Machine Learning Platform#
The Machine Learning Platform component is implemented with H2O.ai. They are the open source leader in AI and machine learning with a mission to democratize AI for everyone. H2O is a fully open source, distributed in-memory machine learning platform with linear scalability. H2O supports the most widely used statistical & machine learning algorithms including gradient boosted machines, generalized linear models, deep learning and more. H2O also has an industry leading AutoML functionality that automatically runs through all the algorithms and their hyperparameters to produce a leaderboard of the best models. The H2O platform is used by over 18,000 organizations globally and is extremely popular in both the R & Python communities.
Please refer to H2O.ai's website for more information.
Hardware Configuration#
The available hardware configuration options are listed below.
Standard Compute Machines#
The currently available compute machines are virtual machines of the m5 type (GHz Intel Xeon® Platinum 8175 processors with new Intel Advanced Vector Extension instruction set). The following list is supported:
| Virtual Machine Size | vCPU | Memory (GiB) | VM Storage (GiB) | Network Bandwidth (Gbps) | Block Storage Bandwidth (Mbps) |
|---|---|---|---|---|---|
| m5.large | 2 | 8 | Block storage-Only | Up to 10 | Up to 3,500 |
| m5.xlarge | 4 | 16 | Block storage-Only | Up to 10 | Up to 3,500 |
| m5.2xlarge | 8 | 32 | Block storage-Only | Up to 10 | Up to 3,500 |
| m5.4xlarge | 16 | 64 | Block storage-Only | Up to 10 | 3,500 |
| m5.8xlarge | 32 | 128 | Block storage-Only | 10 | 5,000 |
| m5.12xlarge | 48 | 192 | Block storage-Only | 10 | 7,000 |
| m5.16xlarge | 64 | 256 | Block storage-Only | 20 | 10,000 |
| m5.24xlarge | 96 | 384 | Block storage-Only | 25 | 14,000 |
GPU Compute Machines (Deep Learning)#
The currently available compute machines are AWS EC2 instances of the P3 type (NVIDIA Tesla V100 GPU and High frequency Intel Xeon E5-2686 v4 CPU). The following list is supported:
| Virtual Machine | GPUs | vCPU | Mem (GiB) | GPU Mem (GiB) | GPU P2P | Storage (GB) | Dedicated EBS Bandwidth | Networking Performance |
|---|---|---|---|---|---|---|---|---|
| p3.2xlarge | 1 | 8 | 61 | 16 | - | Block storage-Only | 1.5 Gbps | Up to 10 Gigabit |
| p3.8xlarge | 4 | 32 | 244 | 64 | NVLink | Block storage-Only | 7 Gbps | 10 Gigabit |
| p3.16xlarge | 8 | 64 | 488 | 128 | NVLink | Block storageBS-Only | 14 Gbps | 25 Gigabit |
Block Storage#
The local storage of a selected compute machine (Standard Compute Machine or a GPU Compute Machine) will be block storage:
| Volume Type | Volume Size | Max Throughput/Volume |
|---|---|---|
| Block storage | 1 GB - 16 TB | 250 MB/s |