Merge pull request #4 from Informatievlaanderen/feat/graphdb-performance

Feat/graphdb performance

README.md

@@ -1,10 +1,12 @@
 # Jmeter performance testing
 
 - [LDES Server](./server)
-  - Ensures that the server can handle significant load and its performance doesn't unknowingly degrade.
+    - Ensures that the server can handle significant load and its performance doesn't unknowingly degrade.
 - [LDIO](./ldio)
-  - Not yet implemented
+    - Not yet implemented
 - [LDES Server + LDIO](./ldio-with-server)
-  - Not yet implemented
+    - Not yet implemented
 - [Binary Representation (LDIO + LDES Server)](./binary-representations)
-  - Verifying whether binary data is faster parsed than string data
+    - Verifying whether binary data is faster parsed than string data
+- [Performance of the Repository Materialiser](./graphdb)
+    - Verifying the performance improvement of using batch-size for storing members in a triples store

graphdb/.env

@@ -0,0 +1,7 @@
+# Windows
+DISPLAY=host.docker.internal:0.0
+X11=./tests/x11
+
+# Unix
+# DISPLAY=:0
+# X11=/tmp/.X11-unix

graphdb/Dockerfile

@@ -0,0 +1,11 @@
+FROM ubuntu:latest
+RUN  useradd -ms /bin/bash jmeter
+RUN apt-get update && apt-get install -y curl && apt-get clean
+RUN curl https://dlcdn.apache.org//jmeter/binaries/apache-jmeter-5.6.2.tgz --output apache-jmeter-5.6.2.tgz && tar -xf apache-jmeter-5.6.2.tgz && rm apache-jmeter-5.6.2.tgz && chown jmeter -R /apache-jmeter-5.6.2 && ln -s apache-jmeter-5.6.2 apache-jmeter
+RUN apt-get install -y openjdk-19-jre && apt-get clean
+RUN curl https://dlcdn.apache.org/jena/binaries/apache-jena-4.10.0.tar.gz --output  apache-jena-4.10.0.tar.gz && tar -xf apache-jena-4.10.0.tar.gz && rm apache-jena-4.10.0.tar.gz && chown jmeter -R apache-jena-4.10.0 && ln -s apache-jena-4.10.0 apache-jena
+RUN rm apache-jmeter/lib/mongo-java-driver-2*
+RUN curl https://repo1.maven.org/maven2/org/mongodb/mongo-java-driver/3.12.14/mongo-java-driver-3.12.14.jar --output apache-jmeter/lib/mongo-java-driver-3.12.14.jar
+USER jmeter
+WORKDIR /home/jmeter
+ENTRYPOINT ["/apache-jmeter/bin/jmeter"]

graphdb/README.md

@@ -0,0 +1,97 @@
+# LDIO Repository Materialiser performance test
+
+## Test Description
+
+This test will test the performance of processing models to the RDF4J repository materialiser
+
+The test will send 10,000 members to ldi-orchestrator, the orchestrator will process and send the members to the RDF4J
+repository
+
+## Test Setup
+
+> **Note**: if needed, copy the [environment file (.env)](./.env) to a personal file (e.g. `user.env`) and change the
+> settings as needed. If you do, you need to add ` --env-file user.env` to each `docker compose` command.
+> **_NOTE for Windows users:_**
+>
+Use [this tutorial](https://medium.com/@potatowagon/how-to-use-gui-apps-in-linux-docker-container-from-windows-host-485d3e1c64a3)
+> to run JMeter in the docker-compose file.
+> You can use DISPLAY=host.docker.internal:0.0 as environment variable to skip the ipconfig step.
+
+# Steps:
+
+Prerequisites:
+
+- Pull the latest version of the ldi orchestrator
+
+    ```bash
+    docker pull ghcr.io/informatievlaanderen/ldi-orchestrator:latest
+    ```
+
+1. Prepare directory that will contain the results
+    ```bash
+    mkdir tests/results
+    chmod 777 tests/results
+    rm -rf tests/results/output
+    rm -rf tests/results/report.jtl
+    ```
+
+2. Start test
+   POST 10k members to LDIO to materialise to the RDF4J repository
+    ```bash
+    export BATCH_SIZE=500
+    docker compose up 
+    ```
+
+3. End test
+    ```bash
+    docker compose down
+    ```
+
+4. Test results
+
+To run the different tests, repeat step 1 -> 3 above and insert the results below.
+
+- Test run:       February 23, 2023
+- Docker image
+-
+
+[//]: # (TODO add correct image tag)
+
+    - ldi-orchestrator: ghcr.io/informatievlaanderen/ldi-orchestrator: 
+
+- Resources for both the orchestrator as server:
+    ```yaml
+        deploy:
+          resources:
+            limits:
+              cpus: '4'
+              memory: '2GB'
+            reservations:
+              cpus: '2'
+              memory: '1GB'
+    ```
+
+# Results
+
+- `ldes/ldi-orchestrator:2.0.0-SNAPSHOT` \
+  Following error was counted 230 times, or for sample it has received:
+
+```text
+Non HTTP response code: java.net.SocketTimeoutException/Non HTTP response message: Read timed out
+```
+
+- `ghcr.io/informatievlaanderen/ldi-orchestrator:20240301142108`
+
+Last run on: March 4th 2024
+
+10,000 members were sent to the ldio workbench in 10 different threads, which all have a life span of 45 seconds, in the
+table below, you can see how members were successfully posted to the workbench for some batch sizes and how many members
+were processed per second
+
+| Batch Size | Members successfully posted | Members processed per second |   
+|------------|-----------------------------|------------------------------|
+| 50         | 216                         | 4.69                         |  
+| 100        | 361                         | 7.75                         |  
+| 250        | 561                         | 10.71                        |  
+| 500        | 788                         | 17.58                        |  
+| 1000       | 1210                        | 26.78                        |

