E2E Networks

https://docs.e2enetworks.com/docs/
Admin
E2E Networks is India's leading hyperscaler providing advanced cloud GPU infrastructure including...
Tokens:1,779
Snippets:10
Trust Score:9.9
Update:1 year ago
Show doc for...
Tokens:
Raw
### Provision GPU Instance and Model Endpoint with Terraform

Source: https://context7.com/context7/e2enetworks/llms.txt

This Terraform configuration demonstrates how to manage E2E Networks infrastructure as code. It shows the configuration of the E2E Networks provider, provisioning a GPU instance for ML training, and creating a model endpoint for inference with autoscaling capabilities. Outputs provide connection details for the provisioned instance and endpoint.

```hcl
terraform {
  required_providers {
    e2e = {
      source  = "e2enetworks/e2e"
      version = ">~ 1.0"
    }
  }
}

provider "e2e" {
  api_key = var.e2e_api_key
  region  = "india-mumbai"
}

resource "e2e_instance" "gpu_node" {
  name          = "ml-training-node"
  instance_type = "h100-80gb"
  image         = "ubuntu-22.04-cuda-12"

  disk_size_gb = 500

  tags = {
    environment = "production"
    project     = "llm-training"
  }
}

resource "e2e_model_endpoint" "inference_api" {
  name         = "production-inference"
  model_path   = "s3://models/my-model-v1"
  gpu_type     = "a100-40gb"
  replicas     = 3

  autoscaling {
    min_replicas = 1
    max_replicas = 10
    target_gpu_utilization = 75
  }
}

output "instance_ip" {
  value = e2e_instance.gpu_node.public_ip
}

output "endpoint_url" {
  value = e2e_model_endpoint.inference_api.endpoint_url
}
```

--------------------------------

### Python Use Case: AI Research and Experimentation

Source: https://context7.com/context7/e2enetworks/llms.txt

This Python snippet details configuration for AI research and experimentation environments on E2E Networks H100 or H200 instances. It specifies the environment, GPU type, storage size, and pre-installed frameworks.

```python
# Use Case 4: AI Research and Experimentation
"""
Scenario: Rapid prototyping and model experiments
Recommended: NVIDIA H100 or H200 instances
Configuration: Jupyter notebook environment
"""
research_config = {
    "environment": "jupyter_lab",
    "gpu_type": "h100-80gb",
    "storage_gb": 1000,
    "preinstalled_frameworks": [
        "pytorch",
        "tensorflow",
        "jax",
        "transformers"
    ]
}
```

--------------------------------

### Configure E2E Networks Environment Variables

Source: https://context7.com/context7/e2enetworks/llms.txt

This snippet shows how to set environment variables for E2E Networks API key, region, and project ID. These variables are typically used by SDKs or CLIs to authenticate and configure access to the platform.

```bash
export E2E_API_KEY="your_api_key_here"
export E2E_REGION="india-mumbai"
export E2E_PROJECT_ID="your_project_id"
```

--------------------------------

### Python Use Case: Batch Processing and Data Analytics

Source: https://context7.com/context7/e2enetworks/llms.txt

This Python snippet defines configuration for batch processing and data analytics using E2E Networks A100 instances. It includes settings for spot instances, auto-shutdown, and checkpoint intervals for cost optimization and efficient job management.

```python
# Use Case 3: Batch Processing and Data Analytics
"""
Scenario: Processing large datasets with GPU acceleration
Recommended: NVIDIA A100 instances
Configuration: On-demand instances for batch jobs
"""
batch_config = {
    "job_type": "batch_processing",
    "instance_type": "a100-40gb",
    "spot_instances": True,  # Cost optimization
    "auto_shutdown": True,   # Shutdown after job completion
    "checkpoint_interval": "1h"
}
```

