#
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements. See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
"""An extensible run inference transform.
Users of this module can extend the ModelLoader class for any MLframework. Then
pass their extended ModelLoader object into RunInference to create a
RunInference Beam transform for that framework.
The transform will handle standard inference functionality like metric
collection, sharing model between threads and batching elements.
Note: This module is still actively being developed and users should have
no expectation that these interfaces will not change.
"""
import logging
import os
import pickle
import platform
import sys
import time
from typing import Any
from typing import Generic
from typing import Iterable
from typing import List
from typing import TypeVar
import apache_beam as beam
from apache_beam.utils import shared
try:
# pylint: disable=wrong-import-order, wrong-import-position
import resource
except ImportError:
resource = None # type: ignore[assignment]
_MICROSECOND_TO_MILLISECOND = 1000
_NANOSECOND_TO_MICROSECOND = 1000
_SECOND_TO_MICROSECOND = 1_000_000
T = TypeVar('T')
[docs]class InferenceRunner():
"""Implements running inferences for a framework."""
[docs] def run_inference(self, batch: List[Any], model: Any) -> Iterable[Any]:
"""Runs inferences on a batch of examples and
returns an Iterable of Predictions."""
raise NotImplementedError(type(self))
[docs] def get_num_bytes(self, batch: Any) -> int:
"""Returns the number of bytes of data for a batch."""
return len(pickle.dumps(batch))
[docs] def get_metrics_namespace(self) -> str:
"""Returns a namespace for metrics collected by RunInference transform."""
return 'RunInference'
[docs]class ModelLoader(Generic[T]):
"""Has the ability to load an ML model."""
[docs] def load_model(self) -> T:
"""Loads and initializes a model for processing."""
raise NotImplementedError(type(self))
[docs] def get_inference_runner(self) -> InferenceRunner:
"""Returns an implementation of InferenceRunner for this model."""
raise NotImplementedError(type(self))
[docs]class RunInference(beam.PTransform):
"""An extensible transform for running inferences."""
def __init__(self, model_loader: ModelLoader, clock=None):
self._model_loader = model_loader
self._clock = clock
# TODO(BEAM-14208): Add batch_size back off in the case there
# are functional reasons large batch sizes cannot be handled.
[docs] def expand(self, pcoll: beam.PCollection) -> beam.PCollection:
return (
pcoll
# TODO(BEAM-14044): Hook into the batching DoFn APIs.
| beam.BatchElements()
| beam.ParDo(_RunInferenceDoFn(self._model_loader, self._clock)))
class _MetricsCollector:
"""A metrics collector that tracks ML related performance and memory usage."""
def __init__(self, namespace: str):
# Metrics
self._inference_counter = beam.metrics.Metrics.counter(
namespace, 'num_inferences')
self._inference_request_batch_size = beam.metrics.Metrics.distribution(
namespace, 'inference_request_batch_size')
self._inference_request_batch_byte_size = (
beam.metrics.Metrics.distribution(
namespace, 'inference_request_batch_byte_size'))
# Batch inference latency in microseconds.
self._inference_batch_latency_micro_secs = (
beam.metrics.Metrics.distribution(
namespace, 'inference_batch_latency_micro_secs'))
self._model_byte_size = beam.metrics.Metrics.distribution(
namespace, 'model_byte_size')
# Model load latency in milliseconds.
self._load_model_latency_milli_secs = beam.metrics.Metrics.distribution(
namespace, 'load_model_latency_milli_secs')
# Metrics cache
self._load_model_latency_milli_secs_cache = None
self._model_byte_size_cache = None
def update_metrics_with_cache(self):
if self._load_model_latency_milli_secs_cache is not None:
self._load_model_latency_milli_secs.update(
self._load_model_latency_milli_secs_cache)
self._load_model_latency_milli_secs_cache = None
if self._model_byte_size_cache is not None:
self._model_byte_size.update(self._model_byte_size_cache)
self._model_byte_size_cache = None
def cache_load_model_metrics(self, load_model_latency_ms, model_byte_size):
self._load_model_latency_milli_secs_cache = load_model_latency_ms
self._model_byte_size_cache = model_byte_size
def update(
self,
examples_count: int,
examples_byte_size: int,
latency_micro_secs: int):
self._inference_batch_latency_micro_secs.update(latency_micro_secs)
self._inference_counter.inc(examples_count)
self._inference_request_batch_size.update(examples_count)
self._inference_request_batch_byte_size.update(examples_byte_size)
class _RunInferenceDoFn(beam.DoFn):
"""A DoFn implementation generic to frameworks."""
