diff --git a/cpp/core/CMakeLists.txt b/cpp/core/CMakeLists.txt
index fb7c714a3ef89..73330ade181a1 100644
--- a/cpp/core/CMakeLists.txt
+++ b/cpp/core/CMakeLists.txt
@@ -325,7 +325,7 @@ else()
 endif()
 
 target_link_libraries(gluten
-    PUBLIC Arrow::arrow Arrow::parquet)
+    PUBLIC Arrow::arrow Arrow::parquet glog::glog)
 
 install(TARGETS gluten
         DESTINATION ${CMAKE_INSTALL_LIBDIR})
diff --git a/cpp/core/benchmarks/CompressionBenchmark.cc b/cpp/core/benchmarks/CompressionBenchmark.cc
index 3e2396818235c..1966c8ef1c189 100644
--- a/cpp/core/benchmarks/CompressionBenchmark.cc
+++ b/cpp/core/benchmarks/CompressionBenchmark.cc
@@ -25,6 +25,7 @@
 #include <arrow/type_fwd.h>
 #include <benchmark/benchmark.h>
 #include <execinfo.h>
+#include <glog/logging.h>
 #include <parquet/arrow/reader.h>
 #include <parquet/file_reader.h>
 #include <sched.h>
@@ -126,7 +127,7 @@ class BenchmarkCompression {
       }
       case gluten::kQatZstd: {
         ipcWriteOptions.codec = createArrowIpcCodec(arrow::Compression::ZSTD, CodecBackend::QAT);
-        std::cout << "load qatzstd" << std::endl;
+        LOG(INFO) << "load qatzstd";
         break;
       }
 #endif
@@ -163,7 +164,7 @@ class BenchmarkCompression {
         state);
     auto endTime = std::chrono::steady_clock::now();
     auto totalTime = (endTime - startTime).count();
-    std::cout << "Thread " << state.thread_index() << " took " << (1.0 * totalTime / 1e9) << "s" << std::endl;
+    LOG(INFO) << "Thread " << state.thread_index() << " took " << (1.0 * totalTime / 1e9) << "s";
 
     state.counters["rowgroups"] =
         benchmark::Counter(rowGroupIndices_.size(), benchmark::Counter::kAvgThreads, benchmark::Counter::OneK::kIs1000);
@@ -301,8 +302,8 @@ class BenchmarkCompressionCacheScanBenchmark final : public BenchmarkCompression
       }
     } while (recordBatch);
 
-    std::cout << "parquet parse done elapsed time " << elapseRead / 1e6 << " ms " << std::endl;
-    std::cout << "batches = " << numBatches << " rows = " << numRows << std::endl;
+    LOG(INFO) << "parquet parse done elapsed time " << elapseRead / 1e6 << " ms ";
+    LOG(INFO) << "batches = " << numBatches << " rows = " << numRows;
 
     std::vector<std::shared_ptr<arrow::ipc::IpcPayload>> payloads(batches.size());
     std::vector<std::vector<int64_t>> uncompressedBufferSize(batches.size());
@@ -418,16 +419,16 @@ int main(int argc, char** argv) {
     } else if (strcmp(argv[i], "--file") == 0) {
       datafile = argv[i + 1];
     } else if (strcmp(argv[i], "--qat-gzip") == 0) {
-      std::cout << "QAT gzip is used as codec" << std::endl;
+      LOG(INFO) << "QAT gzip is used as codec";
       codec = gluten::kQatGzip;
     } else if (strcmp(argv[i], "--qat-zstd") == 0) {
-      std::cout << "QAT zstd is used as codec" << std::endl;
+      LOG(INFO) << "QAT zstd is used as codec";
       codec = gluten::kQatZstd;
     } else if (strcmp(argv[i], "--qpl-gzip") == 0) {
-      std::cout << "QPL gzip is used as codec" << std::endl;
+      LOG(INFO) << "QPL gzip is used as codec";
       codec = gluten::kQplGzip;
     } else if (strcmp(argv[i], "--zstd") == 0) {
-      std::cout << "CPU zstd is used as codec" << std::endl;
+      LOG(INFO) << "CPU zstd is used as codec";
       codec = gluten::kZstd;
     } else if (strcmp(argv[i], "--buffer-size") == 0) {
       compressBufferSize = atol(argv[i + 1]);
@@ -435,9 +436,9 @@ int main(int argc, char** argv) {
       cpuOffset = atol(argv[i + 1]);
     }
   }
-  std::cout << "iterations = " << iterations << std::endl;
-  std::cout << "threads = " << threads << std::endl;
-  std::cout << "datafile = " << datafile << std::endl;
+  LOG(INFO) << "iterations = " << iterations;
+  LOG(INFO) << "threads = " << threads;
+  LOG(INFO) << "datafile = " << datafile;
 
