Python driver benchmarks by patrycja-ziemkiewicz · Pull Request #55 · scylladb/python-rs-driver

patrycja-ziemkiewicz · 2026-03-17T17:21:07Z

This PR add benchmark scenarios for the python-rs driver

Example plots generated by the benchmarker:

@wprzytula

When creating nodejs-rs driver, we have created a set of benchmarks, that we have used to compare the performance of the nodejs-rs driver with the node.js cassandra-driver and the scylla rust driver. This meant, that we had some benchmark code for rust driver inside nodejs driver repository. With two other wrappers creating benchmarks*, that are heavily based, on the benchmarks from nodejs ZPP, we have decided with @wprzytula to move the code for the rust part of the benchmarks to the rust driver repository, so that it can be used by all three wrappers and maintained in one place. * C# driver: scylladb/csharp-rs-driver#100 * Python driver: scylladb/python-rs-driver#55

@wprzytula

When creating nodejs-rs driver, we have created a set of benchmarks, that we have used to compare the performance of the nodejs-rs driver with the node.js cassandra-driver and the scylla rust driver. This meant, that we had some benchmark code for rust driver inside nodejs driver repository. With two other wrappers creating benchmarks*, that are heavily based, on the benchmarks from nodejs ZPP, we have decided with @wprzytula to move the code for the rust part of the benchmarks to the rust driver repository, so that it can be used by all three wrappers and maintained in one place. * C# driver: scylladb/csharp-rs-driver#100 * Python driver: scylladb/python-rs-driver#55

Copilot

Pull request overview

Adds an SDB (ScyllaDB Drivers Benchmarker) benchmark suite for the Python-RS driver, including single-threaded and concurrent scenarios plus documentation/config to run and plot results.

Changes:

Introduces benchmark scenario scripts for insert, select (small/large), (de)serialization, batch, and paging.
Adds concurrent variants of key scenarios (insert/select/(de)serialization/paging).
Adds SDB configuration (config.yml) and a README with instructions for running/plotting.

Reviewed changes

Copilot reviewed 16 out of 16 changed files in this pull request and generated 11 comments.

Show a summary per file

File	Description
benchmark/common.py	Shared DB setup/teardown helpers and data generators for all benchmark scenarios.
benchmark/insert.py	Insert throughput benchmark using prepared statements.
benchmark/ser.py	Serialization-focused benchmark inserting complex rows.
benchmark/deser.py	Deserialization-focused benchmark selecting complex rows.
benchmark/batch.py	Batch execution benchmark using logged batches.
benchmark/paging.py	Manual paging benchmark using `iter_current_page()` + `fetch_next_page()`.
benchmark/select_runner.py	Shared select benchmark runner used by small/large select wrappers.
benchmark/small_select.py	Wrapper to run select benchmark with a small row count.
benchmark/large_select.py	Wrapper to run select benchmark with a large row count.
benchmark/concurrent_insert.py	Concurrent insert benchmark using asyncio tasks.
benchmark/concurrent_select.py	Concurrent select benchmark using asyncio tasks.
benchmark/concurrent_ser.py	Concurrent serialization benchmark using asyncio tasks.
benchmark/concurrent_deser.py	Concurrent deserialization benchmark using asyncio tasks.
benchmark/concurrent_paging.py	Concurrent paging benchmark using asyncio tasks.
benchmark/config.yml	SDB benchmark definitions and backend run commands for this repo.
benchmark/README.md	Instructions for building SDB, running benchmarks, and plotting results.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

jakub-graczyk · 2026-05-16T22:05:09Z

+
+
+async def test(session: Session, cnt: int, num_of_rows: int):
+    prepared = await session.prepare(SIMPLE_SELECT_QUERY)


For consistency prepare outside test(), same in complex

Copilot

Pull request overview

Copilot reviewed 32 out of 34 changed files in this pull request and generated 10 comments.

