diff --git a/configure b/configure
index 00274f1b34..e95915f2de 100755
--- a/configure
+++ b/configure
@@ -35,6 +35,8 @@ function usage()
 	echo "                           No path required."
 	echo " rdma                      [disabled]"
 	echo "                           No path required."
+	echo " vtune                     Required to profile I/O under Intel VTune Amplifier XE."
+	echo "                           example: /opt/intel/vtune_amplifier_xe_version"
 	echo ""
 }
 
@@ -115,6 +117,14 @@ for i in "$@"; do
 			FIO_SOURCE_DIR=
 			CONFIG_FIO_PLUGIN=n
 			;;
+		--with-vtune=*)
+			VTUNE_SOURCE_DIR="${i#*=}"
+			CONFIG_VTUNE=y
+			;;
+		--without-vtune)
+			VTUNE_SOURCE_DIR=
+			CONFIG_VTUNE=n
+			;;
 		--)
 			break
 			;;
@@ -145,6 +155,13 @@ if [ "$CONFIG_FIO_PLUGIN" = "y" ]; then
 	fi
 fi
 
+if [ "$CONFIG_VTUNE" = "y" ]; then
+	if [ -z "$VTUNE_SOURCE_DIR" ]; then
+		echo "When VTune is enabled, you must specify the VTune directory using --with-vtune=path"
+		exit 1
+	fi
+fi
+
 if [ "$CONFIG_ASAN" = "y" -a "$CONFIG_TSAN" = "y" ]; then
 	echo "ERROR: ASAN and TSAN cannot be enabled at the same time."
 	exit 1
@@ -193,6 +210,12 @@ fi
 if [ -n "$CONFIG_RBD" ]; then
 	echo "CONFIG_RBD?=$CONFIG_RBD" >> CONFIG.local
 fi
+if [ -n "$CONFIG_VTUNE" ]; then
+	echo "CONFIG_VTUNE?=$CONFIG_VTUNE" >> CONFIG.local
+fi
+if [ -n "$VTUNE_SOURCE_DIR" ]; then
+	echo "VTUNE_SOURCE_DIR?=$VTUNE_SOURCE_DIR" >> CONFIG.local
+fi
 
 python scripts/genconfig.py > config.h
 
diff --git a/lib/bdev/Makefile b/lib/bdev/Makefile
index 4e93af215e..539eaa01ec 100644
--- a/lib/bdev/Makefile
+++ b/lib/bdev/Makefile
@@ -35,7 +35,13 @@ SPDK_ROOT_DIR := $(abspath $(CURDIR)/../..)
 include $(SPDK_ROOT_DIR)/mk/spdk.common.mk
 
 CFLAGS += $(ENV_CFLAGS) -I.
+ifeq ($(CONFIG_VTUNE),y)
+CFLAGS += -I$(VTUNE_SOURCE_DIR)/include
+endif
+
 C_SRCS = bdev.c scsi_nvme.c
+C_SRCS-$(CONFIG_VTUNE) += $(VTUNE_SOURCE_DIR)/sdk/src/ittnotify/ittnotify_static.c
+
 LIBNAME = bdev
 
 DIRS-y += error malloc null nvme rpc split
diff --git a/lib/bdev/bdev.c b/lib/bdev/bdev.c
index fed537a058..87856da73a 100644
--- a/lib/bdev/bdev.c
+++ b/lib/bdev/bdev.c
@@ -46,6 +46,11 @@
 #include "spdk_internal/bdev.h"
 #include "spdk_internal/event.h"
 #include "spdk_internal/log.h"
+#include "spdk/string.h"
+
+#ifdef SPDK_CONFIG_VTUNE
+#include "ittnotify.h"
+#endif
 
 #define SPDK_BDEV_IO_POOL_SIZE	(64 * 1024)
 #define BUF_SMALL_POOL_SIZE	8192
@@ -63,6 +68,10 @@ struct spdk_bdev_mgr {
 	TAILQ_HEAD(, spdk_bdev_module_if) vbdev_modules;
 
