In our test setup, we used two EC2 instances taking the roles of a client and an UltiHash cluster respectively. Each instance was of type m7g.2xlarge and had an gp3 EBS volume attached. Both instances were situated within the same VPC subnet in the eu-central-1 region --the same region hosting the S3 bucket used for throughput comparison.
Format | Description | Size | Space savings | PUT throughput (S3) | PUT throughput (UH) | PUT time (S3) | PUT time (UH) | GET throughput (S3) | GET throughput (UH) | GET time (S3) | GET time (UH) | Link to dataset |
---|---|---|---|---|---|---|---|---|---|---|---|---|
RAW | Human and animal scans | 1.48 GB | 74% | 99.20 MB/s | ▼ 32.03 MB/s | 45.6 s | ▼ 141.3 s | 60.76 MB/s | ▼ 52.11 MB/s | 74.5 s | ▼ 86.8 s | https://www.kaggle.com/datasets/imaginar2t/cbctdata |
TIFF | Images of climate data | 16 GB | 46% | 84.44 MB/s | ▼ 37.26 MB/s | 183.8 s | ▼ 416.6 s | 61.33 MB/s | ▲ 110.21 MB/s | 253.1 s | ▲ 140.9 s | https://www.kaggle.com/datasets/abireltaief/highresolution-geotiff-images-of-climatic-data |
CSV | Disease prediction symptoms | 1.4 MB | 42% | 3.00 MB/s | ▼ 0.15 MB/s | 0.4 s | ▼ 8.9 s | 4.19 MB/s | ▲ 55.20 MB/s | 0.3 s | ▲ 0.02 s | https://www.kaggle.com/datasets/kaushil268/disease-prediction-using-machine-learning |
PNG | Car license plate images | 213.6 MB | 29% | 4.31 MB/s | ▼ 0.46 MB/s | 47.0 s | ▼ 441.3 s | 6.07 MB/s | ▲ 56.71 MB/s | 33.4 s | ▲ 3.6 s | https://www.kaggle.com/datasets/andrewmvd/car-plate-detection |
XML | Car license plate annotations | 1.8 MB | 15% | 0.01 MB/s | ▼ 0.0006 MB/s | 17.53 s | ▼ 436.6 s | 0.02 MB/s | ▲ 0.314 MB/s | 11.1 s | ▲ 0.8 s | https://www.kaggle.com/datasets/andrewmvd/car-plate-detection |