diff --git a/.github/workflows/common-server-ci.yml b/.github/workflows/common-server-ci.yml
index 966f4e5..e6639bb 100644
--- a/.github/workflows/common-server-ci.yml
+++ b/.github/workflows/common-server-ci.yml
@@ -20,7 +20,7 @@ jobs:
   Server-Tests:
     strategy:
       matrix:
-        python_version: ['3.8', '3.9', '3.10', '3.11', '3.12']
+        python_version: ['3.9', '3.10', '3.11', '3.12']
         pool_vmImage: ['ubuntu-latest', 'windows-latest']
     runs-on: ${{matrix.pool_vmImage}}
     environment: Server-CI
diff --git a/.github/workflows/inference-http-server-E2E-ci.yml b/.github/workflows/inference-http-server-E2E-ci.yml
index 8455fd8..8964003 100644
--- a/.github/workflows/inference-http-server-E2E-ci.yml
+++ b/.github/workflows/inference-http-server-E2E-ci.yml
@@ -20,7 +20,7 @@ jobs:
   azureml-inference-http-server-E2E-CI:
     strategy:
       matrix:
-        python_version: ['3.8', '3.9', '3.10', '3.11', '3.12']
+        python_version: ['3.9', '3.10', '3.11', '3.12']
         pool_vmImage: ['ubuntu-latest', 'windows-latest']
     runs-on: ${{matrix.pool_vmImage}}
     environment: Server-CI
diff --git a/.github/workflows/publish.yml b/.github/workflows/publish.yml
index 609ac85..1b95726 100644
--- a/.github/workflows/publish.yml
+++ b/.github/workflows/publish.yml
@@ -15,7 +15,7 @@ jobs:
     - name: Setup python version
       uses: actions/setup-python@v4
       with:
-        python-version: '3.8'
+        python-version: '3.9'
 
     - name: Build wheels
       run: |  
diff --git a/CHANGELOG.rst b/CHANGELOG.rst
index a4f5d5b..267fe00 100644
--- a/CHANGELOG.rst
+++ b/CHANGELOG.rst
@@ -1,3 +1,31 @@
+1.4.0 (2024-11-13)
+~~~~~~~~~~~~~~~~~~
+Azureml_Inference_Server_Http 1.4.0 (2024-11-13)
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Deprecation
+-----------
+
+- Python 3.8 has reached the end-of-maintenance update support. The lifespan notification can be found 
+  at https://peps.python.org/pep-0569/#lifespan. The Azureml_Inference_Server_Http dropped the support of Python 3.8
+  to avoid patching unsupported Python 3.8 packages.
+
+- Deprecated the previous added support for Flask 2.0. A compatibility layer was introduced to ensure the flask 2.0 upgrade
+  doesn't break the users who use ``@rawhttp`` as the methods on the Flask request object. Specifically,
+
+  * ``request.headers.has_keys()`` was removed
+  * ``request.json`` throws an exception if the content-type is not "application/json". Previously it returns ``None``.
+
+  The compatibility layer which restored these functionalities to their previous behaviors is now removed. Users 
+  are encouraged to audit their score scripts and migrate your score script to be compatible with Flask 2.
+ 
+  Flask's full changelog can be found here: https://flask.palletsprojects.com/en/2.1.x/changes/
+
+Enhancements
+------------
+
+- Upgraded waitress (only windows) package to 3.0.1
+
 1.3.4 (2024-10-21)
 ~~~~~~~~~~~~~~~~~~
 Azureml_Inference_Server_Http 1.3.4 (2024-10-21)
diff --git a/README.md b/README.md
index 1b8866b..90018e5 100644
--- a/README.md
+++ b/README.md
@@ -9,7 +9,7 @@ This is the Flask server or the Sanic server code. The azureml-inference-server-
 ### <a name="virtualenv">Setting your environment</a>
 
 - Clone the [azureml-inference-server](https://github.com/microsoft/azureml-inference-server) repository.
