diff --git a/pom.xml b/pom.xml
index af793cb30..5a417f740 100644
--- a/pom.xml
+++ b/pom.xml
@@ -95,6 +95,13 @@
 			<version>3.23.1-GA</version>
 			<optional>true</optional>
 		</dependency>
+		<dependency>
+			<groupId>io.netty</groupId>
+			<artifactId>netty-all</artifactId>
+			<version>4.0.20.Final</version>
+			<scope>provided</scope>
+			<optional>true</optional>
+		</dependency>
 		<!-- Testing frameworks and related dependencies -->
 		<dependency>
 			<groupId>junit</groupId>
diff --git a/pom4ide.xml b/pom4ide.xml
index f2e731d0c..bdaa1a9c7 100644
--- a/pom4ide.xml
+++ b/pom4ide.xml
@@ -95,6 +95,13 @@
 			<version>3.23.1-GA</version>
 			<optional>true</optional>
 		</dependency>
+		<dependency>
+			<groupId>io.netty</groupId>
+			<artifactId>netty-all</artifactId>
+			<version>4.0.20.Final</version>
+			<scope>provided</scope>
+			<optional>true</optional>
+		</dependency>
 		<!-- Testing frameworks and related dependencies -->
 		<dependency>
 			<groupId>junit</groupId>
diff --git a/src/main/java/com/alibaba/ttl/TransmittableThreadLocal.java b/src/main/java/com/alibaba/ttl/TransmittableThreadLocal.java
index c310157eb..d7e35797e 100644
--- a/src/main/java/com/alibaba/ttl/TransmittableThreadLocal.java
+++ b/src/main/java/com/alibaba/ttl/TransmittableThreadLocal.java
@@ -1,5 +1,8 @@
 package com.alibaba.ttl;
 
+import com.alibaba.ttl.internal.TtlValue;
+import com.alibaba.ttl.internal.TtlValueFactory;
+
 import java.util.HashMap;
 import java.util.Iterator;
 import java.util.Map;
@@ -14,6 +17,10 @@
  * <p>
  * Note: {@link TransmittableThreadLocal} extends {@link java.lang.InheritableThreadLocal},
  * so {@link TransmittableThreadLocal} first is a {@link java.lang.InheritableThreadLocal}.
+ * <p>
+ * If you have netty in the runtime and {@link io.netty.util.internal.FastThreadLocal} is supported. You can store
+ * {@link TransmittableThreadLocal} in {@link io.netty.util.internal.FastThreadLocal} to give up inheritance for better performance.
+ * </p>
  *
  * @author Jerry Lee (oldratlee at gmail dot com)
  * @see TtlRunnable
@@ -23,6 +30,9 @@
 public class TransmittableThreadLocal<T> extends InheritableThreadLocal<T> {
     private static final Logger logger = Logger.getLogger(TransmittableThreadLocal.class.getName());
 
+    // Hold the real value to support special optimization for thread local
+    private final TtlValue<T> ttlValue = TtlValueFactory.create();
+
     /**
      * Computes the value for this transmittable thread-local variable
      * as a function of the source thread's value at the time the task
@@ -66,7 +76,7 @@ protected void afterExecute() {
 
     @Override
     public final T get() {
-        T value = super.get();
+        T value = ttlValue != null ? ttlValue.get() : super.get();
         if (null != value) {
             addValue();
         }
@@ -75,7 +85,7 @@ public final T get() {
 
     @Override
     public final void set(T value) {
-        super.set(value);
+        setTtlValue(value);
         if (null == value) { // may set null to remove value
             removeValue();
         } else {
@@ -83,10 +93,22 @@ public final void set(T value) {
         }
     }
 
