feat: use ColumnEncoding_Kind_DIRECT_DELTA as default in offset stream (#1337)

* feat: use ColumnEncoding_Kind_DIRECT_DELTA as default in offset stream

Optimize performance of variable-length column offsets by switching from
Zstd to delta encoding. This approach better compresses incremental integer
sequences, cutting disk space by more than half while maintaining performance.

The following is a comparison of file sizes for different encoding methods on TPC-DS 20G:

Name                   PAX(ZSTD)    AOCS_SIZE    PAX(Delta)    PAX SIZE / AOCS * 100%
call_center               12 kB       231 kB      10185 bytes        4.31%
catalog_page             499 kB       653 kB       393 kB           60.18%
catalog_returns          240 MB       171 MB       178 MB          104.09%
catalog_sales           3033 MB      1837 MB      1977 MB          107.63%
customer                  16 MB        12 MB        12 MB          100.00%
customer_address        7008 kB      3161 kB      3115 kB           98.54%
customer_demographics     28 MB      8164 kB      9292 kB          113.82%
date_dim                3193 kB      1406 kB      1249 kB           88.85%
household_demographics    42 kB       248 kB        28 kB           11.29%
income_band            1239 bytes     225 kB      1239 bytes         0.54%
inventory                 36 MB        71 MB        36 MB           50.70%
item                    3084 kB      2479 kB      2227 kB           89.84%
promotion                 27 kB       239 kB        18 kB            7.53%
reason                 2730 bytes     226 kB      2280 bytes         0.99%
ship_mode              3894 bytes     227 kB      3315 bytes         1.43%
store                     23 kB       239 kB        18 kB            7.53%
store_returns            400 MB       265 MB       277 MB          104.53%
store_sales             4173 MB      2384 MB      2554 MB          107.12%
time_dim                1702 kB       819 kB       627 kB           76.56%
warehouse              5394 bytes     227 kB      4698 bytes         2.02%
web_page                  21 kB       236 kB        14 kB            5.93%
web_returns              116 MB        83 MB        85 MB          102.41%
web_sales               1513 MB       908 MB       982 MB          108.15%

Author: gongxun0928

contrib/pax_storage/.gitignore

@@ -12,6 +12,7 @@
 Thumbs.db
 
 # Temp files dir
+bench_data
 .tmp/**
 build*/**
 results/**

contrib/pax_storage/src/cpp/cmake/pax.cmake

@@ -51,6 +51,7 @@ set(pax_storage_src
     storage/columns/pax_dict_encoding.cc
     storage/columns/pax_decoding.cc
     storage/columns/pax_encoding.cc
+    storage/columns/pax_delta_encoding.cc
     storage/columns/pax_rlev2_decoding.cc
     storage/columns/pax_rlev2_encoding.cc
     storage/columns/pax_vec_bitpacked_column.cc

contrib/pax_storage/src/cpp/cmake/pax_format.cmake

@@ -41,6 +41,7 @@ set(pax_storage_src
     storage/columns/pax_dict_encoding.cc
     storage/columns/pax_decoding.cc
     storage/columns/pax_encoding.cc
+    storage/columns/pax_delta_encoding.cc
     storage/columns/pax_rlev2_decoding.cc
     storage/columns/pax_rlev2_encoding.cc
     storage/columns/pax_vec_column.cc

contrib/pax_storage/src/cpp/pax_gbench.cc

@@ -25,12 +25,310 @@
  *-------------------------------------------------------------------------
  */
 
+#include "pax_gbench.h"
+
+#include "comm/cbdb_api.h"
+
 #include <benchmark/benchmark.h>
 
