Skip to content
This repository was archived by the owner on Jul 11, 2025. It is now read-only.

Commit f100cdf

Browse files
cybermaggedoncybermaggedon
committed
Deployment updates
1 parent dd1f584 commit f100cdf

File tree

4 files changed

+326
-3
lines changed

4 files changed

+326
-3
lines changed

‎_config.yml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -59,7 +59,7 @@ heading_anchors: true
5959
# Note: The footer_content option is deprecated and will be removed in a
6060
# future major release. Please use `_includes/footer_custom.html` for more
6161
# robust markup / liquid-based content.
62-
footer_content: "Copyright &copy; 2017-2020 Patrick Marsceill. Distributed by an <a href=\"https://github.com/just-the-docs/just-the-docs/tree/main/LICENSE.txt\">MIT license.</a>"
62+
footer_content: "Copyright &copy; 2024-2025 TrustGraph.ai. Distributed under an <a href=\"https://github.com/trustgraph-ai/trustgraph/blob/master/LICENSE\">Apache 2 license.</a>"
6363

6464
# Footer last edited timestamp
6565
last_edit_timestamp: true # show or hide edit time - page must have `last_modified_date` defined in the frontmatter

‎deployment/azure.md

Lines changed: 153 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,153 @@
1+
---
2+
title: Azure AKS
3+
layout: default
4+
nav_order: 3.5
5+
parent: Deployment
6+
grand_parent: TrustGraph Documentation
7+
---
8+
9+
# Microsoft Azure AKS Deployment
10+
11+
Deploy TrustGraph on Microsoft Azure using Azure Kubernetes Service (AKS) with comprehensive AI integration.
12+
13+
## Overview
14+
15+
TrustGraph provides a complete Azure deployment solution using **Pulumi** (Infrastructure as Code) that automatically provisions and configures an AKS cluster with Azure's AI services for a production-ready TrustGraph deployment.
16+
17+
## What You Get
18+
19+
The Azure deployment includes:
20+
21+
- **Dedicated resource group** for complete resource isolation
22+
- **Azure Identity service principal** for secure component authentication
23+
- **AKS cluster** deployed in managed resource group
24+
- **Key Vault and Storage Account** for AI component requirements
25+
- **Azure AI Foundry integration** with AI hub, workspace, and serverless endpoints
26+
- **Azure Cognitive Services** with OpenAI GPT-4o-mini deployment
27+
- **Dual AI model support**: Phi-4 and OpenAI models
28+
- **Complete TrustGraph stack** deployed and configured
29+
- **Secrets management** for secure credential handling
30+
- **Monitoring and observability** with Grafana
31+
- **Web workbench** for document processing and Graph RAG
32+
33+
## Deployment Method
34+
35+
The deployment uses **Pulumi**, an Infrastructure as Code tool that:
36+
37+
- Has an open-source license
38+
- Uses general-purpose programming languages (TypeScript/JavaScript)
39+
- Provides testable infrastructure code
40+
- Offers retryable deployments
41+
- Supports local or cloud state management
42+
43+
## Architecture
44+
45+
**Kubernetes Platform**: Azure Kubernetes Service (AKS)
46+
**AI Services**: Azure AI Foundry + Cognitive Services
47+
**Default Models**: Phi-4 (Machine Learning) and GPT-4o-mini (OpenAI)
48+
**Identity Management**: Azure Identity service principal
49+
**Storage**: Azure Key Vault and Storage Account
50+
**Network**: Managed AKS networking with Azure CNI
51+
**Monitoring**: Integrated Azure monitoring and Grafana
52+
53+
## AI Model Options
54+
55+
Choose between two AI configurations:
56+
57+
### Machine Learning Services (AI Foundry)
58+
- **Model**: Phi-4 (serverless endpoint)
59+
- **Configuration**: Copy `resources.yaml.mls` to `resources.yaml`
60+
- **Features**: Azure-native model hosting
61+
62+
### Cognitive Services (OpenAI)
63+
- **Model**: GPT-4o-mini
64+
- **Configuration**: Copy `resources.yaml.cs` to `resources.yaml`
65+
- **Features**: OpenAI API compatibility
66+
67+
## Quick Process Overview
68+
69+
1. **Choose AI model** (Phi-4 or OpenAI)
70+
2. **Install Pulumi** and dependencies
71+
3. **Configure Azure credentials** using `az login`
72+
4. **Customize configuration** in `Pulumi.azure.yaml`
73+
5. **Deploy** with `pulumi up`
74+
6. **Access services** via port-forwarding
75+
76+
## Configuration Options
77+
78+
Customizable settings include:
79+
80+
- **Location**: Azure deployment region
81+
- **Environment**: dev, staging, production
82+
- **AI Endpoint Model**: e.g., `azureml://registries/azureml/models/Phi-4`
83+
- **OpenAI Model**: e.g., `gpt-4o-mini`
84+
- **OpenAI Version**: e.g., `"2024-07-18"` (quoted for date format)
85+
- **Content Filtering**: e.g., `Microsoft.DefaultV2` (Responsible AI policy)
86+
87+
## Access Points
88+
89+
Once deployed, you'll have access to:
90+
91+
- **TrustGraph API**: Port 8088
92+
- **Web Workbench**: Port 8888 (document processing, Graph RAG)
93+
- **Grafana Monitoring**: Port 3000
94+
95+
## Azure AI Integration
96+
97+
The deployment includes comprehensive Azure AI integration:
98+
99+
### Machine Learning Services
100+
- **AI Hub**: Central workspace for ML operations
101+
- **AI Workspace**: Project-specific environment
102+
- **Serverless Endpoints**: Scalable model hosting
103+
- **Model Catalog**: Access to Azure's model library
104+
105+
### Cognitive Services
106+
- **OpenAI Service**: GPT models via Azure
107+
- **Content Filtering**: Responsible AI policies
108+
- **API Management**: Secure API access
109+
- **Usage Monitoring**: Built-in analytics
110+
111+
## Complete Documentation
112+
113+
For detailed step-by-step instructions, configuration options, and troubleshooting, visit:
114+
115+
**[TrustGraph Azure AKS Deployment Guide](https://github.com/trustgraph-ai/pulumi-trustgraph-aks)**
116+
117+
The repository contains:
118+
- Complete Pulumi deployment code
119+
- AKS cluster configuration
120+
- Azure AI services integration
121+
- Dual model configuration templates
122+
- Detailed setup instructions
123+
- Troubleshooting guides
124+
- Customization options
125+
126+
## Important Notes
127+
128+
**Storage Account Issues**: Azure occasionally reports "parallel access to resources" errors during Storage Account creation. If deployment fails, retry with `pulumi up` - it's retryable and continues from where it left off.
129+
130+
**Model Selection**: Choose your AI model before deployment by copying the appropriate `resources.yaml.*` file to `resources.yaml`.
131+
132+
**Resource Groups**: Azure automatically creates separate resource groups for AKS cluster components.
133+
134+
**SSH Keys**: An SSH private key is generated but typically not needed for AKS management.
135+
136+
## Azure Enterprise Features
137+
138+
**Enterprise Integration**: Native integration with Azure Active Directory
139+
**Compliance**: Built-in compliance tools and reporting
140+
**Security**: Azure Security Center integration
141+
**Networking**: Advanced networking with Azure CNI
142+
**Monitoring**: Azure Monitor and Application Insights
143+
**Backup**: Azure Backup integration for persistent data
144+
145+
## Next Steps
146+
147+
After deployment, you can:
148+
- Load documents through the web workbench
149+
- Test Graph RAG queries with Phi-4 or OpenAI models
150+
- Monitor processing through Grafana and Azure Monitor
151+
- Scale the AKS cluster as needed
152+
- Integrate with other Azure services
153+
- Leverage Azure's enterprise security features

‎deployment/intel.md

Lines changed: 170 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,170 @@
1+
---
2+
title: Intel GPU / Tiber Cloud
3+
layout: default
4+
nav_order: 2.7
5+
parent: Deployment
6+
grand_parent: TrustGraph Documentation
7+
---
8+
9+
# Intel Tiber Cloud Deployment
10+
11+
Deploy TrustGraph on Intel Tiber Cloud with Intel GPU and Gaudi accelerated systems for high-performance AI workloads.
12+
13+
## Overview
14+
15+
TrustGraph provides specialized deployment configurations for Intel's advanced hardware platforms, including Intel Tiber Cloud and bare-metal Intel GPU/Gaudi systems. This deployment method is designed for:
16+
17+
- **High-performance AI inference** with Intel accelerators
18+
- **Self-hosted deployments** on Intel hardware
19+
- **Research and development** with cutting-edge Intel AI technologies
20+
- **Enterprise workloads** requiring Intel-optimized performance
21+
22+
⚠️ **Work in Progress**: This deployment method is actively being developed and optimized for Intel's latest hardware platforms.
23+
24+
## What You Get
25+
26+
The Intel Tiber Cloud deployment includes:
27+
28+
- **Intel accelerated systems** (Gaudi, GPU, or GR platforms)
29+
- **Optimized AI inference** with vLLM or TGI servers
30+
- **Large language models** like Llama 3.3 70B
31+
- **Complete TrustGraph stack** deployed via containerization
32+
- **SSH-based deployment** with automated scripts
33+
- **Port forwarding setup** for secure access
34+
- **Monitoring and observability** with Grafana
35+
- **Web workbench** for document processing and Graph RAG
36+
37+
## Intel Hardware Platforms
38+
39+
### Intel Gaudi Systems
40+
- **Software**: TrustGraph 1.0.13 + vLLM (HabanaAI fork)
41+
- **Model**: Llama 3.3 70B
42+
- **Deployment**: `deploy-gaudi-vllm.zip`
43+
- **Optimized for**: AI training and inference workloads
44+
45+
### Intel GPU Multi-GPU 1550 Systems
46+
- **Software**: TrustGraph 1.0.13 + TGI Server 3.3.1-intel-xpu
47+
- **Deployment**: `deploy-gpu-tgi.zip`
48+
- **Optimized for**: High-throughput GPU inference
49+
50+
### Intel GR Systems
51+
- **Software**: TrustGraph 1.0.13 + TGI Server 3.3.1-intel-xpu
52+
- **Deployment**: `deploy-gr.zip`
53+
- **Optimized for**: Specialized Intel AI workloads
54+
55+
## Why Choose Intel Tiber Cloud?
56+
57+
### 🚀 **Cutting-Edge AI Hardware**
58+
- **Intel Gaudi**: Purpose-built for AI training and inference
59+
- **Intel GPU**: High-performance parallel processing
60+
- **Specialized Architecture**: Optimized for AI/ML workloads
61+
62+
### ⚡ **Performance Optimization**
63+
- **Hardware-Accelerated Inference**: Native Intel optimization
64+
- **Large Model Support**: Handle models like Llama 3.3 70B efficiently
65+
- **Reduced Latency**: Direct hardware acceleration
66+
67+
### 🔒 **Self-Hosted Control**
68+
- **Data Sovereignty**: Complete control over data and models
69+
- **Custom Configuration**: Tailor deployments to specific needs
70+
- **Enterprise Security**: Self-hosted infrastructure
71+
72+
### 🛠️ **Developer Access**
73+
- **Research Platform**: Access to latest Intel AI technologies
74+
- **Experimentation**: Test advanced AI configurations
75+
- **Direct Hardware Access**: Low-level optimization capabilities
76+
77+
## Deployment Method
78+
79+
The Intel deployment uses automated deployment scripts that:
80+
81+
- Connect via SSH jump host to Intel Tiber Cloud
82+
- