HubSpot Backport: HBASE-27687 Enhance quotas to consume blockBytesSca…

…nned rather than response size (apache#5654)

Signed-off-by: Bryan Beaudreault <bbeaudreault@apache.org>

hbase-server/src/main/java/org/apache/hadoop/hbase/quotas/DefaultOperationQuota.java

@@ -22,6 +22,8 @@
 import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.hbase.client.Mutation;
 import org.apache.hadoop.hbase.client.Result;
+import org.apache.hadoop.hbase.ipc.RpcCall;
+import org.apache.hadoop.hbase.ipc.RpcServer;
 import org.apache.yetus.audience.InterfaceAudience;
 import org.apache.yetus.audience.InterfaceStability;
 
@@ -49,9 +51,15 @@ public class DefaultOperationQuota implements OperationQuota {
   protected long readDiff = 0;
   protected long writeCapacityUnitDiff = 0;
   protected long readCapacityUnitDiff = 0;
+  private boolean useResultSizeBytes;
+  private long blockSizeBytes;
 
-  public DefaultOperationQuota(final Configuration conf, final QuotaLimiter... limiters) {
+  public DefaultOperationQuota(final Configuration conf, final int blockSizeBytes,
+    final QuotaLimiter... limiters) {
     this(conf, Arrays.asList(limiters));
+    this.useResultSizeBytes =
+      conf.getBoolean(OperationQuota.USE_RESULT_SIZE_BYTES, USE_RESULT_SIZE_BYTES_DEFAULT);
+    this.blockSizeBytes = blockSizeBytes;
   }
 
   /**
@@ -94,8 +102,17 @@ public void checkQuota(int numWrites, int numReads, int numScans) throws RpcThro
   public void close() {
     // Adjust the quota consumed for the specified operation
     writeDiff = operationSize[OperationType.MUTATE.ordinal()] - writeConsumed;
-    readDiff = operationSize[OperationType.GET.ordinal()]
-      + operationSize[OperationType.SCAN.ordinal()] - readConsumed;
+
+    long resultSize =
+      operationSize[OperationType.GET.ordinal()] + operationSize[OperationType.SCAN.ordinal()];
+    if (useResultSizeBytes) {
+      readDiff = resultSize - readConsumed;
+    } else {
+      long blockBytesScanned =
+        RpcServer.getCurrentCall().map(RpcCall::getBlockBytesScanned).orElse(0L);
+      readDiff = Math.max(blockBytesScanned, resultSize) - readConsumed;
+    }
+
     writeCapacityUnitDiff =
       calculateWriteCapacityUnitDiff(operationSize[OperationType.MUTATE.ordinal()], writeConsumed);
     readCapacityUnitDiff = calculateReadCapacityUnitDiff(
@@ -140,8 +157,15 @@ public void addMutation(final Mutation mutation) {
    */
   protected void updateEstimateConsumeQuota(int numWrites, int numReads, int numScans) {
     writeConsumed = estimateConsume(OperationType.MUTATE, numWrites, 100);
-    readConsumed = estimateConsume(OperationType.GET, numReads, 100);
-    readConsumed += estimateConsume(OperationType.SCAN, numScans, 1000);
+
+    if (useResultSizeBytes) {
+      readConsumed = estimateConsume(OperationType.GET, numReads, 100);
+      readConsumed += estimateConsume(OperationType.SCAN, numScans, 1000);
+    } else {
+      // assume 1 block required for reads. this is probably a low estimate, which is okay
+      readConsumed = numReads > 0 ? blockSizeBytes : 0;
+      readConsumed += numScans > 0 ? blockSizeBytes : 0;
+    }
 
     writeCapacityUnitConsumed = calculateWriteCapacityUnit(writeConsumed);
     readCapacityUnitConsumed = calculateReadCapacityUnit(readConsumed);

hbase-server/src/main/java/org/apache/hadoop/hbase/quotas/ExceedOperationQuota.java

@@ -40,9 +40,9 @@ public class ExceedOperationQuota extends DefaultOperationQuota {
   private static final Logger LOG = LoggerFactory.getLogger(ExceedOperationQuota.class);
   private QuotaLimiter regionServerLimiter;
 
-  public ExceedOperationQuota(final Configuration conf, QuotaLimiter regionServerLimiter,
-    final QuotaLimiter... limiters) {
-    super(conf, limiters);
+  public ExceedOperationQuota(final Configuration conf, int blockSizeBytes,
+    QuotaLimiter regionServerLimiter, final QuotaLimiter... limiters) {
+    super(conf, blockSizeBytes, limiters);
     this.regionServerLimiter = regionServerLimiter;
   }
 

hbase-server/src/main/java/org/apache/hadoop/hbase/quotas/OperationQuota.java

@@ -35,6 +35,16 @@ public enum OperationType {
     SCAN
   }
 
+  /**
+   * If false, the default, then IO based throttles will consume read availability based on the
+   * block bytes scanned by the given request. If true then IO based throttles will use result size
+   * rather than block bytes scanned. Using block bytes scanned should be preferable to using result
+   * size, because otherwise access patterns like heavily filtered scans may be able to produce a
+   * significant and effectively un-throttled workload.
+   */
+  String USE_RESULT_SIZE_BYTES = "hbase.quota.use.result.size.bytes";
+  boolean USE_RESULT_SIZE_BYTES_DEFAULT = false;
+
   /**
    * Checks if it is possible to execute the specified operation. The quota will be estimated based
    * on the number of operations to perform and the average size accumulated during time.