-static void example_benchmark(benchmark::State &state) {
+#include <functional>
+#include <memory>
+#include <vector>
+
+#include "access/paxc_rel_options.h"
+#include "comm/cbdb_wrappers.h"
+#include "cpp-stub/src/stub.h"
+#include "storage/micro_partition_iterator.h"
+#include "storage/pax.h"
+#include "storage/strategy.h"
+
+namespace pax::bench {
+
+// Create memory context for benchmark
+void CreateMemoryContext() {
+  MemoryContext test_memory_context = AllocSetContextCreate(
+      (MemoryContext)NULL, "TestMemoryContext", 80 * 1024 * 1024,
+      80 * 1024 * 1024, 80 * 1024 * 1024);
+  MemoryContextSwitchTo(test_memory_context);
+}
+
+// Global registry
+class BenchmarkRegistry {
+ private:
+  std::vector<InitFunction> init_functions_;
+  std::vector<CleanupFunction> cleanup_functions_;
+  bool initialized_ = false;
+
+ public:
+  void RegisterInitFunction(InitFunction func) {
+    init_functions_.push_back(func);
+  }
+
+  void RegisterCleanupFunction(CleanupFunction func) {
+    cleanup_functions_.push_back(func);
+  }
+
+  void RunAllInitFunctions() {
+    if (initialized_) return;
+
+    printf("Running PAX Benchmark Suite...\n");
+    printf("Initializing all benchmark modules...\n\n");
+
+    for (const auto &func : init_functions_) {
+      func();
+    }
+    initialized_ = true;
+  }
+
+  void RunAllCleanupFunctions() {
+    if (!initialized_) return;
+
+    printf("\nCleaning up all benchmark modules...\n");
+
+    // Cleanup functions executed in reverse order
+    for (auto it = cleanup_functions_.rbegin(); it != cleanup_functions_.rend();
+         ++it) {
+      (*it)();
+    }
+    initialized_ = false;
+  }
+};
+
+// Global registry access function
+BenchmarkRegistry &GetBenchmarkRegistry() {
+  static BenchmarkRegistry instance;
+  return instance;
+}
+
+// Registration functions
+void RegisterBenchmarkInit(InitFunction func) {
+  GetBenchmarkRegistry().RegisterInitFunction(func);
+}
+
+void RegisterBenchmarkCleanup(CleanupFunction func) {
+  GetBenchmarkRegistry().RegisterCleanupFunction(func);
+}
+
+// Global Mock functions for benchmark framework
+bool MockMinMaxGetStrategyProcinfo(Oid, Oid, Oid *, FmgrInfo *,
+                                   StrategyNumber) {
+  return false;
+}
+
+int32 MockGetFastSequences(Oid) {
+  static int32 mock_id = 0;
+  return mock_id++;
+}
+
+void MockInsertMicroPartitionPlaceHolder(Oid, int) {}
+void MockDeleteMicroPartitionEntry(Oid, Snapshot, int) {}
+void MockExecStoreVirtualTuple(TupleTableSlot *) {}
+
+std::string MockBuildPaxDirectoryPath(RelFileNode rnode, BackendId backend_id) {
+  // Create a simple file path for benchmarks
+  return std::string("./bench_data");
+}
+
+std::vector<int> MockGetMinMaxColumnIndexes(Relation) {
+  return std::vector<int>();
+}
+
+std::vector<int> MockBloomFilterColumnIndexes(Relation) {
+  return std::vector<int>();
+}
+
+std::vector<std::tuple<ColumnEncoding_Kind, int>> MockGetRelEncodingOptions(
+    Relation relation) {
+  std::vector<std::tuple<ColumnEncoding_Kind, int>> encoding_opts;
+
+  // Get number of columns from relation
+  int num_columns = 10;  // default for benchmark
+  if (relation && relation->rd_att) {
+    num_columns = relation->rd_att->natts;
+  }
+
+  // Create encoding options for each column (NO_ENCODED, 0)
+  for (int i = 0; i < num_columns; i++) {
+    encoding_opts.emplace_back(
+        std::make_tuple(ColumnEncoding_Kind_NO_ENCODED, 0));
+  }
+
+  return encoding_opts;
+}
+
+// Mock TupleDescInitEntry that doesn't rely on SYSCACHE
+void MockTupleDescInitEntry(TupleDesc desc, AttrNumber attributeNumber,
+                            const char *attributeName, Oid oidtypeid,
+                            int32 typmod, int attdim) {
+  // Basic validation
+  if (attributeNumber < 1 || attributeNumber > desc->natts) {
+    return;
+  }
+
+  Form_pg_attribute att = TupleDescAttr(desc, attributeNumber - 1);
+
+  // Set basic attribute properties
+  namestrcpy(&(att->attname), attributeName);
+  att->atttypid = oidtypeid;
+  att->atttypmod = typmod;
+  att->attndims = attdim;
+  att->attnum = attributeNumber;
+  att->attnotnull = false;
+  att->atthasdef = false;
+  att->attidentity = '\0';
+  att->attgenerated = '\0';
+  att->attisdropped = false;
+  att->attislocal = true;
+  att->attinhcount = 0;
+  att->attcollation = InvalidOid;
+
+  // Set type-specific properties based on OID (hardcoded for common types)
+  switch (oidtypeid) {
+    case INT2OID:  // smallint
+      att->attlen = 2;