+dependencies = [
+    "scylla-driver>=3.29.10",
+]


+        fut = to_asyncio(
+            session.execute_async(  # pyright: ignore[reportUnknownMemberType]
+                query=prepared,
+                parameters=get_simple_data(),


+        fut = to_asyncio(
+            session.execute_async(  # pyright: ignore[reportUnknownMemberType]
+                query=prepared,
+                parameters=get_simple_data(),


+async def test(session: Session, prepared: PreparedStatement, cnt: int, num_of_rows: int):
+    for _ in range(cnt):
+        fut = to_asyncio(
+            session.execute_async(  # pyright: ignore[reportUnknownMemberType]
+                query=prepared,
+                parameters=None,
+                paging_state=None,
+            )
+        )
+        result = await fut
+        assert len(result) == num_of_rows  # pyright: ignore[reportUnknownArgumentType, reportUnknownMemberType]
+
+


+async def test(session: Session, prepared: PreparedStatement, cnt: int, num_of_rows: int):
+    for _ in range(cnt):
+        fut = to_asyncio(
+            session.execute_async(  # pyright: ignore[reportUnknownMemberType]
+                query=prepared,
+                parameters=None,
+                paging_state=None,
+            )
+        )
+        result = await fut
+        assert len(result) == num_of_rows  # pyright: ignore[reportUnknownArgumentType, reportUnknownMemberType]
+
+


+import sys
+
+from common import (
+    SIMPLE_SELECT_QUERY,
+)
+from python_rs_helpers import connect
+from scylla.session import Session
+
+


+import sys
+
+from cassandra.cluster import Session  # pyright: ignore[reportMissingTypeStubs]
+from cassandra.query import PreparedStatement  # pyright: ignore[reportMissingTypeStubs]
+from common import (
+    SIMPLE_SELECT_QUERY,
+)
+from python_helpers import PAGE_SIZE, connect, to_asyncio
+
+


+from common import (
+    SIMPLE_INSERT_QUERY,
+    get_simple_data,
+)
+from python_rs_helpers import init_simple_table, init_with_inserts


+from scylla.session import Session
+from scylla.statement import PreparedStatement, Statement
+
+STEP = 3971


+)
+from python_helpers import connect, to_asyncio
+
+STEP = 3971


Copilot

Pull request overview

Copilot reviewed 32 out of 34 changed files in this pull request and generated 5 comments.

 dynamic = ["version"]
+dependencies = [
+    "scylla-driver>=3.29.10",
+]


+sdb -d test.db plot \
+  -b config.yml \
+  -o concurrent-result.png \
+  --series Python-RS-concurrent/ \
+  series


+    tasks = [session.execute(prepared, get_data(), paged=False) for _ in range(num_of_rows)]
+
+    await asyncio.gather(*tasks)


+async def test(session: Session, prepared: PreparedStatement, cnt: int, num_of_rows: int):
+    for _ in range(cnt):
+        fut = to_asyncio(
+            session.execute_async(  # pyright: ignore[reportUnknownMemberType]
+                query=prepared,
+                parameters=None,
+                paging_state=None,
+            )
+        )
+        result = await fut
+        assert len(result) == num_of_rows  # pyright: ignore[reportUnknownArgumentType, reportUnknownMemberType]
+


+async def test(session: Session, prepared: PreparedStatement, cnt: int, num_of_rows: int):
+    for _ in range(cnt):
+        fut = to_asyncio(
+            session.execute_async(  # pyright: ignore[reportUnknownMemberType]
+                query=prepared,
+                parameters=None,
+                paging_state=None,
+            )
+        )
+        result = await fut
+        assert len(result) == num_of_rows  # pyright: ignore[reportUnknownArgumentType, reportUnknownMemberType]
+


To accurately compare old Python driver to Python-Rs driver we will interact with the old driver exclusively in an asynchronous manner using asyncio. For that we need to translate `ResponseFuture` callbacks to asyncio future. This translation approach is sensible and provides at least somewhat usable interface for interaction with Python driver with minimal overhead required to make sure scheduled `RequestFuture` runs to completion.

footnote: To be consistent with asyncio usage (and actual asynchronous execution instead of awaitng immidiately in "concurrent" test scenarios) we are intentionally not using the blocking `.result()` but the asyncio compatible awaitable protocol with `to_asyncio()`.

wprzytula

IIRC, you want to merge this for more convenient work with benchmarks, until the benchmarker is improved.