def __init__(self, model_loader: ModelLoader, clock=None):
self._model_loader = model_loader
self._inference_runner = model_loader.get_inference_runner()
self._shared_model_handle = shared.Shared()
self._metrics_collector = _MetricsCollector(
self._inference_runner.get_metrics_namespace())
self._clock = clock
if not clock:
self._clock = _ClockFactory.make_clock()
self._model = None
def _load_model(self):
def load():
"""Function for constructing shared LoadedModel."""
memory_before = _get_current_process_memory_in_bytes()
start_time = self._clock.get_current_time_in_microseconds()
model = self._model_loader.load_model()
end_time = self._clock.get_current_time_in_microseconds()
memory_after = _get_current_process_memory_in_bytes()
load_model_latency_ms = ((end_time - start_time) /
_MICROSECOND_TO_MILLISECOND)
model_byte_size = memory_after - memory_before
self._metrics_collector.cache_load_model_metrics(
load_model_latency_ms, model_byte_size)
return model
# TODO(BEAM-14207): Investigate releasing model.
return self._shared_model_handle.acquire(load)
def setup(self):
self._model = self._load_model()
def process(self, batch):
# Process supports both keyed data, and example only data.
# First keys and samples are separated (if there are keys)
has_keys = isinstance(batch[0], tuple)
if has_keys:
examples = [example for _, example in batch]
keys = [key for key, _ in batch]
else:
examples = batch
keys = None
start_time = self._clock.get_current_time_in_microseconds()
result_generator = self._inference_runner.run_inference(
examples, self._model)
predictions = list(result_generator)
inference_latency = self._clock.get_current_time_in_microseconds(
) - start_time
num_bytes = self._inference_runner.get_num_bytes(examples)
num_elements = len(batch)
self._metrics_collector.update(num_elements, num_bytes, inference_latency)
# Keys are recombined with predictions in the RunInference PTransform.
if has_keys:
yield from zip(keys, predictions)
else:
yield from predictions
def finish_bundle(self):
# TODO(BEAM-13970): Figure out why there is a cache.
self._metrics_collector.update_metrics_with_cache()
def _is_darwin() -> bool:
return sys.platform == 'darwin'
def _get_current_process_memory_in_bytes():
"""Returns memory usage in bytes."""
if resource is not None:
usage = resource.getrusage(resource.RUSAGE_SELF).ru_maxrss
if _is_darwin():
return usage
return usage * 1024
else:
logging.warning(
'Resource module is not available for current platform, '
'memory usage cannot be fetched.')
return 0
def _is_windows() -> bool:
return platform.system() == 'Windows' or os.name == 'nt'
def _is_cygwin() -> bool:
return platform.system().startswith('CYGWIN_NT')
class _Clock(object):
def get_current_time_in_microseconds(self) -> int:
return int(time.time() * _SECOND_TO_MICROSECOND)
class _FineGrainedClock(_Clock):
def get_current_time_in_microseconds(self) -> int:
return int(
time.clock_gettime_ns(time.CLOCK_REALTIME) / # type: ignore[attr-defined]
_NANOSECOND_TO_MICROSECOND)
#TODO(BEAM-14255): Research simplifying the internal clock and just using time.
class _ClockFactory(object):
@staticmethod
def make_clock() -> _Clock:
if (hasattr(time, 'clock_gettime_ns') and not _is_windows() and
not _is_cygwin()):
return _FineGrainedClock()
return _Clock()