https://github.com/conker84/streams-benchmark
https://github.com/conker84/streams-benchmark
Last synced: 2 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/conker84/streams-benchmark
- Owner: conker84
- Created: 2020-10-23T10:32:05.000Z (over 4 years ago)
- Default Branch: master
- Last Pushed: 2020-10-30T08:29:41.000Z (over 4 years ago)
- Last Synced: 2025-02-10T15:50:19.172Z (4 months ago)
- Size: 59.8 MB
- Stars: 0
- Watchers: 4
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# How to do a benchmark
Starting from [this document](https://docs.google.com/document/u/2/d/1HpqHSRzelAN1QT0JSJK7ywDOU_ZRgHlr5h_AIxfOgt0/edit#) we're trying to create a reproducible env in order to compare and improve the performances of the Neo4j Streams Plugin.
## Docker
### Prerequisites
Go into [Neo4j](http://localhost:8080) and stop the Sink via the follwing query:
```cypher
CALL streams.sink.stop()
```
Define the constrains:
```cypher
CREATE CONSTRAINT ON (tx:TxStr) ASSERT tx.txId IS UNIQUE;
CREATE CONSTRAINT ON (c:CustomerStr) ASSERT c.custId IS UNIQUE;
```
Check if the topic `benchmark` exists:
```bash
docker run --network streams-benchmark_default \
--volume $PWD/kafka-data:/data \
confluentinc/cp-kafkacat \
kafkacat -b broker:9093 -L | grep benchmark -A 2
```
Create a Kafka topic `benchmark` (if necessary):
```bash
docker exec -it broker kafka-topics --create \
--zookeeper zookeeper:2181 \
--replication-factor 1 --partitions 1 \
--topic benchmark
```
Publish `1M messages` into the `benchmark` topic:
```bash
docker run --network streams-benchmark_default \
--volume $PWD/kafka-data:/data \
confluentinc/cp-kafkacat \
kafkacat -b broker:9093 \
-t benchmark \
-P -l /data/data.1M.json
```
Wait until it ends... (it takes a while)
Count the messages into the topic just to be sure:
```bash
docker run --network streams-benchmark_default \
--volume $PWD/kafka-data:/data \
confluentinc/cp-kafkacat \
kafkacat -b broker:9093 \
-t benchmark \
-C -e -q | wc -l
```
Now choose what you want to test.
### Test with the Neo4j Streams Plugin Sink
You need to (de)comment the following line inside the compose file in order to manage the auto commit:
`NEO4J_kafka_enable_auto_commit: "false"`
In the beta release we added a new feature that, in case of `kafka.enable.auto.commit=false`, allows committing the offsets in async way, you can simply manage it by (de)commenting the following line:
`NEO4J_kafka_streams_async_commit: "true"`
Start the Sink
```cypher
CALL streams.sink.start()
```
### Test with the Neo4j Kafka Connect Sink
You can choose between two options:
#### Install the Sink with parallelization (default behaviour)
```bash
curl -X POST localhost:8083/connectors \
-H 'Content-Type:application/json' \
-H 'Accept:application/json' \
-d @connect.neo4j.json
```
#### Install the Non Parallelized Sink
```bash
curl -X POST localhost:8083/connectors \
-H 'Content-Type:application/json' \
-H 'Accept:application/json' \
-d @connect.neo4.non-parallelized.json
```
### Test with a simple kafka consumer
Run:
`java -jar simple-neo4j-consumer-1.0.jar localhost:9092 benchmark bolt://localhost:7687 neo4j benchmark`
N.b:
`java -jar simple-neo4j-consumer-1.0.jar []`
`auto_commit` values:
* `true` (default)
* `false` disable auto commit and uses the `Consumer#commitSync` as the Neo4j Streams plugin
### How to compute the result
**N.b.** At the end of the test (indipendentely the one you choose) you **must** have:
- 1245480 nodes
- 1000000 relationships
Wait the ingestion ends and then run the following cypher query:
```cypher
MATCH (n:TxStr) WITH count(n) AS countTx
MATCH (n:TxStr) WITH countTx, max(n.finalTimestamp) AS maxTime
MATCH (n:TxStr) WITH countTx, maxTime, min(n.finalTimestamp) AS minTime
RETURN (countTx * 1.000) / ((maxTime - minTime) / 1000)
```