+    private void setTtlValue(T value) {
+        if (ttlValue != null) {
+            ttlValue.set(value);
+        } else {
+            super.set(value);
+        }
+    }
+
     @Override
     public final void remove() {
         removeValue();
-        super.remove();
+        if (ttlValue != null) {
+            ttlValue.remove();
+        } else {
+            super.remove();
+        }
     }
 
     private void superRemove() {
diff --git a/src/main/java/com/alibaba/ttl/internal/FastThreadLocalValue.java b/src/main/java/com/alibaba/ttl/internal/FastThreadLocalValue.java
new file mode 100644
index 000000000..de2992caa
--- /dev/null
+++ b/src/main/java/com/alibaba/ttl/internal/FastThreadLocalValue.java
@@ -0,0 +1,29 @@
+package com.alibaba.ttl.internal;
+
+import io.netty.util.internal.FastThreadLocal;
+
+/**
+ * {@link FastThreadLocal} implementation for {@link com.alibaba.ttl.TransmittableThreadLocal} value holder
+ *
+ * @author Yang Fang (snoop dot fy at gmail dot com)
+ * @since 2.7.0
+ */
+public class FastThreadLocalValue<T> implements TtlValue<T> {
+
+    private FastThreadLocal<T> holder = new FastThreadLocal<T>();
+
+    @Override
+    public T get() {
+        return holder.get();
+    }
+
+    @Override
+    public void set(T t) {
+        holder.set(t);
+    }
+
+    @Override
+    public void remove() {
+        holder.remove();
+    }
+}
diff --git a/src/main/java/com/alibaba/ttl/internal/TtlValue.java b/src/main/java/com/alibaba/ttl/internal/TtlValue.java
new file mode 100644
index 000000000..d8c41496e
--- /dev/null
+++ b/src/main/java/com/alibaba/ttl/internal/TtlValue.java
@@ -0,0 +1,17 @@
+package com.alibaba.ttl.internal;
+
+/**
+ * Hold {@link com.alibaba.ttl.TransmittableThreadLocal} value, can be implemented in varied ways depending on the runtime.
+ *
+ * @author Yang Fang (snoop dot fy at gmail dot com)
+ * @since 2.7.0
+ */
+public interface TtlValue<T> {
+
+    T get();
+
+    void set(T t);
+
+    void remove();
+
+}
diff --git a/src/main/java/com/alibaba/ttl/internal/TtlValueFactory.java b/src/main/java/com/alibaba/ttl/internal/TtlValueFactory.java
new file mode 100644
index 000000000..5e95dfa92
--- /dev/null
+++ b/src/main/java/com/alibaba/ttl/internal/TtlValueFactory.java
@@ -0,0 +1,49 @@
+package com.alibaba.ttl.internal;
+
+/**
+ * {@link com.alibaba.ttl.TransmittableThreadLocal} value holder factory.
+ *
+ * <p>
+ * If there is netty in the runtime and
+ * {@link io.netty.util.internal.FastThreadLocal} is supported. You can just set
+ *
+ * <pre>-Dttl.fastthreadlocal.enable=true</pre>
+ *
+ * to enable FastThreadLocal mode for better performance.
+ * </p>
+ *
+ * <p>
+ * Caution: If FastThreadLocal mode is enabled, {@link com.alibaba.ttl.TransmittableThreadLocal}
+ * will NEVER be inheritable!
+ * </p>
+ *
+ * @author Yang Fang (snoop dot fy at gmail dot com)
+ * @since 2.7.0
+ */
+public class TtlValueFactory {
+
+    private static final String FAST_THREAD_LOCAL_ENABLE = "ttl.fastthreadlocal.enable";
+
+    private static final boolean IS_FAST_THREAD_LOCAL_SUPPORT = isClassPresent("io.netty.util.internal.FastThreadLocal");
+
+    public static <T> TtlValue<T> create() {
+        if (isFastThreadLocalEnabled() && IS_FAST_THREAD_LOCAL_SUPPORT) {
+            return new FastThreadLocalValue<T>();
+        } else {
+            return null;
+        }
+    }
+
+    public static boolean isFastThreadLocalEnabled() {
+        return Boolean.parseBoolean(System.getProperty(FAST_THREAD_LOCAL_ENABLE, "false"));
+    }
+
+    private static boolean isClassPresent(String className) {
+        try {
+            Class.forName(className);
+            return true;
+        } catch (ClassNotFoundException e) {
+            return false;
+        }
+    }
+}
diff --git a/src/test/java/com/alibaba/ttl/TtlCallableTest.kt b/src/test/java/com/alibaba/ttl/TtlCallableTest.kt
index a4c275c1d..2ae1ebb82 100644
--- a/src/test/java/com/alibaba/ttl/TtlCallableTest.kt
+++ b/src/test/java/com/alibaba/ttl/TtlCallableTest.kt
@@ -1,6 +1,7 @@
 package com.alibaba.ttl
 
