YCSB: customizable zipfian constant (#321)

Co-authored-by: Andy Pavlo <[email protected]>

Author: MohammadKhalaji

config/cockroachdb/sample_ycsb_config.xml

@@ -16,6 +16,9 @@
     <!-- Optional: Override the field size for each column in USERTABLE -->
     <!-- <fieldSize>8</fieldSize> -->
 
+    <!-- Optional: Override the zipfian constant to modify the skew -->
+    <!-- <skewFactor>0.99</skewFactor> -->
+
     <!-- The workload -->
     <terminals>1</terminals>
     <works>

config/mariadb/sample_ycsb_config.xml

@@ -16,6 +16,9 @@
     <!-- Optional: Override the field size for each column in USERTABLE -->
     <!-- <fieldSize>8</fieldSize> -->
 
+    <!-- Optional: Override the zipfian constant to modify the skew -->
+    <!-- <skewFactor>0.99</skewFactor> -->
+
     <!-- The workload -->
     <terminals>1</terminals>
     <works>

config/mysql/sample_ycsb_config.xml

@@ -12,6 +12,9 @@
 
     <!-- Scalefactor in YCSB is *1000 the number of rows in the USERTABLE-->
     <scalefactor>1</scalefactor>
+    
+    <!-- Optional: Override the zipfian constant to modify the skew -->
+    <!-- <skewFactor>0.99</skewFactor> -->
 
     <!-- The workload -->
     <terminals>1</terminals>

config/noisepage/sample_ycsb_config.xml

@@ -13,6 +13,9 @@
     <!-- Scalefactor in YCSB is *1000 the number of rows in the USERTABLE-->
     <scalefactor>1</scalefactor>
 
+    <!-- Optional: Override the zipfian constant to modify the skew -->
+    <!-- <skewFactor>0.99</skewFactor> -->
+
     <!-- The workload -->
     <terminals>1</terminals>
     <works>

config/postgres/sample_ycsb_config.xml

@@ -16,6 +16,9 @@
     <!-- Optional: Override the field size for each column in USERTABLE -->
     <!-- <fieldSize>8</fieldSize> -->
 
+    <!-- Optional: Override the zipfian constant to modify the skew -->
+    <!-- <skewFactor>0.99</skewFactor> -->
+
     <!-- The workload -->
     <terminals>1</terminals>
     <works>

config/spanner/sample_ycsb_config.xml

@@ -14,6 +14,9 @@
     <!-- Optional: Override the field size for each column in USERTABLE -->
     <!-- <fieldSize>8</fieldSize> -->
 
+    <!-- Optional: Override the zipfian constant to modify the skew -->
+    <!-- <skewFactor>0.99</skewFactor> -->
+
     <!-- The workload -->
     <terminals>1</terminals>
     <works>

config/sqlite/sample_ycsb_config.xml

@@ -10,6 +10,9 @@
 
     <!-- Scalefactor in YCSB is *1000 the number of rows in the USERTABLE-->
     <scalefactor>1</scalefactor>
+
+    <!-- Optional: Override the zipfian constant to modify the skew -->
+    <!-- <skewFactor>0.99</skewFactor> -->
 
     <!-- SQLITE only supports one writer thread -->
     <loaderThreads>1</loaderThreads>

config/sqlserver/sample_ycsb_config.xml

@@ -16,6 +16,9 @@
     <!-- Optional: Override the field size for each column in USERTABLE -->
     <!-- <fieldSize>8</fieldSize> -->
 
+    <!-- Optional: Override the zipfian constant to modify the skew -->
+    <!-- <skewFactor>0.99</skewFactor> -->
+
     <!-- The workload -->
     <terminals>1</terminals>
     <works>

src/main/java/com/oltpbenchmark/benchmarks/ycsb/YCSBBenchmark.java

@@ -43,6 +43,11 @@ public class YCSBBenchmark extends BenchmarkModule {
      */
     protected final int fieldSize;
 
+    /**
+     * The constant used in the zipfian distribution (to modify the skew)
+     */
+    protected final double skewFactor;
+
     public YCSBBenchmark(WorkloadConfiguration workConf) {
         super(workConf);
 
@@ -54,6 +59,15 @@ public YCSBBenchmark(WorkloadConfiguration workConf) {
         if (this.fieldSize <= 0) {
             throw new RuntimeException("Invalid YCSB fieldSize '" + this.fieldSize + "'");
         }
+
+        double skewFactor = 0.99;
+        if (workConf.getXmlConfig() != null && workConf.getXmlConfig().containsKey("skewFactor")) {
+            skewFactor = workConf.getXmlConfig().getDouble("skewFactor");
+            if (skewFactor <= 0 || skewFactor >= 1) {
+                throw new RuntimeException("Invalid YCSB skewFactor '" + skewFactor + "'");
+            }
+        }
+        this.skewFactor = skewFactor;
     }
 
     @Override

src/main/java/com/oltpbenchmark/benchmarks/ycsb/YCSBWorker.java

@@ -58,8 +58,8 @@ class YCSBWorker extends Worker<YCSBBenchmark> {
     public YCSBWorker(YCSBBenchmark benchmarkModule, int id, int init_record_count) {
         super(benchmarkModule, id);
         this.data = new char[benchmarkModule.fieldSize];
-        this.readRecord = new ZipfianGenerator(rng(), init_record_count);// pool for read keys
-        this.randScan = new ZipfianGenerator(rng(), YCSBConstants.MAX_SCAN);
+        this.readRecord = new ZipfianGenerator(rng(), init_record_count, benchmarkModule.skewFactor);// pool for read keys
+        this.randScan = new ZipfianGenerator(rng(), YCSBConstants.MAX_SCAN, benchmarkModule.skewFactor);
 
         synchronized (YCSBWorker.class) {
             // We must know where to start inserting