   gluten::BenchmarkCompressionIterateScanBenchmark bmIterateScan(datafile, compressBufferSize);
   gluten::BenchmarkCompressionCacheScanBenchmark bmCacheScan(datafile, compressBufferSize);
diff --git a/cpp/core/compute/Runtime.h b/cpp/core/compute/Runtime.h
index beb9bb9cc1ef3..62281304e9e8f 100644
--- a/cpp/core/compute/Runtime.h
+++ b/cpp/core/compute/Runtime.h
@@ -17,6 +17,8 @@
 
 #pragma once
 
+#include <glog/logging.h>
+
 #include "compute/ProtobufUtils.h"
 #include "compute/ResultIterator.h"
 #include "memory/ArrowMemoryPool.h"
diff --git a/cpp/core/jni/JniCommon.h b/cpp/core/jni/JniCommon.h
index 4d4bf5e3fe564..fd953992781bd 100644
--- a/cpp/core/jni/JniCommon.h
+++ b/cpp/core/jni/JniCommon.h
@@ -52,7 +52,7 @@ static inline void checkException(JNIEnv* env) {
     std::string description =
         jStringToCString(env, (jstring)env->CallStaticObjectMethod(describerClass, describeMethod, t));
     if (env->ExceptionCheck()) {
-      std::cerr << "Fatal: Uncaught Java exception during calling the Java exception describer method! " << std::endl;
+      LOG(WARNING) << "Fatal: Uncaught Java exception during calling the Java exception describer method! ";
     }
     throw gluten::GlutenException("Error during calling Java code from native code: " + description);
   }
@@ -161,7 +161,7 @@ static inline void backtrace() {
   auto size = backtrace(array, 1024);
   char** strings = backtrace_symbols(array, size);
   for (size_t i = 0; i < size; ++i) {
-    std::cout << strings[i] << std::endl;
+    LOG(INFO) << strings[i];
   }
   free(strings);
 }
@@ -232,8 +232,8 @@ class SparkAllocationListener final : public gluten::AllocationListener {
   ~SparkAllocationListener() override {
     JNIEnv* env;
     if (vm_->GetEnv(reinterpret_cast<void**>(&env), jniVersion) != JNI_OK) {
-      std::cerr << "SparkAllocationListener#~SparkAllocationListener(): "
-                << "JNIEnv was not attached to current thread" << std::endl;
+      LOG(WARNING) << "SparkAllocationListener#~SparkAllocationListener(): "
+                   << "JNIEnv was not attached to current thread";
       return;
     }
     env->DeleteGlobalRef(jListenerGlobalRef_);
@@ -332,8 +332,8 @@ class CelebornClient : public RssClient {
   ~CelebornClient() {
     JNIEnv* env;
     if (vm_->GetEnv(reinterpret_cast<void**>(&env), jniVersion) != JNI_OK) {
-      std::cerr << "CelebornClient#~CelebornClient(): "
-                << "JNIEnv was not attached to current thread" << std::endl;
+      LOG(WARNING) << "CelebornClient#~CelebornClient(): "
+                   << "JNIEnv was not attached to current thread";
       return;
     }
     env->DeleteGlobalRef(javaCelebornShuffleWriter_);
diff --git a/cpp/core/jni/JniWrapper.cc b/cpp/core/jni/JniWrapper.cc
index e311da0d61776..1bb496ef108d6 100644
--- a/cpp/core/jni/JniWrapper.cc
+++ b/cpp/core/jni/JniWrapper.cc
@@ -18,7 +18,6 @@
 #include <jni.h>
 #include <filesystem>
 