 import com.alibaba.*
+import com.alibaba.ttl.internal.TtlValueFactory
 import com.alibaba.ttl.testmodel.Call
 import org.hamcrest.CoreMatchers.containsString
 import org.hamcrest.CoreMatchers.instanceOf
@@ -33,12 +34,14 @@ class TtlCallableTest {
         val ret = ttlCallable.call()
         assertEquals("ok", ret)
 
+        // inheritable is not supported by FastThreadLocal mode
+        if (!TtlValueFactory.isFastThreadLocalEnabled()) {
+            // child Inheritable
+            assertChildTtlValues("1", call.copied)
 
-        // child Inheritable
-        assertChildTtlValues("1", call.copied)
-
-        // child do not effect parent
-        assertParentTtlValues(copyTtlValues(ttlInstances))
+            // child do not effect parent
+            assertParentTtlValues(copyTtlValues(ttlInstances))
+        }
     }
 
     @Test
diff --git a/src/test/java/com/alibaba/ttl/TtlRunnableTest.kt b/src/test/java/com/alibaba/ttl/TtlRunnableTest.kt
index db6575cfa..da0ec54a1 100644
--- a/src/test/java/com/alibaba/ttl/TtlRunnableTest.kt
+++ b/src/test/java/com/alibaba/ttl/TtlRunnableTest.kt
@@ -1,6 +1,7 @@
 package com.alibaba.ttl
 
 import com.alibaba.*
+import com.alibaba.ttl.internal.TtlValueFactory
 import com.alibaba.ttl.testmodel.*
 import org.hamcrest.CoreMatchers.containsString
 import org.hamcrest.CoreMatchers.instanceOf
@@ -33,11 +34,14 @@ class TtlRunnableTest {
         ttlRunnable.run()
 
 
-        // child Inheritable
-        assertChildTtlValues("1", task.copied)
+        // inheritable is not supported by FastThreadLocal mode
+        if (!TtlValueFactory.isFastThreadLocalEnabled()) {
+            // child Inheritable
+            assertChildTtlValues("1", task.copied)
 
-        // child do not effect parent
-        assertParentTtlValues(copyTtlValues(ttlInstances))
+            // child do not effect parent
+            assertParentTtlValues(copyTtlValues(ttlInstances))
+        }
     }
 
     @Test
@@ -55,8 +59,11 @@ class TtlRunnableTest {
         thread1.join()
 
 
-        // child Inheritable
-        assertChildTtlValues("1", task.copied)
+        // inheritable is not supported by FastThreadLocal mode
+        if (!TtlValueFactory.isFastThreadLocalEnabled()) {
+            // child Inheritable
+            assertChildTtlValues("1", task.copied)
+        }
 
         // child do not effect parent
         assertParentTtlValues(copyTtlValues(ttlInstances))
@@ -77,8 +84,11 @@ class TtlRunnableTest {
         submit.get()
 
 
-        // child Inheritable
-        assertChildTtlValues("1", task.copied)
+        // inheritable is not supported by FastThreadLocal mode
+        if (!TtlValueFactory.isFastThreadLocalEnabled()) {
+            // child Inheritable
+            assertChildTtlValues("1", task.copied)
+        }
 