adespawn mentioned this pull request Mar 19, 2026

Perform some benchmarks against old driver #17

Open

patrycja-ziemkiewicz force-pushed the python-driver-benchmarks branch 2 times, most recently from 01c17af to 09080d8 Compare March 19, 2026 14:56

adespawn mentioned this pull request Mar 19, 2026

Move throughput benchmarks from nodejs-rs driver repository scylladb/scylla-rust-driver#1589

Closed

4 tasks

patrycja-ziemkiewicz force-pushed the python-driver-benchmarks branch 5 times, most recently from 1415afb to c254469 Compare March 24, 2026 12:25

patrycja-ziemkiewicz force-pushed the python-driver-benchmarks branch 2 times, most recently from b326ab3 to 62e4c14 Compare April 30, 2026 13:38

patrycja-ziemkiewicz marked this pull request as ready for review April 30, 2026 14:15

patrycja-ziemkiewicz requested review from Lorak-mmk, Copilot, jakub-graczyk, pasinskik, paulinaczajkowska and wprzytula April 30, 2026 14:15

Copilot started reviewing on behalf of patrycja-ziemkiewicz April 30, 2026 14:16 View session

Copilot AI reviewed Apr 30, 2026

View reviewed changes

patrycja-ziemkiewicz force-pushed the python-driver-benchmarks branch from 62e4c14 to dfbdafa Compare April 30, 2026 14:36

Lorak-mmk requested changes May 4, 2026

View reviewed changes

Comment thread benchmark/common.py Outdated

Comment thread benchmark/common.py

Comment thread benchmark/common.py Outdated

Comment thread benchmark/common.py Outdated

Comment thread benchmark/concurrent_insert.py Outdated

Comment thread benchmark/concurrent_ser.py Outdated

Copilot AI review requested due to automatic review settings May 7, 2026 10:54

jakub-graczyk approved these changes May 7, 2026

View reviewed changes

Copilot AI review requested due to automatic review settings May 16, 2026 22:03

Copilot started reviewing on behalf of jakub-graczyk May 16, 2026 22:04 View session

jakub-graczyk reviewed May 16, 2026

View reviewed changes

Copilot AI reviewed May 16, 2026

View reviewed changes

Copilot AI review requested due to automatic review settings July 9, 2026 14:40

patrycja-ziemkiewicz force-pushed the python-driver-benchmarks branch 2 times, most recently from 6231818 to 0b157a1 Compare July 9, 2026 14:47

Copilot started reviewing on behalf of patrycja-ziemkiewicz July 9, 2026 14:51 View session

Copilot AI reviewed Jul 9, 2026

View reviewed changes

patrycja-ziemkiewicz and others added 19 commits July 9, 2026 18:09

Add shared helpers

cb5f20c

Add insert scenarios

e96a52d

Add select scenarios

138dbf5

Add paging scenarios

a6d3622

Add batch scenarios

0bd8c06

Add config.yml

46638bc

Add README.md

07e4291

Add benchmarker output to gitignore

f90b72e

add: python driver dep

a378b41

move: helpers out of common.py

54b176c

add: insert scenarios

cc90c10

add: concurrent insert scenarios

7880fe6

add: concurrent select scenarios

ff9bd8b

add: paging scenarios

95ef5ae

add: batch scenario

4b8f8bd

update: config

2b2bf8e

update: readme

0b157a1

wprzytula approved these changes Jul 28, 2026

View reviewed changes

Navigation Menu

Uh oh!

Python driver benchmarks - #55

Python driver benchmarks#55
patrycja-ziemkiewicz wants to merge 19 commits into
scylladb:mainfrom
patrycja-ziemkiewicz:python-driver-benchmarks

patrycja-ziemkiewicz commented Mar 17, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

jakub-graczyk May 16, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI left a comment

Uh oh!

wprzytula left a comment •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants



		async def test(session: Session, cnt: int, num_of_rows: int):
		prepared = await session.prepare(SIMPLE_SELECT_QUERY)

		tasks = [session.execute(prepared, get_data(), paged=False) for _ in range(num_of_rows)]

		await asyncio.gather(*tasks)

Uh oh!

Conversation

patrycja-ziemkiewicz commented Mar 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

jakub-graczyk May 16, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

wprzytula left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

patrycja-ziemkiewicz commented Mar 17, 2026 •

edited

Loading

wprzytula left a comment •

edited

Loading