_SID constr1=${constr1}@${inst1} fi function check_connectivity { Num=`echo show user | $ORACLE_HOME/bin/sqlplus -s $1| grep -i 'USER ' | wc -l` if [ $Num -gt 0 ] then ## ok - instance is accessible echo '************************************************' else ## inst is Inaccessible echo Connection Details: `echo $1 ` is Inaccessible... echo '************************************************' exit; fi } check_connectivity $constr1 sleep 1; sleep 1; mkdir `pwd`/Get_tab_cnt_${user1}_${inst1}_${DATE} mkdir `pwd`/Get_tab_cnt_${user1}_${inst1}_${DATE}/List small2='1' small='10' medium='30' big_par=8 medium_par=4 small_par=2 small2_par=1 COUNT_DIR=`pwd`/Get_tab_cnt_${user1}_${inst1}_${DATE} exclude_tab=x sqlplus -s ${constr1} <<EOF set head off set line 100 set pages 50000 set long 10000 col segment_name for a40 spool ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_size_temp.lst select distinct segment_name,to_number(to_char(sum(bytes/1024/1024/1024))) from dba_segments where owner=upper('$1') and segment_name =upper('$2') group by segment_name order by 2 desc; spool off; spool ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_with_pk.lst select table_name from dba_tables where owner=upper('$1') and table_name in ( select table_name from dba_constraints where owner=upper('$1') and table_name =upper('$2') and constraint_type='P'); spool off; EOF ################## Getting Big Tables ###################### sed '/^$/d' ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_size_temp.lst | grep -v SQL | grep -v select | grep -v ERROR | grep -v ORA- | grep -v '*' > ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_size.lst echo ${medium} | awk -v medium=${medium} ' $2 > medium { print $1 } ' ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_size.lst > ${COUNT_DIR}/List/${user1}_${inst1}_big_tab.lst echo ${medium} | echo ${small} | awk -v medium=${medium} -v small=${small} ' $2 < medium && $2 > small { print $1 } ' ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_size.lst > ${COUNT_DIR}/List/${user1}_${inst1}_medium_tab.lst echo ${small} | echo ${small2} | awk -v small=${small} -v small2=${small2} ' $2 < small && $2 > small2 { print $1 } ' ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_size.lst > ${COUNT_DIR}/List/${user1}_${inst1}_small_tab.lst echo ${small2} | awk -v small2=${small2} ' $2 < small2 { print $1 } ' ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_size.lst > ${COUNT_DIR}/List/${user1}_${inst1}_small2_tab.lst ###############################( Generating Count Files ( Big ) ############################################# while read table do is_pk=`grep -i ${table} ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_with_pk.lst | wc -l ` is_pk=`echo $is_pk` if [ is_pk -ne 0 ] then echo "select /*+ index_ffs(${table},${table}_pk ) parallel_index(${table},${table}_pk,${big_par}) */ '${table},', count(*) from ${table} ${time_vs} ;" >> ${COUNT_DIR}/sqls/${user1}_${inst1}_${table}_tab_count.sql else echo "select /*+ PARALLEL(${table},${big_par}) */ '${table} ,' , count (1) from ${table} ${time_vs} ;" >> ${COUNT_DIR}/sqls/${user1}_${inst1}_${table}_tab_count.sql fi done < ${COUNT_DIR}/List/${user1}_${inst1}_big_tab.lst ################################ Generating Count Files ( Medium ) ############################################# while read table do #echo " Table Name is : $table " is_pk=`grep -i ${table} ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_with_pk.lst | wc -l ` is_pk=`echo $is_pk` if [ is_pk -ne 0 ] then echo "select /*+ index_ffs(${table},${table}_pk ) parallel_index(${table},${table}_pk,${medium_par}) */ '${table},', count(*) from ${table} ${time_vt} ;" >> ${COUNT_DIR}/List/${user1}_${inst1}_medium_tab_count.lst else echo "select /*+ PARALLEL(${table},${medium_par}) */ '${table} ,' , count (1) from ${table} ${time_vs} ;" >> ${COUNT_DIR}/List/${user1}_${inst1}_medium_tab_count.lst fi done < ${COUNT_DIR}/List/${user1}_${inst1}_medium_tab.lst ################################ Generating Count Files ( Small ) ############################################# while read table do #echo " Table Name is : $table " is_pk=`grep -i ${table} ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_with_pk.lst | wc -l ` is_pk=`echo $is_pk` if [ is_pk -ne 0 ] then echo "select /*+ index_ffs(${table},${table}_pk ) parallel_index(${table},${table}_pk,${small_par}) */ '${table},', count(*) from ${table} ${time_vs} ;" >> ${COUNT_DIR}/List/${user1}_${inst1}_small_tab_count.lst else echo "select /*+ PARALLEL(${table},${small_par}) */ '${table} ,' , count (1) from ${table} ${time_vs} ;" >> ${COUNT_DIR}/List/${user1}_${inst1}_small_tab_count.lst fi done < ${COUNT_DIR}/List/${user1}_${inst1}_small_tab.lst ################################ Generating Count Files (Very Small ) ############################################# while read table do #echo " Table Name is : $table " is_pk=`grep -i ${table} ${COUNT_DIR}/List/${user1}_${inst1}_all_tab_with_pk.lst | wc -l ` is_pk=`echo $is_pk` if [ is_pk -ne 0 ] then echo "select /*+ index_ffs(${table},${table}_pk ) parallel_index(${table},${table}_pk,${small2_par}) */ '${table},', count(*) from ${table} ${time_vs} ;" >> ${COUNT_DIR}/List/${user1}_${inst1}_small2_tab_count.lst else echo "select /*+ PARALLEL(${table},${small2_par}) */ '${table} ,' , count (1) from ${table} ${time_vs} ;" >> ${COUNT_DIR}/List/${user1}_${inst1}_small2_tab_count.lst fi done < ${COUNT_DIR}/List/${user1}_${inst1}_small2_tab.lst rm ${COUNT_DIR}/List/*temp.lst #################### get table list from schemas ########################## touch ${COUNT_DIR}/${user1}_${inst1}_final_tab_cnt_stats.sql schema_con=${user1}_${inst1} function get_final_tab_cnt_stats { schema_type=$1 if [ -s ${COUNT_DIR}/List/${schema_con}_big_tab_count.lst ];then cat ${COUNT_DIR}/List/${schema_con}_big_tab_count.lst >> ${COUNT_DIR}/${schema_con}_final_tab_cnt_stats.sql fi if [ -s ${COUNT_DIR}/List/${schema_con}_medium_tab_count.lst ];then cat ${COUNT_DIR}/List/${schema_con}_medium_tab_count.lst >> ${COUNT_DIR}/${schema_con}_final_tab_cnt_stats.sql fi if [ -s ${COUNT_DIR}/List/${schema_con}_small2_tab_count.lst ];then cat ${COUNT_DIR}/List/${schema_con}_small2_tab_count.lst >> ${COUNT_DIR}/${schema_con}_final_tab_cnt_stats.sql fi if [ -s ${COUNT_DIR}/List/${schema_con}_small_tab_count.lst ];then cat ${COUNT_DIR}/List/${schema_con}_small_tab_count.lst >> ${COUNT_DIR}/${schema_con}_final_tab_cnt_stats.sql fi } get_final_tab_cnt_stats $schema_con cat ${COUNT_DIR}/${schema_con}_final_tab_cnt_stats.sql rm -rf ${COUNT_DIR}