         // child do not effect parent
         assertParentTtlValues(copyTtlValues(ttlInstances))
@@ -102,8 +112,11 @@ class TtlRunnableTest {
         submit.get()
 
 
-        // child Inheritable
-        assertChildTtlValues("1", task.copied)
+        // inheritable is not supported by FastThreadLocal mode
+        if (!TtlValueFactory.isFastThreadLocalEnabled()) {
+            // child Inheritable
+            assertChildTtlValues("1", task.copied)
+        }
 
         // child do not effect parent
         assertParentTtlValues(copyTtlValues(ttlInstances))
@@ -168,11 +181,14 @@ class TtlRunnableTest {
         val submit = executorService.submit(ttlRunnable)
         submit.get()
 
-        // child Inheritable
-        assertEquals(3, task.copied.size.toLong())
-        assertEquals(FooPojo(PARENT_CREATE_UNMODIFIED_IN_CHILD, 1), task.copied[PARENT_CREATE_UNMODIFIED_IN_CHILD])
-        assertEquals(FooPojo(PARENT_CREATE_MODIFIED_IN_CHILD + "1", 2), task.copied[PARENT_CREATE_MODIFIED_IN_CHILD])
-        assertEquals(FooPojo(CHILD_CREATE + 1, 3), task.copied[CHILD_CREATE + 1])
+        // inheritable is not supported by FastThreadLocal mode
+        if (!TtlValueFactory.isFastThreadLocalEnabled()) {
+            // child Inheritable
+            assertEquals(3, task.copied.size.toLong())
+            assertEquals(FooPojo(PARENT_CREATE_UNMODIFIED_IN_CHILD, 1), task.copied[PARENT_CREATE_UNMODIFIED_IN_CHILD])
+            assertEquals(FooPojo(PARENT_CREATE_MODIFIED_IN_CHILD + "1", 2), task.copied[PARENT_CREATE_MODIFIED_IN_CHILD])
+            assertEquals(FooPojo(CHILD_CREATE + 1, 3), task.copied[CHILD_CREATE + 1])
+        }
 
         // child do not effect parent
         val copied = copyTtlValues(ttlInstances)
diff --git a/src/test/java/com/alibaba/ttl/reported_bugs/Bug70_Test.kt b/src/test/java/com/alibaba/ttl/reported_bugs/Bug70_Test.kt
index 08b836f6b..d12a93cff 100644
--- a/src/test/java/com/alibaba/ttl/reported_bugs/Bug70_Test.kt
+++ b/src/test/java/com/alibaba/ttl/reported_bugs/Bug70_Test.kt
@@ -2,6 +2,7 @@ package com.alibaba.ttl.reported_bugs
 
 import com.alibaba.ttl.TransmittableThreadLocal
 import com.alibaba.ttl.TtlRunnable
+import com.alibaba.ttl.internal.TtlValueFactory
 import org.junit.Assert.assertEquals
 import org.junit.Test
 import java.util.concurrent.Executors
@@ -29,15 +30,18 @@ class Bug70_Test {
                 .get()
         assertEquals(hello, task1.get())
 
-        val taskRef = AtomicReference<FutureTask<String>>()
-        val thread = Thread {
-            val task2 = FutureTask<String> { threadLocal.get() }
-            val runnable = TtlRunnable.get(task2, false, false)
-            executorService.submit(runnable)
-            taskRef.set(task2)
+        // inheritable is not supported by FastThreadLocal mode
+        if (!TtlValueFactory.isFastThreadLocalEnabled()) {
+            val taskRef = AtomicReference<FutureTask<String>>()
+            val thread = Thread {
+                val task2 = FutureTask<String> { threadLocal.get() }
+                val runnable = TtlRunnable.get(task2, false, false)
+                executorService.submit(runnable)
+                taskRef.set(task2)
+            }
+            thread.start()
+            thread.join()
+            assertEquals(hello, taskRef.get().get())
         }
-        thread.start()
-        thread.join()
-        assertEquals(hello, taskRef.get().get())
     }
 }