--------------------------------

### Deploy AI Model Endpoint with Inference API Call

Source: https://context7.com/context7/e2enetworks/llms.txt

This snippet demonstrates how to deploy and manage AI model inference endpoints, including versioning. It shows a typical `curl` command for making inference requests to a deployed model endpoint, specifying the endpoint name, version, and input data. The expected response includes status, predictions, and latency.

```bash
curl -X POST https://api.e2enetworks.com/v1/inference \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "endpoint": "model-endpoint-name",
    "version": "v1.2.3",
    "inputs": {
      "text": "Input data for model inference"
    }
  }'
```

--------------------------------

### Use API Token for Subsequent Requests

Source: https://context7.com/context7/e2enetworks/llms.txt

This snippet demonstrates how to use the obtained API access token to make subsequent authenticated requests to the E2E Networks API. It shows a cURL request to list instances, including the 'Authorization: Bearer' header.

```bash
curl https://api.e2enetworks.com/v1/instances \
  -H "Authorization: Bearer eyJhbGc..." \
  -H "Content-Type: application/json"
```

--------------------------------

### Python Use Case: Large Language Model Training

Source: https://context7.com/context7/e2enetworks/llms.txt

This Python snippet outlines configuration for training a large language model using E2E Networks' H100 instances. It specifies instance type, number of nodes, GPUs per node, distributed strategy, and mixed precision.

```python
# Use Case 1: Large Language Model Training
"""
Scenario: Training a 13B parameter language model
Recommended: NVIDIA H100 instances
Configuration: Multi-node distributed training
"""
import torch
import transformers

# Configuration for E2E Networks H100 cluster
training_config = {
    "instance_type": "h100-80gb",
    "num_nodes": 8,
    "gpus_per_node": 8,
    "total_gpus": 64,
    "distributed_strategy": "FSDP",
    "mixed_precision": "bf16"
}
```

--------------------------------

### Python Use Case: Real-time Model Inference API

Source: https://context7.com/context7/e2enetworks/llms.txt

This Python snippet configures settings for a real-time model inference API with autoscaling on E2E Networks A100 instances. It includes endpoint type, GPU type, replica counts, autoscaling metrics, and batch size.

```python
# Use Case 2: Real-time Model Inference API
"""
Scenario: Serving a deployed model with autoscaling
Recommended: NVIDIA A100 instances
Configuration: Load-balanced endpoint with 3-10 replicas
"""
inference_config = {
    "endpoint_type": "tir_model_endpoint",
    "gpu_type": "a100-40gb",
    "min_replicas": 3,
    "max_replicas": 10,
    "autoscale_metric": "gpu_utilization",
    "target_utilization": 75,
    "max_batch_size": 32
}
```

--------------------------------

### E2E Networks Service Commitments Configuration

Source: https://context7.com/context7/e2enetworks/llms.txt

This snippet defines the service level commitments for E2E Networks in YAML format. It includes details on uptime guarantee, monthly downtime allowance, support availability, response times, and reliability features.

```yaml
# E2E Networks Service Commitments:

uptime_sla:
  guarantee: 99.99%
  monthly_downtime_allowance: 4.38 minutes
  compensation: Credit based on service level breach

support:
  availability: 24/7 human support
  response_time:
    critical: < 1 hour
    high: < 4 hours
    normal: < 24 hours
  channels:
    - email
    - phone
    - chat
    - ticket_system

sales_response:
  initial_contact: 24-48 hours
  custom_solutions: Available
  enterprise_agreements: Available

reliability_features:
  - Robust engineering infrastructure
  - Automated failover
  - Data redundancy
  - Network resilience
  - GPU health monitoring
```

--------------------------------

### Authenticate with E2E Networks API

Source: https://context7.com/context7/e2enetworks/llms.txt

This snippet shows how to authenticate with the E2E Networks API using cURL to obtain an access token. It requires a POST request with JSON payload containing email and password. The response contains access and refresh tokens.

```bash
curl https://api.e2enetworks.com/v1/auth/token \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{
    "email": "user@example.com",
    "password": "your_password"
  }'
```
Tokens:
Raw