-#include <glog/logging.h>
 #include "compute/ProtobufUtils.h"
 #include "compute/Runtime.h"
 #include "config/GlutenConfig.h"
@@ -798,7 +797,7 @@ JNIEXPORT jlong JNICALL Java_io_glutenproject_vectorized_ShuffleWriterJniWrapper
   jobject thread = env->CallStaticObjectMethod(cls, mid);
   checkException(env);
   if (thread == NULL) {
-    std::cerr << "Thread.currentThread() return NULL" << std::endl;
+    LOG(WARNING) << "Thread.currentThread() return NULL";
   } else {
     jmethodID midGetid = getMethodIdOrError(env, cls, "getId", "()J");
     jlong sid = env->CallLongMethod(thread, midGetid);
diff --git a/cpp/core/tests/RoundRobinPartitionerTest.cc b/cpp/core/tests/RoundRobinPartitionerTest.cc
index 5fb3e00feb191..c500c025a8995 100644
--- a/cpp/core/tests/RoundRobinPartitionerTest.cc
+++ b/cpp/core/tests/RoundRobinPartitionerTest.cc
@@ -15,6 +15,7 @@
  * limitations under the License.
  */
 #include "shuffle/RoundRobinPartitioner.h"
+#include <glog/logging.h>
 #include <gtest/gtest.h>
 #include <numeric>
 
@@ -46,10 +47,11 @@ class RoundRobinPartitionerTest : public ::testing::Test {
 
   template <typename T>
   void toString(const std::vector<T>& vec, const std::string& name) const {
-    std::cout << name << " = [";
-    std::copy(vec.cbegin(), vec.cend(), std::ostream_iterator<T>(std::cout, ","));
-    std::cout << " ]";
-    std::cout << std::endl;
+    std::stringstream ss;
+    ss << name << " = [";
+    std::copy(vec.cbegin(), vec.cend(), std::ostream_iterator<T>(ss, ","));
+    ss << " ]";
+    LOG(INFO) << ss.str();
   }
 
   int32_t getPidSelection() const {
diff --git a/cpp/core/utils/ObjectStore.cc b/cpp/core/utils/ObjectStore.cc
index 4a8e7c7ea2f44..5dc9c7843a73a 100644
--- a/cpp/core/utils/ObjectStore.cc
+++ b/cpp/core/utils/ObjectStore.cc
@@ -16,6 +16,7 @@
  */
 
 #include "ObjectStore.h"
+#include <glog/logging.h>
 #include <iostream>
 
 gluten::ObjectStore::~ObjectStore() {
@@ -24,7 +25,7 @@ gluten::ObjectStore::~ObjectStore() {
   for (auto itr = aliveObjectHandles_.rbegin(); itr != aliveObjectHandles_.rend(); itr++) {
     ResourceHandle handle = *itr;
     if (store_.lookup(handle) == nullptr) {
-      std::cerr << "Fatal: resource handle " + std::to_string(handle) + " not found in store." << std::endl;
+      LOG(WARNING) << "Fatal: resource handle " + std::to_string(handle) + " not found in store.";
     }
     store_.erase(handle);
   }
diff --git a/cpp/core/utils/macros.h b/cpp/core/utils/macros.h
index 6fd0f15c04542..fef8ae1e75de9 100644
--- a/cpp/core/utils/macros.h
+++ b/cpp/core/utils/macros.h
@@ -20,6 +20,7 @@
 #include <time.h>
 
 #include <arrow/status.h>
+#include <glog/logging.h>
 #include <chrono>
 
 #include "utils/exception.h"
diff --git a/cpp/velox/benchmarks/ColumnarToRowBenchmark.cc b/cpp/velox/benchmarks/ColumnarToRowBenchmark.cc
index 0951b08bf7fc9..73af9b04eeee5 100644
--- a/cpp/velox/benchmarks/ColumnarToRowBenchmark.cc
+++ b/cpp/velox/benchmarks/ColumnarToRowBenchmark.cc
@@ -129,7 +129,7 @@ class GoogleBenchmarkColumnarToRowCacheScanBenchmark : public GoogleBenchmarkCol
     localSchema = std::make_shared<arrow::Schema>(*schema_.get());
 
     if (state.thread_index() == 0)
-      std::cout << localSchema->ToString() << std::endl;
+      LOG(INFO) << localSchema->ToString();
 
     std::unique_ptr<::parquet::arrow::FileReader> parquetReader;
     std::shared_ptr<RecordBatchReader> recordBatchReader;
@@ -148,7 +148,7 @@ class GoogleBenchmarkColumnarToRowCacheScanBenchmark : public GoogleBenchmarkCol
       }
     } while (recordBatch);
 
-    std::cout << " parquet parse done elapsed time = " << elapseRead / 1000000 << " rows = " << numRows << std::endl;
+    LOG(INFO) << " parquet parse done elapsed time = " << elapseRead / 1000000 << " rows = " << numRows;
 
     // reuse the columnarToRowConverter for batches caused system % increase a lot
     auto ctxPool = defaultLeafVeloxMemoryPool();
@@ -261,10 +261,10 @@ int main(int argc, char** argv) {
       cpu = atol(argv[i + 1]);
     }
   }
-  std::cout << "iterations = " << iterations << std::endl;
-  std::cout << "threads = " << threads << std::endl;
-  std::cout << "datafile = " << datafile << std::endl;
-  std::cout << "cpu = " << cpu << std::endl;
+  LOG(INFO) << "iterations = " << iterations;
+  LOG(INFO) << "threads = " << threads;
+  LOG(INFO) << "datafile = " << datafile;
+  LOG(INFO) << "cpu = " << cpu;
 
   gluten::GoogleBenchmarkColumnarToRowCacheScanBenchmark bck(datafile);
 
diff --git a/cpp/velox/benchmarks/GenericBenchmark.cc b/cpp/velox/benchmarks/GenericBenchmark.cc
index 02f6a98d34e06..24c066c135c06 100644
--- a/cpp/velox/benchmarks/GenericBenchmark.cc
+++ b/cpp/velox/benchmarks/GenericBenchmark.cc
@@ -183,7 +183,7 @@ auto BM_Generic = [](::benchmark::State& state,
           return;
         }
         if (FLAGS_print_result) {
-          std::cout << maybeBatch.ValueOrDie()->ToString() << std::endl;
+          LOG(INFO) << maybeBatch.ValueOrDie()->ToString();
         }
       }
 
@@ -203,7 +203,7 @@ auto BM_Generic = [](::benchmark::State& state,
     const auto& task = rawIter->task_;
     const auto& planNode = rawIter->veloxPlan_;
     auto statsStr = facebook::velox::exec::printPlanWithStats(*planNode, task->taskStats(), true);
-    std::cout << statsStr << std::endl;
+    LOG(INFO) << statsStr;
   }
   Runtime::release(runtime);
 
@@ -237,10 +237,9 @@ int main(int argc, char** argv) {
 
   try {
     if (argc < 2) {
-      std::cout << "No input args. Usage: " << std::endl
-                << "./generic_benchmark /path/to/substrait_json_file /path/to/data_file_1 /path/to/data_file_2 ..."
-                << std::endl;
-      std::cout << "Running example..." << std::endl;
+      LOG(INFO) << "No input args. Usage: " << std::endl
+                << "./generic_benchmark /path/to/substrait_json_file /path/to/data_file_1 /path/to/data_file_2 ...";
+      LOG(INFO) << "Running example...";
       inputFiles.resize(2);
       substraitJsonFile = getGeneratedFilePath("example.json");
       inputFiles[0] = getGeneratedFilePath("example_orders");
@@ -248,16 +247,16 @@ int main(int argc, char** argv) {
     } else {
       substraitJsonFile = argv[1];
       abortIfFileNotExists(substraitJsonFile);
-      std::cout << "Using substrait json file: " << std::endl << substraitJsonFile << std::endl;
-      std::cout << "Using " << argc - 2 << " input data file(s): " << std::endl;
+      LOG(INFO) << "Using substrait json file: " << std::endl << substraitJsonFile;
+      LOG(INFO) << "Using " << argc - 2 << " input data file(s): ";
       for (auto i = 2; i < argc; ++i) {
         inputFiles.emplace_back(argv[i]);
         abortIfFileNotExists(inputFiles.back());
-        std::cout << inputFiles.back() << std::endl;
+        LOG(INFO) << inputFiles.back();
       }
     }
   } catch (const std::exception& e) {
-    std::cout << "Failed to run benchmark: " << e.what() << std::endl;
+    LOG(INFO) << "Failed to run benchmark: " << e.what();
     ::benchmark::Shutdown();
     std::exit(EXIT_FAILURE);
   }
@@ -278,12 +277,12 @@ int main(int argc, char** argv) {
   } while (0)
 