hbase-server/src/main/java/org/apache/hadoop/hbase/quotas/RegionServerRpcQuotaManager.java

@@ -21,6 +21,7 @@
 import java.util.List;
 import java.util.Optional;
 import org.apache.hadoop.hbase.TableName;
+import org.apache.hadoop.hbase.client.TableDescriptor;
 import org.apache.hadoop.hbase.ipc.RpcScheduler;
 import org.apache.hadoop.hbase.ipc.RpcServer;
 import org.apache.hadoop.hbase.regionserver.Region;
@@ -113,7 +114,8 @@ QuotaCache getQuotaCache() {
    * @param table the table where the operation will be executed
    * @return the OperationQuota
    */
-  public OperationQuota getQuota(final UserGroupInformation ugi, final TableName table) {
+  public OperationQuota getQuota(final UserGroupInformation ugi, final TableName table,
+    final int blockSizeBytes) {
     if (isQuotaEnabled() && !table.isSystemTable() && isRpcThrottleEnabled()) {
       UserQuotaState userQuotaState = quotaCache.getUserQuotaState(ugi);
       QuotaLimiter userLimiter = userQuotaState.getTableLimiter(table);
@@ -123,7 +125,8 @@ public OperationQuota getQuota(final UserGroupInformation ugi, final TableName t
           LOG.trace("get quota for ugi=" + ugi + " table=" + table + " userLimiter=" + userLimiter);
         }
         if (!useNoop) {
-          return new DefaultOperationQuota(this.rsServices.getConfiguration(), userLimiter);
+          return new DefaultOperationQuota(this.rsServices.getConfiguration(), blockSizeBytes,
+            userLimiter);
         }
       } else {
         QuotaLimiter nsLimiter = quotaCache.getNamespaceLimiter(table.getNamespaceAsString());
@@ -139,11 +142,11 @@ public OperationQuota getQuota(final UserGroupInformation ugi, final TableName t
         }
         if (!useNoop) {
           if (exceedThrottleQuotaEnabled) {
-            return new ExceedOperationQuota(this.rsServices.getConfiguration(), rsLimiter,
-              userLimiter, tableLimiter, nsLimiter);
+            return new ExceedOperationQuota(this.rsServices.getConfiguration(), blockSizeBytes,
+              rsLimiter, userLimiter, tableLimiter, nsLimiter);
           } else {
-            return new DefaultOperationQuota(this.rsServices.getConfiguration(), userLimiter,
-              tableLimiter, nsLimiter, rsLimiter);
+            return new DefaultOperationQuota(this.rsServices.getConfiguration(), blockSizeBytes,
+              userLimiter, tableLimiter, nsLimiter, rsLimiter);
           }
         }
       }
@@ -213,9 +216,10 @@ private OperationQuota checkQuota(final Region region, final int numWrites, fina
     } else {
       ugi = User.getCurrent().getUGI();
     }
-    TableName table = region.getTableDescriptor().getTableName();
+    TableDescriptor tableDescriptor = region.getTableDescriptor();
+    TableName table = tableDescriptor.getTableName();
 
-    OperationQuota quota = getQuota(ugi, table);
+    OperationQuota quota = getQuota(ugi, table, region.getMinBlockSizeBytes());
     try {
       quota.checkQuota(numWrites, numReads, numScans);
     } catch (RpcThrottlingException e) {

hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/HRegion.java

@@ -438,6 +438,8 @@ public MetricsTableRequests getMetricsTableRequests() {
 
   private final CellComparator cellComparator;
 
+  private final int minBlockSizeBytes;
+
   /**
    * @return The smallest mvcc readPoint across all the scanners in this region. Writes older than
    *         this readPoint, are included in every read operation.
@@ -900,6 +902,9 @@ public HRegion(final HRegionFileSystem fs, final WAL wal, final Configuration co
           .remove(getRegionInfo().getEncodedName());
       }
     }
+
+    minBlockSizeBytes = Arrays.stream(this.htableDescriptor.getColumnFamilies())
+      .mapToInt(ColumnFamilyDescriptor::getBlocksize).min().orElse(HConstants.DEFAULT_BLOCKSIZE);
   }
 
   private void setHTableSpecificConf() {
@@ -1941,6 +1946,11 @@ public Configuration getReadOnlyConfiguration() {
     return new ReadOnlyConfiguration(this.conf);
   }
 
+  @Override
+  public int getMinBlockSizeBytes() {
+    return minBlockSizeBytes;
+  }
+
   private ThreadPoolExecutor getStoreOpenAndCloseThreadPool(final String threadNamePrefix) {
     int numStores = Math.max(1, this.htableDescriptor.getColumnFamilyCount());
     int maxThreads = Math.min(numStores, conf.getInt(HConstants.HSTORE_OPEN_AND_CLOSE_THREADS_MAX,

hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/Region.java

@@ -611,4 +611,10 @@ void requestCompaction(byte[] family, String why, int priority, boolean major,
    *         if you try to set a configuration.
    */
   Configuration getReadOnlyConfiguration();
+
+  /**
+   * The minimum block size configuration from all relevant column families. This is used when
+   * estimating quota consumption.
+   */
+  int getMinBlockSizeBytes();
 }