 	TAILQ_HEAD(, spdk_bdev) bdevs;
+
+#ifdef SPDK_CONFIG_VTUNE
+	__itt_domain	*domain;
+#endif
 };
 
 static struct spdk_bdev_mgr g_bdev_mgr = {
@@ -89,6 +98,13 @@ struct spdk_bdev_channel {
 	struct spdk_io_channel *mgmt_channel;
 
 	struct spdk_bdev_io_stat stat;
+
+#ifdef SPDK_CONFIG_VTUNE
+	uint64_t		start_tsc;
+	uint64_t		interval_tsc;
+	__itt_string_handle	*handle;
+#endif
+
 };
 
 struct spdk_bdev *
@@ -367,8 +383,13 @@ spdk_bdev_initialize(void)
 	if (!g_bdev_mgr.buf_large_pool) {
 		SPDK_ERRLOG("create rbuf large pool failed\n");
 		rc = -1;
+		goto end;
 	}
 
+#ifdef SPDK_CONFIG_VTUNE
+	g_bdev_mgr.domain = __itt_domain_create("spdk_bdev");
+#endif
+
 	spdk_io_device_register(&g_bdev_mgr, spdk_bdev_mgmt_channel_create,
 				spdk_bdev_mgmt_channel_destroy,
 				sizeof(struct spdk_bdev_mgmt_channel));
@@ -563,6 +584,21 @@ spdk_bdev_channel_create(void *io_device, void *ctx_buf)
 	ch->mgmt_channel = spdk_get_io_channel(&g_bdev_mgr);
 	memset(&ch->stat, 0, sizeof(ch->stat));
 
+#ifdef SPDK_CONFIG_VTUNE
+	{
+		char *name;
+
+		name = spdk_sprintf_alloc("spdk_bdev_%s_%p", ch->bdev->name, ch);
+		if (!name) {
+			return -1;
+		}
+		ch->handle = __itt_string_handle_create(name);
+		free(name);
+		ch->start_tsc = spdk_get_ticks();
+		ch->interval_tsc = spdk_get_ticks_hz() / 100;
+	}
+#endif
+
 	return 0;
 }
 
@@ -957,6 +993,11 @@ void
 spdk_bdev_get_io_stat(struct spdk_bdev *bdev, struct spdk_io_channel *ch,
 		      struct spdk_bdev_io_stat *stat)
 {
+#ifdef SPDK_CONFIG_VTUNE
+	SPDK_ERRLOG("Calling spdk_bdev_get_io_stat is not allowed when VTune integration is enabled.\n");
+	memset(stat, 0, sizeof(*stat));
+	return;
+#endif
 
 	struct spdk_bdev_channel *channel = spdk_io_channel_get_ctx(ch);
 
@@ -1128,6 +1169,24 @@ spdk_bdev_io_complete(struct spdk_bdev_io *bdev_io, enum spdk_bdev_io_status sta
 		}
 	}
 
+#ifdef SPDK_CONFIG_VTUNE
+	uint64_t now_tsc = spdk_get_ticks();
+	if (now_tsc > (bdev_io->ch->start_tsc + bdev_io->ch->interval_tsc)) {
+		uint64_t data[4];
+
+		data[0] = bdev_io->ch->stat.num_read_ops;
+		data[1] = bdev_io->ch->stat.bytes_read;
+		data[2] = bdev_io->ch->stat.num_write_ops;
+		data[3] = bdev_io->ch->stat.bytes_written;
+
+		__itt_metadata_add(g_bdev_mgr.domain, __itt_null, bdev_io->ch->handle,
+				   __itt_metadata_u64, 4, data);
+
+		memset(&bdev_io->ch->stat, 0, sizeof(bdev_io->ch->stat));
+		bdev_io->ch->start_tsc = now_tsc;
+	}
+#endif
+
 	assert(bdev_io->cb != NULL);
 	bdev_io->cb(bdev_io, status == SPDK_BDEV_IO_STATUS_SUCCESS, bdev_io->caller_ctx);
 }