Apply the count-min-sketch.

apavlo · Jan 5, 2016 · b654e8b · b654e8b
1 parent 6a73d4f
commit b654e8b
Show file tree

Hide file tree

Showing 6 changed files with 625 additions and 2 deletions.
diff --git a/build.py b/build.py
@@ -312,6 +312,9 @@
  json_spirit_reader.cpp
  json_spirit_value.cpp
 """
+CTX.THIRD_PARTY_INPUT['mmh3'] = """
+ MurmurHash3.cpp
+"""
 
 ###############################################################################
 # SPECIFY THE TESTS

diff --git a/src/ee/anticache/AntiCacheEvictionManager.cpp b/src/ee/anticache/AntiCacheEvictionManager.cpp
@@ -1811,6 +1811,10 @@ bool AntiCacheEvictionManager::mergeUnevictedTuples(PersistentTable *table) {
 // Evicted Access Tracking Methods
 // -----------------------------------------
 
+#ifdef ANTICACHE_COUNTER
+const uint32_t AntiCacheEvictionManager::m_hash_seed[3] = {41, 239785, 878124560};
+#endif
+
 void AntiCacheEvictionManager::recordEvictedAccess(catalog::Table* catalogTable, TableTuple *tuple) {
     // FIXME: HACK HACK HACK
     if (m_evicted_block_ids.size() > 100000) {
@@ -1838,6 +1842,35 @@ void AntiCacheEvictionManager::recordEvictedAccess(catalog::Table* catalogTable,
         if (!m_update_access && (block_id & 0x10000000)) {
             uint32_t _block_id = (uint32_t)(block_id & 0x0FFFFFFF);
             int16_t ACID = (int16_t)((block_id & 0xE0000000) >> 29);
+
+
+
+            uint64_t key = ((uint64_t)_block_id << 32) + tuple_id;
+            uint64_t hash[2];
+
+            unsigned char min_sketch = 255;
+            for (int i = 0; i < SKETCH_HEIGHT; ++i) {
+                MurmurHash3_x64_128(&key, 8, m_hash_seed[i], hash);
+                //hash = MurmurHash64A(&key, 8, m_hash_seed[i]);
+                int j = hash[0] & SKETCH_MASK;
+                if (m_sketch[i][j] < 254) {
+                    m_sketch[i][j]++;
+                }
+                if (m_sketch[i][j] < min_sketch)
+                    min_sketch = m_sketch[i][j];
+            }
+
+            if (min_sketch > 10) {
+                m_evicted_tables_sync.push_back(catalogTable);
+                m_evicted_block_ids_sync.push_back(block_id); 
+                m_evicted_offsets_sync.push_back(tuple_id);
+                m_update_access = true;
+                return;
+                //printf("greater than: %u %d %d\n", min_sketch, _block_id, tuple_id);
+            } 
+
+
+
             AntiCacheDB* antiCacheDB = m_db_lookup[ACID]; 
             AntiCacheBlock* value = antiCacheDB->readBlock(_block_id, 0);
             //char* unevicted_tuples = new char[value->getSize()];

diff --git a/src/ee/anticache/AntiCacheEvictionManager.h b/src/ee/anticache/AntiCacheEvictionManager.h
@@ -35,13 +35,24 @@
 #include "common/NValue.hpp"
 #include "common/ValuePeeker.hpp"
 #include "anticache/AntiCacheDB.h"
+#include "mmh3/MurmurHash3.h"
 
 #include <vector>
 #include <map>
 #include <pthread.h>
 
 #define MAX_DBS 8
 
+#ifdef ANTICACHE_COUNTER
+    //#define SKETCH_WIDTH 32768
+    //#define SKETCH_MASK 32767
+    #define SKETCH_WIDTH 16384
+    #define SKETCH_MASK 16383
+    //#define SKETCH_WIDTH 1048576
+    //#define SKETCH_MASK 1048575
+    #define SKETCH_HEIGHT 3
+#endif
+
 namespace voltdb {
 
 class Table;
@@ -108,7 +119,8 @@ class AntiCacheEvictionManager {
 
 #ifdef ANTICACHE_COUNTER
     bool m_update_access;
-    unsigned char m_sketch[3][150000];
+    unsigned char m_sketch[SKETCH_HEIGHT][SKETCH_WIDTH];
+    static const uint32_t m_hash_seed[3];
 #endif
 
 protected:

diff --git a/src/ee/execution/VoltDBEngine.cpp b/src/ee/execution/VoltDBEngine.cpp
@@ -1344,7 +1344,20 @@ int VoltDBEngine::getStats(int selector, int locators[], int numLocators,
                             VOLT_EE_EXCEPTION_TYPE_EEEXCEPTION, message);
                 }
             }*/
-
+#ifdef ANTICACHE_COUNTER
+            AntiCacheEvictionManager* eviction_manager = m_executorContext->getAntiCacheEvictionManager();
+            if (eviction_manager != NULL) {
+                VOLT_ERROR("We have reset the sketch counter of size %ld :)", sizeof(eviction_manager->m_sketch));
+                /*
+                int sum = 0;
+                for (int i = 0; i < SKETCH_WIDTH; ++i)
+                    sum += eviction_manager->m_sketch[0][i];
+                printf("cardinality: %d\n", sum);
+                */
+
+                memset(eviction_manager->m_sketch, 0, sizeof(eviction_manager->m_sketch));
+            }
+#endif
             resultTable = m_statsManager.getStats(
                     (StatisticsSelectorType) selector, locatorIds, interval,
                     now);