graphdb/config/ldio-workbench.config.yml

@@ -0,0 +1,24 @@
+server:
+  port: 8080
+orchestrator:
+  pipelines:
+    - name: gipod
+      description: "Simple http in, version creation, http out pipeline allowing to pause output."
+      input:
+        name: Ldio:HttpIn
+        adapter:
+          name: Ldio:RdfAdapter
+      outputs:
+        - name: Ldio:RepositoryMaterialiser
+          config:
+            sparql-host: http://graphdb:8080/rdf4j-server
+            repository-id: test
+            batch-size: 50
+management:
+  tracing:
+    sampling:
+      probability: 1.0
+  endpoints:
+    web:
+      exposure:
+        include: "*"

graphdb/docker-compose.yml

@@ -0,0 +1,58 @@
+version: '3.5'
+services:
+  jmeter:
+    container_name: graphdb_jmeter
+    image: jmeter:latest
+    build: .
+    environment:
+      - DISPLAY=${DISPLAY}
+    volumes:
+      - ${X11}:/tmp/.X11-unix
+      - ./tests:/home/jmeter
+    command: -n -t graphdb-performance-testing.jmx -l results/report.jtl -e -o results/output/
+    depends_on:
+      repository-initializer:
+        condition: service_completed_successfully
+      ldio-workbench:
+        condition: service_healthy
+
+  graphdb:
+    container_name: rdf4j_server
+    image: eclipse/rdf4j-workbench:4.3.4
+    ports:
+      - "8080:8080"
+
+  repository-initializer:
+    image: ubuntu
+    command: "/bin/sh /initializer/initialize.sh"
+    container_name: graphdb_repo-initializer
+    volumes:
+      - ./repository-initializer:/initializer:ro
+    depends_on:
+      graphdb:
+        condition: service_started
+
+  ldio-workbench:
+    image: ghcr.io/informatievlaanderen/ldi-orchestrator:20240301142108
+    container_name: graphdb_ldio-workbench
+    environment:
+      - SPRING_CONFIG_NAME=application
+      - SPRING_CONFIG_LOCATION=/config/
+    volumes:
+      - ./config/ldio-workbench.config.yml:/config/application.yml:ro
+    healthcheck:
+      test: ["CMD", "wget", "-qO-", "http://localhost:8080/actuator/health"]
+      interval: 10s
+      timeout: 2s
+      retries: 10
+    ports:
+      - "8082:8080"
+    deploy:
+      resources:
+        limits:
+          cpus: '4'
+          memory: '2GB'
+        reservations:
+          cpus: '2'
+          memory: '1GB'
+

graphdb/repository-initializer/initialize.sh

@@ -0,0 +1,3 @@
+#!/bin/sh
+apt-get update && apt-get install curl -y && apt-get clean
+curl -X PUT http://graphdb:8080/rdf4j-server/repositories/test -H "Content-Type: text/turtle" -d "@/initializer/repo-definition.ttl"

graphdb/repository-initializer/repo-definition.ttl

@@ -0,0 +1,12 @@
+@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>.
+@prefix config: <tag:rdf4j.org,2023:config/>.
+
+[] a config:Repository ;
+   config:rep.id "test" ;
+   config:rep.impl [
+      config:rep.type "openrdf:SailRepository" ;
+      config:sail.impl [
+         config:sail.type "openrdf:MemoryStore" ;
+         config:mem.persist true ;
+      ]
+   ].