Skip to content

perf: Parallelize read calls by table and batch #4619

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
franciscojavierarceo merged 4 commits into from
Oct 12, 2024
Merged

perf: Parallelize read calls by table and batch #4619

Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
3b37b8a
put all dynamo calls in an asyncio.gather
robhowley Oct 11, 2024
b3b7ed1
fix: indexing
robhowley Oct 11, 2024
3a30284
parallelize the per table lookups
robhowley Oct 12, 2024
e6fcd78
ruff
robhowley Oct 12, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Next Next commit
put all dynamo calls in an asyncio.gather 
Signed-off-by: Rob Howley <howley.robert@gmail.com>
  • Loading branch information
@robhowley
robhowley committed Oct 12, 2024
commit 3b37b8a8fb07fde3efc77874827828e2400d8661
56 changes: 35 additions & 21 deletions sdk/python/feast/infra/online_stores/dynamodb.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,7 @@
# 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.
import asyncio
import itertools
import logging
from datetime import datetime
Expand Down Expand Up @@ -297,36 +298,49 @@ async def online_read_async(
batch_size = online_config.batch_size
entity_ids = self._to_entity_ids(config, entity_keys)
entity_ids_iter = iter(entity_ids)
result: List[Tuple[Optional[datetime], Optional[Dict[str, ValueProto]]]] = []
table_name = _get_table_name(online_config, config, table)

deserialize = TypeDeserializer().deserialize

def to_tbl_resp(raw_client_response):
return {
"entity_id": deserialize(raw_client_response["entity_id"]),
"event_ts": deserialize(raw_client_response["event_ts"]),
"values": deserialize(raw_client_response["values"]),
}
entity_id_batches = []
while True:
batch = list(itertools.islice(entity_ids_iter, batch_size))
if not batch:
break
entity_id_batch = self._to_client_batch_get_payload(
online_config, table_name, batch
)
entity_id_batches.append(entity_id_batch)
Comment on lines +312 to +322
Copy link
Copy Markdown
Contributor Author

@robhowley robhowley Oct 12, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

construct the batches of ids/entity_ids that we'll be looking up


async with self._get_aiodynamodb_client(online_config.region) as client:
while True:
batch = list(itertools.islice(entity_ids_iter, batch_size))

# No more items to insert
if len(batch) == 0:
break
batch_entity_ids = self._to_client_batch_get_payload(
online_config, table_name, batch
)

async def get_and_format(entity_id_batch):
def to_tbl_resp(raw_client_response):
return {
"entity_id": deserialize(raw_client_response["entity_id"]),
"event_ts": deserialize(raw_client_response["event_ts"]),
"values": deserialize(raw_client_response["values"]),
}

response = await client.batch_get_item(
RequestItems=batch_entity_ids,
RequestItems=entity_id_batch,
)
batch_result = self._process_batch_get_response(
table_name, response, entity_ids, batch, to_tbl_response=to_tbl_resp
return self._process_batch_get_response(
table_name,
response,
entity_ids,
batch,
to_tbl_response=to_tbl_resp,
)
result.extend(batch_result)
return result

result_batches = await asyncio.gather(
*[
get_and_format(entity_id_batch)
for entity_id_batch in entity_id_batches
]
)
Comment on lines +325 to +332
Copy link
Copy Markdown
Contributor Author

@robhowley robhowley Oct 12, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

make those batch requests in parallel.

note: gather maintains order


return list(itertools.chain(*result_batches))

def _get_aioboto_session(self):
if self._aioboto_session is None:
Expand Down