Setting up the benchmark for S3

To set up the benchmark for the S3 resource, install and configure the GOSBench tool.

Prerequisites

• The load generator VMs are deployed, as described in Deploying virtual machines with load generators.
• The target storage is created, as described in Creating the target storage.

To install and configure GOSBench

1. Download and extract the tool from GitHub on each load generator and the coordinator node:

   # curl -OL https://github.com/mulbc/gosbench/releases/download/v0.4/gosbench_0.4_Linux_x86_64.tar.gz
   # tar -xzf gosbench_0.4_Linux_x86_64.tar.gz

2. On the coordinator node, create the file gosbench_script.yaml with the following configuration:

   s3_config:
     - access_key: <ACCESS_KEY>
       secret_key: <SECRET_KEY>
       region: eu-central-1
       endpoint: <S3_ENDPOINT_URL>
       skipSSLverify: true
   
   grafana_config:
     endpoint: https://<GRAFANA_ENDPOINT>
     username: admin
     password: <ADMIN_PASSWORD>
   
   tests:
     - name: write test
       read_weight: 0
       existing_read_weight: 0
       write_weight: 100
       delete_weight: 0
       list_weight: 0
       objects:
         size_min: 1
         size_max: 100000
         part_size: 0
         size_distribution: random
         unit: KB
         number_min: 10
         number_max: 10
         number_distribution: constant
       buckets:
         number_min: 1
         number_max: 10
         number_distribution: constant
       bucket_prefix: 1255gosbenchobject_
       prefix: obj
       stop_with_runtime: 900s
       stop_with_ops:
       workers: <NUMBER_OF_WORKERS>
       workers_share_buckets: True
       parallel_clients: 3
       clean_after: True
   
     - name: read test
       read_weight: 100
       existing_read_weight: 0
       write_weight: 0
       delete_weight: 0
       list_weight: 0
       objects:
         size_min: 1
         size_max: 100000
         part_size: 0
         size_distribution: random
         unit: KB
         number_min: 10
         number_max: 10
         number_distribution: constant
       buckets:
         number_min: 1
         number_max: 10
         number_distribution: constant
       bucket_prefix: 1255gosbenchobject_
       prefix: obj
       stop_with_runtime: 900s
       stop_with_ops:
       workers: <NUMBER_OF_WORKERS>
       workers_share_buckets: True
       parallel_clients: 3
       clean_after: True

3. Update the required fields accordingly, in particular, these fields:

   • access_key is the S3 access key to access the resource.
   • secret_key is the S3 secret key to access the resource,
   • endpoint in the s3_config section is the S3 endpoint URL to access the resource.
   • endpoint in the grafana_config section is the Grafana dashboard URL, which is https://<admin_panel_IP>:8888/grafana/d/dashboard-directory-en-US/.
   • password is the Grafana administrator password.
   • workers is the number of load generators. Note that there are multiple occurrences of this parameter that need to be set to the same value.

For more details, refer to the GOSBench documentation.