-- Install [Python 3.8](https://www.python.org/downloads/).
+- Install [Python 3.12](https://www.python.org/downloads/).
 - Install the virtualenv python package with `pip install virtualenv`.
 - Create a new virtual environment with `virtualenv <env name>`, for example `virtualenv amlinf`.
 - Activate the new environment.
diff --git a/azureml_inference_server_http/_version.py b/azureml_inference_server_http/_version.py
index 0e8d75c..fc19b39 100644
--- a/azureml_inference_server_http/_version.py
+++ b/azureml_inference_server_http/_version.py
@@ -1,4 +1,4 @@
 # Copyright (c) Microsoft Corporation.
 # Licensed under the MIT License.
 
-__version__ = "1.3.4"
+__version__ = "1.4.0"
diff --git a/azureml_inference_server_http/server/appinsights_client.py b/azureml_inference_server_http/server/appinsights_client.py
index 4736484..972265e 100644
--- a/azureml_inference_server_http/server/appinsights_client.py
+++ b/azureml_inference_server_http/server/appinsights_client.py
@@ -72,7 +72,7 @@ def log_app_insights_exception(self, ex):
 
     def send_model_data_log(self, request_id, client_request_id, model_input, prediction):
         try:
-            if not self.enabled or not config.model_dc_storage_enabled:
+            if not self.enabled or not config.mdc_storage_enabled:
                 return
             properties = {
                 "custom_dimensions": {
diff --git a/azureml_inference_server_http/server/config.py b/azureml_inference_server_http/server/config.py
index 8dc200f..544b639 100644
--- a/azureml_inference_server_http/server/config.py
+++ b/azureml_inference_server_http/server/config.py
@@ -29,7 +29,6 @@
     "SERVICE_PATH_PREFIX": "service_path_prefix",
     "SERVICE_VERSION": "service_version",
     "SCORING_TIMEOUT_MS": "scoring_timeout",
-    "AML_FLASK_ONE_COMPATIBILITY": "flask_one_compatibility",
     "AZUREML_LOG_LEVEL": "log_level",
     "AML_APP_INSIGHTS_ENABLED": "app_insights_enabled",
     "AML_APP_INSIGHTS_KEY": "app_insights_key",
@@ -119,9 +118,6 @@ class AMLInferenceServerConfig(BaseSettings):
     # Dictates how long scoring function with run before timeout in milliseconds.
     scoring_timeout: int = pydantic.Field(default=3600 * 1000, alias="SCORING_TIMEOUT_MS")
 
-    # When @rawhttp is used, whether the user requires on `request` object to have the flask v1 properties/behavior.
-    flask_one_compatibility: bool = pydantic.Field(default=True)
-
     # Sets the Logging level
     log_level: str = pydantic.Field(default="INFO", alias="AZUREML_LOG_LEVEL")
 
@@ -132,7 +128,7 @@ class AMLInferenceServerConfig(BaseSettings):
     app_insights_key: Optional[pydantic.SecretStr] = pydantic.Field(default=None)
 
     # Whether to enable model data collection
-    model_dc_storage_enabled: bool = pydantic.Field(default=False)
+    mdc_storage_enabled: bool = pydantic.Field(default=False)
 
     # Whether to log response to AppInsights
     app_insights_log_response_enabled: bool = pydantic.Field(default=True, alias="APP_INSIGHTS_LOG_RESPONSE_ENABLED")
diff --git a/azureml_inference_server_http/server/create_app.py b/azureml_inference_server_http/server/create_app.py
index 5ac283b..f6f093e 100644
--- a/azureml_inference_server_http/server/create_app.py
+++ b/azureml_inference_server_http/server/create_app.py
@@ -1,69 +1,18 @@
 # Copyright (c) Microsoft Corporation.
 # Licensed under the MIT License.
 
-from distutils.version import LooseVersion
-import functools
 import importlib
 import logging
-import traceback
-import warnings
 
-import flask
 from flask import Flask
-import werkzeug
-import werkzeug.datastructures
 from werkzeug.exceptions import HTTPException
 
 from azureml_inference_server_http.api.aml_response import AMLResponse
 from . import routes
-from .config import config
 
 logger = logging.getLogger("azmlinfsrv")
 
 
-def patch_flask():
-    # While "packaging" is the recommended package for comparing versions, we can't introduce the dependency because
-    # we need our server to work even when the user doesn't have latest dependencies installed.
