service-capacity-modeling 0.3.74__tar.gz → 0.3.76__tar.gz

{service_capacity_modeling-0.3.74 → service_capacity_modeling-0.3.76}/PKG-INFO

@@ -1,19 +1,22 @@
 Metadata-Version: 2.4
 Name: service-capacity-modeling
-Version: 0.3.74
+Version: 0.3.76
 Summary: Contains utilities for modeling capacity for pluggable workloads
 Author: Joseph Lynch
 Author-email: josephl@netflix.com
 License: Apache 2.0
 Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
 Classifier: License :: OSI Approved :: Apache Software License
 Classifier: Operating System :: OS Independent
+Requires-Python: >=3.9,<3.12
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: pydantic>2.0
 Requires-Dist: scipy
 Requires-Dist: numpy
-Requires-Dist: importlib_resources; python_version < "3.7"
 Requires-Dist: isodate
 Provides-Extra: aws
 Requires-Dist: boto3; extra == "aws"
@@ -26,6 +29,7 @@ Dynamic: license
 Dynamic: license-file
 Dynamic: provides-extra
 Dynamic: requires-dist
+Dynamic: requires-python
 Dynamic: summary
 
 # Service Capacity Modeling
@@ -43,10 +47,10 @@ remember this repository is public when making changes to it.
 Run the tests:
 ```bash
 # Test the capacity planner on included netflix models
-$ tox -e py38
+$ tox -e py39
 
 # Run a single test with a debugger attached if the test fails
-$ .tox/py38/bin/pytest -n0 -k test_java_heap_heavy --pdb --pdbcls=IPython.terminal.debugger:Pdb
+$ .tox/py39/bin/pytest -n0 -k test_java_heap_heavy --pdb --pdbcls=IPython.terminal.debugger:Pdb
 
 # Verify all type contracts
 $ tox -e mypy
@@ -247,7 +251,7 @@ To contribute to this project:
 2. Write a unit test using `pytest` in the `tests` folder.
 3. Ensure your tests pass via `tox` or debug them with:
 ```
-tox -e py38 -- -k test_<your_functionality> --pdb --pdbcls=IPython.terminal.debugger:Pdb
+tox -e py39 -- -k test_<your_functionality> --pdb --pdbcls=IPython.terminal.debugger:Pdb
 ```
 
 ### Pre-commit / Linting

{service_capacity_modeling-0.3.74 → service_capacity_modeling-0.3.76}/README.md

@@ -13,10 +13,10 @@ remember this repository is public when making changes to it.
 Run the tests:
 ```bash
 # Test the capacity planner on included netflix models
-$ tox -e py38
+$ tox -e py39
 
 # Run a single test with a debugger attached if the test fails
-$ .tox/py38/bin/pytest -n0 -k test_java_heap_heavy --pdb --pdbcls=IPython.terminal.debugger:Pdb
+$ .tox/py39/bin/pytest -n0 -k test_java_heap_heavy --pdb --pdbcls=IPython.terminal.debugger:Pdb
 
 # Verify all type contracts
 $ tox -e mypy
@@ -217,7 +217,7 @@ To contribute to this project:
 2. Write a unit test using `pytest` in the `tests` folder.
 3. Ensure your tests pass via `tox` or debug them with:
 ```
-tox -e py38 -- -k test_<your_functionality> --pdb --pdbcls=IPython.terminal.debugger:Pdb
+tox -e py39 -- -k test_<your_functionality> --pdb --pdbcls=IPython.terminal.debugger:Pdb
 ```
 
 ### Pre-commit / Linting

{service_capacity_modeling-0.3.74 → service_capacity_modeling-0.3.76}/service_capacity_modeling/interface.py

@@ -264,14 +264,14 @@ class Drive(ExcludeUnsetModel):
         return max(self.block_size_kib, self.group_size_kib)
 
     @property
-    def max_size_gib(self):
+    def max_size_gib(self) -> int:
         if self.max_scale_size_gib != 0:
             return self.max_scale_size_gib
         else:
             return self.size_gib
 
     @property
-    def max_io_per_s(self):
+    def max_io_per_s(self) -> int:
         if self.max_scale_io_per_s != 0:
             return self.max_scale_io_per_s
         else:
@@ -779,23 +779,48 @@ class BufferComponent(str, Enum):
     compute = "compute"
     # [Data Shape]    a.k.a. "Dataset" related buffers, e.g. Disk and Memory
     storage = "storage"
-
     # Resource specific component
     cpu = "cpu"
     network = "network"
     disk = "disk"
     memory = "memory"
 
+    @staticmethod
+    def is_generic(component: str) -> bool:
+        return component in {BufferComponent.compute, BufferComponent.storage}
+
+    @staticmethod
+    def is_specific(component: str) -> bool:
+        return not BufferComponent.is_generic(component)
+
 
 class BufferIntent(str, Enum):
     # Most buffers show "desired" buffer, this is the default
     desired = "desired"
     # ratio on top of existing buffers to ensure exists. Generally combined
     # with a different desired buffer to ensure we don't just scale needlessly
+    # This means we can scale up or down as as long as we meet the desired buffer.
     scale = "scale"
-    # Ignore model preferences, just preserve existing buffers
+
+    # DEPRECATED: Use scale_up/scale_down instead
+    # Ignores model preferences, just preserve existing buffers
+    # We rarely actually want to do this since it can cause severe over provisioning
     preserve = "preserve"
 
+    # Scale up if necessary to meet the desired buffer.
+    # If the existing resource is over-provisioned, do not reduce the requirement.
+    # If under-provisioned, the requirement can be increased to meet the desired buffer.
+    # Example: need 20 cores but have 10 → scale up to 20 cores.
+    # Example 2: need 20 cores but have 40 → do not scale down and require at
+    # least 40 cores
+    scale_up = "scale_up"
+    # Scale down if necessary to meet the desired buffer.
+    # If the existing resource is under-provisioned, do not increase the requirement.
+    # If over-provisioned, the requirement can be decreased to meet the desired buffer.
+    # Example: need 20 cores but have 10 → maintain buffer and do not scale up.
+    # Example 2: need 20 cores but have 40 → scale down to 20 cores.
+    scale_down = "scale_down"
+
 
 class Buffer(ExcludeUnsetModel):
     # The value of the buffer expressed as a ratio over "normal" load e.g. 1.5x
@@ -819,7 +844,6 @@ class Buffers(ExcludeUnsetModel):
             "compute": Buffer(ratio: 1.5),
         }
     )
-
     And then models layer in their buffers, for example if a workload
     requires 10 CPU cores, but the operator of that workload  likes to build in
     2x buffer for background work (20 cores provisioned), they would express that