 #if 0
-  std::cout << "FLAGS_threads:" << FLAGS_threads << std::endl;
-  std::cout << "FLAGS_iterations:" << FLAGS_iterations << std::endl;
-  std::cout << "FLAGS_cpu:" << FLAGS_cpu << std::endl;
-  std::cout << "FLAGS_print_result:" << FLAGS_print_result << std::endl;
-  std::cout << "FLAGS_write_file:" << FLAGS_write_file << std::endl;
-  std::cout << "FLAGS_batch_size:" << FLAGS_batch_size << std::endl;
+  LOG(INFO) << "FLAGS_threads:" << FLAGS_threads ;
+  LOG(INFO) << "FLAGS_iterations:" << FLAGS_iterations ;
+  LOG(INFO) << "FLAGS_cpu:" << FLAGS_cpu ;
+  LOG(INFO) << "FLAGS_print_result:" << FLAGS_print_result ;
+  LOG(INFO) << "FLAGS_write_file:" << FLAGS_write_file ;
+  LOG(INFO) << "FLAGS_batch_size:" << FLAGS_batch_size ;
 #endif
 
   if (FLAGS_skip_input) {
diff --git a/cpp/velox/benchmarks/ParquetWriteBenchmark.cc b/cpp/velox/benchmarks/ParquetWriteBenchmark.cc
index 3f0fb2fce0db3..42445f93a346f 100644
--- a/cpp/velox/benchmarks/ParquetWriteBenchmark.cc
+++ b/cpp/velox/benchmarks/ParquetWriteBenchmark.cc
@@ -139,7 +139,7 @@ class GoogleBenchmarkArrowParquetWriteCacheScanBenchmark : public GoogleBenchmar
     localSchema = std::make_shared<arrow::Schema>(*schema_.get());
 
     if (state.thread_index() == 0)
-      std::cout << localSchema->ToString() << std::endl;
+      LOG(INFO) << localSchema->ToString();
 
     std::unique_ptr<::parquet::arrow::FileReader> parquetReader;
     std::shared_ptr<RecordBatchReader> recordBatchReader;
@@ -158,7 +158,7 @@ class GoogleBenchmarkArrowParquetWriteCacheScanBenchmark : public GoogleBenchmar
       }
     } while (recordBatch);
 
-    std::cout << " parquet parse done elapsed time = " << elapseRead / 1000000 << " rows = " << numRows << std::endl;
+    LOG(INFO) << " parquet parse done elapsed time = " << elapseRead / 1000000 << " rows = " << numRows;
 
     // reuse the ParquetWriteConverter for batches caused system % increase a lot
     auto fileName = "arrow_parquet_write.parquet";
@@ -232,7 +232,7 @@ class GoogleBenchmarkVeloxParquetWriteCacheScanBenchmark : public GoogleBenchmar
     localSchema = std::make_shared<arrow::Schema>(*schema_.get());
 
     if (state.thread_index() == 0)
-      std::cout << localSchema->ToString() << std::endl;
+      LOG(INFO) << localSchema->ToString();
 
     std::unique_ptr<::parquet::arrow::FileReader> parquetReader;
     std::shared_ptr<RecordBatchReader> recordBatchReader;
@@ -251,7 +251,7 @@ class GoogleBenchmarkVeloxParquetWriteCacheScanBenchmark : public GoogleBenchmar
       }
     } while (recordBatch);
 
-    std::cout << " parquet parse done elapsed time = " << elapseRead / 1000000 << " rows = " << numRows << std::endl;
+    LOG(INFO) << " parquet parse done elapsed time = " << elapseRead / 1000000 << " rows = " << numRows;
 
     // reuse the ParquetWriteConverter for batches caused system % increase a lot
     auto fileName = "velox_parquet_write.parquet";
@@ -323,11 +323,11 @@ int main(int argc, char** argv) {
       output = (argv[i + 1]);
     }
   }
-  std::cout << "iterations = " << iterations << std::endl;
-  std::cout << "threads = " << threads << std::endl;
-  std::cout << "datafile = " << datafile << std::endl;
-  std::cout << "cpu = " << cpu << std::endl;
-  std::cout << "output = " << output << std::endl;
+  LOG(INFO) << "iterations = " << iterations;
+  LOG(INFO) << "threads = " << threads;
+  LOG(INFO) << "datafile = " << datafile;
+  LOG(INFO) << "cpu = " << cpu;
+  LOG(INFO) << "output = " << output;
 