-    with warnings.catch_warnings():
-        warnings.filterwarnings(
-            action="ignore",
-            category=DeprecationWarning,
-            message="distutils Version classes are deprecated.",
-        )
-        patch_werkzeug = LooseVersion(werkzeug.__version__) >= LooseVersion("2.1")
-
-    if patch_werkzeug:
-        # Request.headers.has_key() was removed in werkzeug 2.1
-        # https://github.com/pallets/werkzeug/commit/03979aaff2b8020fd6fd52e69745950d484e3fa5
-        # Restore the functionality to preserve backwards compatability.
-        werkzeug.datastructures.EnvironHeaders.has_key = werkzeug.datastructures.EnvironHeaders.__contains__
-        werkzeug.datastructures.CombinedMultiDict.has_key = werkzeug.datastructures.CombinedMultiDict.__contains__
-
-        # In werkzeug 2.1, get_json() is modified to throw BadRequest if the content type is not application/json.
-        @functools.wraps(flask.Request.on_json_loading_failed)
-        def on_json_loading_failed(self, e):
-            if e is None:
-                return None
-
-            return on_json_loading_failed.__wrapped__(self, e)
-
-        flask.Request.on_json_loading_failed = on_json_loading_failed
-        logger.info("AML_FLASK_ONE_COMPATIBILITY is set. Patched Flask to ensure compatibility with Flask 1.")
-    else:
-        logger.info("AML_FLASK_ONE_COMPATIBILITY is set, but patching is not necessary.")
-
-
-if config.flask_one_compatibility:
-    try:
-        patch_flask()
-    except Exception:
-        logger.warning(
-            "AML_FLASK_ONE_COMPATIBILITY is set. However, compatibility patch for Flask 1 has failed. "
-            "This is only a problem if you use @rawhttp and relies on deprecated methods such as has_key().\n"
-            + traceback.format_exc()
-        )
-
-
 def create():
     app = Flask(__name__)
     # To create a new instance of main_blueprint each time the create() is called
diff --git a/docs/AzureMLInferenceServer.md b/docs/AzureMLInferenceServer.md
index d7dfea9..402c2ef 100644
--- a/docs/AzureMLInferenceServer.md
+++ b/docs/AzureMLInferenceServer.md
@@ -247,7 +247,6 @@ Config file will support only below keys:
 | SERVICE_PATH_PREFIX    | No  | "" |
 | SERVICE_VERSION  | No| "1.0"  |
 | SCORING_TIMEOUT_MS | No  | 3600 * 1000  |
-| AML_FLASK_ONE_COMPATIBILITY  | No  | "True"  |
 | AZUREML_LOG_LEVEL    | No  | "INFO" |
 | AML_APP_INSIGHTS_ENABLED  | No | False |
 | AML_APP_INSIGHTS_KEY | No  | None  |
diff --git a/pyproject.toml b/pyproject.toml
index 49faceb..8cc885f 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -8,7 +8,7 @@ markers = [
 
 [tool.black]
   line-length = 119
-  target-version = ['py38']
+  target-version = ['py39']
   extend-exclude = '''
   ^/(
       env/
diff --git a/setup.py b/setup.py
index 3b25cdc..563afb5 100644
--- a/setup.py
+++ b/setup.py
@@ -51,7 +51,6 @@ def get_license():
     classifiers=[
         "Programming Language :: Python",
         "Programming Language :: Python :: 3",
-        "Programming Language :: Python :: 3.8",
         "Programming Language :: Python :: 3.9",
         "Programming Language :: Python :: 3.10",
         "Programming Language :: Python :: 3.11",
@@ -63,9 +62,9 @@ def get_license():
         "Operating System :: MacOS",