+      att->attalign = 's';
+      att->attstorage = 'p';
+      att->attbyval = true;
+      break;
+    case INT4OID:  // integer
+      att->attlen = 4;
+      att->attalign = 'i';
+      att->attstorage = TYPSTORAGE_PLAIN;
+      att->attbyval = true;
+      break;
+    case INT8OID:  // bigint
+      att->attlen = 8;
+      att->attalign = 'd';
+      att->attstorage = TYPSTORAGE_PLAIN;
+      att->attbyval = FLOAT8PASSBYVAL;
+      break;
+    case FLOAT8OID:  // double precision
+      att->attlen = 8;
+      att->attalign = 'd';
+      att->attstorage = 'p';
+      att->attbyval = FLOAT8PASSBYVAL;
+      break;
+    case BOOLOID:  // boolean
+      att->attlen = 1;
+      att->attalign = 'c';
+      att->attstorage = 'p';
+      att->attbyval = true;
+      break;
+    case TEXTOID:  // text
+      att->attlen = -1;
+      att->attalign = 'i';
+      att->attstorage = TYPSTORAGE_PLAIN;
+      att->attbyval = false;
+      att->attcollation = DEFAULT_COLLATION_OID;
+      break;
+    case NUMERICOID:  // numeric
+      att->attlen = -1;
+      att->attalign = TYPALIGN_INT;
+      att->attstorage = TYPSTORAGE_PLAIN;
+      att->attbyval = false;
+      break;
+    case TIMESTAMPOID:  // timestamp
+      att->attlen = 8;
+      att->attalign = 'd';
+      att->attstorage = TYPSTORAGE_PLAIN;
+      att->attbyval = FLOAT8PASSBYVAL;
+      break;
+    default:
+      // Default values for unknown types
+      att->attlen = -1;
+      att->attalign = 'i';
+      att->attstorage = 'p';
+      att->attbyval = false;
+      break;
+  }
+}
+
+// Global initialization function for general benchmark framework
+void GlobalBenchmarkInit() {
+  static bool global_initialized = false;
+  if (global_initialized) return;
+
+  printf("Initializing PAX benchmark framework...\n");
+
+  // Initialize memory context
+  MemoryContextInit();
+
+  // Setup global Mock functions
+  static std::unique_ptr<Stub> stub_global = std::make_unique<Stub>();
+
+  stub_global->set(MinMaxGetPgStrategyProcinfo, MockMinMaxGetStrategyProcinfo);
+  stub_global->set(CPaxGetFastSequences, MockGetFastSequences);
+  stub_global->set(cbdb::BuildPaxDirectoryPath, MockBuildPaxDirectoryPath);
+  stub_global->set(cbdb::InsertMicroPartitionPlaceHolder,
+                   MockInsertMicroPartitionPlaceHolder);
+  stub_global->set(cbdb::DeleteMicroPartitionEntry,
+                   MockDeleteMicroPartitionEntry);
+  stub_global->set(cbdb::GetMinMaxColumnIndexes, MockGetMinMaxColumnIndexes);
+  stub_global->set(cbdb::GetBloomFilterColumnIndexes,
+                   MockBloomFilterColumnIndexes);
+  stub_global->set(cbdb::GetRelEncodingOptions, MockGetRelEncodingOptions);
+  stub_global->set(ExecStoreVirtualTuple, MockExecStoreVirtualTuple);
+  stub_global->set(TupleDescInitEntry, MockTupleDescInitEntry);
+
+  // Create basic test directory
+  system("mkdir -p ./bench_data");
+
+  global_initialized = true;
+  printf("PAX benchmark framework initialized.\n");
+}
+
+// Global cleanup function for general benchmark framework
+void GlobalBenchmarkCleanup() {
+  printf("Cleaning up PAX benchmark framework...\n");
+
+  // Clean up test directory
+  // system("rm -rf ./bench_data");
+
+  // Reset memory context
+  if (TopMemoryContext) {
+    MemoryContextReset(TopMemoryContext);
+  }
+
+  printf("PAX benchmark framework cleaned up.\n");
+}
+
+// Example benchmark test
+static void example_benchmark(::benchmark::State &state) {
   for (auto _ : state) {
+    // Empty example test
   }
 }
 BENCHMARK(example_benchmark);
 
-BENCHMARK_MAIN();
+}  // namespace pax::benchmark
+
+// Global cleanup function (C-style for atexit)
+static void cleanup_all() {
+  pax::bench::GetBenchmarkRegistry().RunAllCleanupFunctions();
+  pax::bench::GlobalBenchmarkCleanup();
+}
+
+// Main entry function
+int main(int argc, char **argv) {
+  // Register global cleanup function
+  std::atexit(cleanup_all);
+
+  // Global initialization
+  pax::bench::GlobalBenchmarkInit();
+
+  // Run all registered initialization functions
+  pax::bench::GetBenchmarkRegistry().RunAllInitFunctions();
+
+  // Initialize benchmark framework
+  ::benchmark::Initialize(&argc, argv);
+  if (::benchmark::ReportUnrecognizedArguments(argc, argv)) return 1;
+
+  printf("\n=== Starting PAX Benchmark Suite ===\n");
+  printf("Use --benchmark_filter=<pattern> to run specific tests\n");
+  printf("Use --benchmark_list_tests to see all available tests\n\n");
+
+  // Run benchmark
+  ::benchmark::RunSpecifiedBenchmarks();
+
+  return 0;
+}