Update row-size-sampler.sh

modified default samples

Author: michaelraney

bin/row-size-sampler.sh

@@ -21,6 +21,6 @@ keyspaces=$(echo desc keyspaces | ./cqlsh $@  | xargs -n1 echo | grep -v $SYSTEM
 for ks in $keyspaces; do
   tables=$(echo "SELECT table_name FROM system_schema.tables WHERE keyspace_name='$ks';" | ./cqlsh $@ | xargs -n1 echo | grep -v $TABLEFILTER)
   for tb in $tables; do
-    ./cqlsh $@ -e  "CONSISTENCY LOCAL_ONE; SELECT * FROM \"$ks\".\"$tb\" LIMIT 30000;" | grep -v '\[json\]\|rows)\|-----\|^$' | tr -d ' ' | awk -v keyspace=$ks -v table=$tb -F'|' 'BEGIN {columns=0; numSamples=100000; kilobyte=1024; min = "NaN"; max = -1; lines = 1; }  { if(NR==2){columns=NF;} if(NR>2){thislen=length($0)+107; total+=thislen; squares+=thislen^2;  lines+=1; avg=total/lines;  min = (thislen<min ? thislen : min); max =  (thislen>max ? thislen : max) }} NR==numSamples {exit} END { printf("%s.%s = { lines: %d, columns: %d, average: %d bytes, stdev: %d bytes, min: %d bytes, max: %d bytes}\n", keyspace, table, lines, columns, avg, sqrt(squares/lines - (avg^2)), min, max); }'  >> row-size-estimates.txt 2>&1
+    ./cqlsh $@ -e  "CONSISTENCY LOCAL_ONE; SELECT * FROM \"$ks\".\"$tb\" LIMIT 30000;" | grep -v '\[json\]\|rows)\|-----\|^$' | tr -d ' ' | awk -v keyspace=$ks -v table=$tb -F'|' 'BEGIN {columns=0; numSamples=30000; kilobyte=1024; min = "NaN"; max = -1; lines = 1; }  { if(NR==2){columns=NF;} if(NR>2){thislen=length($0)+107; total+=thislen; squares+=thislen^2;  lines+=1; avg=total/lines;  min = (thislen<min ? thislen : min); max =  (thislen>max ? thislen : max) }} NR==numSamples {exit} END { printf("%s.%s = { lines: %d, columns: %d, average: %d bytes, stdev: %d bytes, min: %d bytes, max: %d bytes}\n", keyspace, table, lines, columns, avg, sqrt(squares/lines - (avg^2)), min, max); }'  >> row-size-estimates.txt 2>&1
   done
 done