         "Operating System :: POSIX :: Linux",
     ],
-    python_requires=">=3.8",
+    python_requires=">=3.9",
     install_requires=[
-        "flask<=2.3.2",  # We aim to be compatible with both flask 1 and 2
+        "flask~=3.0.0",
         "flask-cors~=5.0.0",
         'gunicorn>=23.0.0; platform_system!="Windows"',
         "inference-schema~=1.8.0",
@@ -73,7 +72,7 @@ def get_license():
         'psutil<6.0.0; platform_system=="Windows"',
         "pydantic~=2.9.0",
         "pydantic-settings",
-        'waitress==2.1.2; platform_system=="Windows"',
+        'waitress>=3.0.1; platform_system=="Windows"',
         "werkzeug>=3.0.3",  # Werkzeug 3.x breaks back-compatibility of urls package
         "certifi>=2024.7.4",  # Python (Pip) Security Update for certifi (GHSA-248v-346w-9cwc)
     ],
diff --git a/tests/azmlinfsrv/resources/valid_score.py b/tests/azmlinfsrv/resources/valid_score.py
index fedf70c..8791b6a 100644
--- a/tests/azmlinfsrv/resources/valid_score.py
+++ b/tests/azmlinfsrv/resources/valid_score.py
@@ -38,9 +38,10 @@ def run(input_data):
         }
 
         r['pid'] = os.getpid()
-        if input_data.headers.has_key('sleep-in-sec'):
-            print(f'sleep-in-sec: {input_data.headers["sleep-in-sec"]}')
-            time.sleep(float(input_data.headers['sleep-in-sec']))
+        if 'sleep-in-sec' in r['received_headers']:
+            sleep = r['received_headers']['sleep-in-sec']
+            print(f'sleep-in-sec: {sleep}')
+            time.sleep(float(sleep))
 
         run_time = time.time() - start_time
         return AMLResponse(r, 200, {"pid": str(os.getpid()), "x-ms-run-function-duration": str(run_time)}, json_str=True)
diff --git a/tests/server/conftest.py b/tests/server/conftest.py
index 2bf4d36..2f52b2d 100644
--- a/tests/server/conftest.py
+++ b/tests/server/conftest.py
@@ -57,7 +57,7 @@ def app_cors(config):
 @pytest.fixture()
 def app_appinsights(config):
     config.app_insights_enabled = True
-    config.model_dc_storage_enabled = True
+    config.mdc_storage_enabled = True
     config.app_insights_key = pydantic.SecretStr(str(uuid.uuid4()))
     return create_app()
 
diff --git a/tests/server/test_compat.py b/tests/server/test_compat.py
index d6f1d3b..550e3f1 100644
--- a/tests/server/test_compat.py
+++ b/tests/server/test_compat.py
@@ -10,22 +10,6 @@
 from .common import TestingClient
 
 
-def test_compat_flask2_json(app: flask.Flask, client: TestingClient):
-    """Ensure that `request.json` does not throw in flask 2 when compatibility flag is set."""
-
-    @app.set_user_run
-    @rawhttp
-    def run(request: flask.Request):
-        # This property decodes the request data as JSON. Before flask 2 this property returns `None` when the request
-        # content-type is not json. Flask 2 modified the behavior to throw when content-type is not json.
-        return AMLResponse(str(request.json), 200)
-
-    data = json.dumps({"a": 1})
-    response = client.post_score(data=data)
-    assert response.status_code == 200
-    assert response.data == b"None"
-
-
 def test_compat_flask2_json_error(app: flask.Flask, client: TestingClient):
     """Ensure that `request.json` correctly calls on_json_loading_failed() when it cannot parse the input JSON."""