   gluten::GoogleBenchmarkVeloxParquetWriteCacheScanBenchmark bck(datafile, output);
 
diff --git a/cpp/velox/benchmarks/QueryBenchmark.cc b/cpp/velox/benchmarks/QueryBenchmark.cc
index 9c29edd815caa..021981ab1124e 100644
--- a/cpp/velox/benchmarks/QueryBenchmark.cc
+++ b/cpp/velox/benchmarks/QueryBenchmark.cc
@@ -101,7 +101,7 @@ auto BM = [](::benchmark::State& state,
         state.SkipWithError(maybeBatch.status().message().c_str());
         return;
       }
-      std::cout << maybeBatch.ValueOrDie()->ToString() << std::endl;
+      LOG(INFO) << maybeBatch.ValueOrDie()->ToString();
     }
   }
   Runtime::release(runtime);
diff --git a/cpp/velox/benchmarks/ShuffleSplitBenchmark.cc b/cpp/velox/benchmarks/ShuffleSplitBenchmark.cc
index c4aec9650e1b7..a1e5eb9066502 100644
--- a/cpp/velox/benchmarks/ShuffleSplitBenchmark.cc
+++ b/cpp/velox/benchmarks/ShuffleSplitBenchmark.cc
@@ -254,7 +254,7 @@ class BenchmarkShuffleSplitCacheScanBenchmark : public BenchmarkShuffleSplit {
     localSchema = std::make_shared<arrow::Schema>(fields);
 
     if (state.thread_index() == 0)
-      std::cout << localSchema->ToString() << std::endl;
+      LOG(INFO) << localSchema->ToString();
 
     auto pool = options.memory_pool;
     GLUTEN_ASSIGN_OR_THROW(
@@ -279,8 +279,8 @@ class BenchmarkShuffleSplitCacheScanBenchmark : public BenchmarkShuffleSplit {
         numRows += recordBatch->num_rows();
       }
     } while (recordBatch);
-    std::cout << "parquet parse done elapsed time " << elapseRead / 1000000 << " ms " << std::endl;
-    std::cout << "batches = " << numBatches << " rows = " << numRows << std::endl;
+    LOG(INFO) << "parquet parse done elapsed time " << elapseRead / 1000000 << " ms ";
+    LOG(INFO) << "batches = " << numBatches << " rows = " << numRows;
 
     for (auto _ : state) {
       for_each(
@@ -291,8 +291,8 @@ class BenchmarkShuffleSplitCacheScanBenchmark : public BenchmarkShuffleSplit {
             ARROW_ASSIGN_OR_THROW(cb, recordBatch2VeloxColumnarBatch(*recordBatch));
             TIME_NANO_OR_THROW(splitTime, shuffleWriter->split(cb, ShuffleWriter::kMinMemLimit));
           });
-      // std::cout << " split done memory allocated = " <<
-      // options.memory_pool->bytes_allocated() << std::endl;
+      // LOG(INFO) << " split done memory allocated = " <<
+      // options.memory_pool->bytes_allocated() ;
     }
 
     TIME_NANO_OR_THROW(splitTime, shuffleWriter->stop());
@@ -315,7 +315,7 @@ class BenchmarkShuffleSplitIterateScanBenchmark : public BenchmarkShuffleSplit {
       ShuffleWriterOptions options,
       benchmark::State& state) {
     if (state.thread_index() == 0)
-      std::cout << schema_->ToString() << std::endl;
+      LOG(INFO) << schema_->ToString();
 
     GLUTEN_ASSIGN_OR_THROW(
         shuffleWriter,
@@ -353,7 +353,7 @@ int main(int argc, char** argv) {
   gflags::ParseCommandLineFlags(&argc, &argv, true);
 
   if (FLAGS_file.size() == 0) {
-    std::cerr << "No input data file. Please specify via argument --file" << std::endl;
+    LOG(WARNING) << "No input data file. Please specify via argument --file";
   }
 
   if (FLAGS_partitions == -1) {
diff --git a/cpp/velox/benchmarks/common/BenchmarkUtils.cc b/cpp/velox/benchmarks/common/BenchmarkUtils.cc
index 00dc804389fe4..f7e0a6ec4c2dc 100644
--- a/cpp/velox/benchmarks/common/BenchmarkUtils.cc
+++ b/cpp/velox/benchmarks/common/BenchmarkUtils.cc
@@ -117,7 +117,7 @@ bool checkPathExists(const std::string& filepath) {
 
 void abortIfFileNotExists(const std::string& filepath) {
   if (!checkPathExists(filepath)) {
-    std::cerr << "File path does not exist: " << filepath << std::endl;
+    LOG(WARNING) << "File path does not exist: " << filepath;
     ::benchmark::Shutdown();
     std::exit(EXIT_FAILURE);
   }
@@ -148,7 +148,7 @@ void setCpu(uint32_t cpuindex) {
   CPU_ZERO(&cs);
   CPU_SET(cpuindex, &cs);
   if (sched_setaffinity(0, sizeof(cs), &cs) == -1) {
-    std::cerr << "Error binding CPU " << std::to_string(cpuindex) << std::endl;
+    LOG(WARNING) << "Error binding CPU " << std::to_string(cpuindex);
     exit(EXIT_FAILURE);
   }
 }
diff --git a/cpp/velox/benchmarks/common/BenchmarkUtils.h b/cpp/velox/benchmarks/common/BenchmarkUtils.h
index 16a22cdb1d50f..dc1221e994ed4 100644
--- a/cpp/velox/benchmarks/common/BenchmarkUtils.h
+++ b/cpp/velox/benchmarks/common/BenchmarkUtils.h
@@ -18,11 +18,13 @@
 #pragma once
 
 #include <arrow/c/abi.h>
+#include <glog/logging.h>
 #include <cstdlib>
 #include <filesystem>
 #include <fstream>
 #include <thread>
 #include <utility>
+
 #include "benchmark/benchmark.h"
 #include "substrait/SubstraitToVeloxPlan.h"
 
diff --git a/cpp/velox/benchmarks/common/ParquetReaderIterator.h b/cpp/velox/benchmarks/common/ParquetReaderIterator.h
index f6621719786a9..b315aff3e6266 100644
--- a/cpp/velox/benchmarks/common/ParquetReaderIterator.h
+++ b/cpp/velox/benchmarks/common/ParquetReaderIterator.h
@@ -34,7 +34,7 @@ class ParquetReaderIterator : public FileReaderIterator {
         fileReader_->GetRecordBatchReader(arrow::internal::Iota(fileReader_->num_row_groups()), &recordBatchReader_));
 
     auto schema = recordBatchReader_->schema();
-    std::cout << "schema:\n" << schema->ToString() << std::endl;
+    LOG(INFO) << "schema:\n" << schema->ToString();
   }
 
   std::shared_ptr<arrow::Schema> getSchema() override {
diff --git a/cpp/velox/substrait/SubstraitToVeloxPlanValidator.cc b/cpp/velox/substrait/SubstraitToVeloxPlanValidator.cc
index 2b0e85522836c..110a871a09c5f 100644
--- a/cpp/velox/substrait/SubstraitToVeloxPlanValidator.cc
+++ b/cpp/velox/substrait/SubstraitToVeloxPlanValidator.cc
@@ -69,8 +69,7 @@ bool validateColNames(const ::substrait::NamedStruct& schema) {
     for (auto i = 0; i < name.size(); i++) {
       auto c = name[i];
       if (!isUnquotedPathCharacter(c)) {
-        std::cout << "native validation failed due to: Illegal column charactor " << c << "in column " << name
-                  << std::endl;
+        LOG(WARNING) << "native validation failed due to: Illegal column charactor " << c << "in column " << name;
         return false;
       }
     }
diff --git a/cpp/velox/udf/examples/MyUDF.cpp b/cpp/velox/udf/examples/MyUDF.cpp
index 91ae9f753146c..91d9e295d485d 100644
--- a/cpp/velox/udf/examples/MyUDF.cpp
+++ b/cpp/velox/udf/examples/MyUDF.cpp
@@ -92,5 +92,5 @@ DEFINE_REGISTER_UDF {
       "myudf1", integerSignatures(), std::make_unique<PlusConstantFunction<facebook::velox::TypeKind::INTEGER>>(5));
   facebook::velox::exec::registerVectorFunction(
       "myudf2", bigintSignatures(), std::make_unique<PlusConstantFunction<facebook::velox::TypeKind::BIGINT>>(5));
-  std::cout << "registered myudf1, myudf2" << std::endl;
+  LOG(INFO) << "registered myudf1, myudf2";
 }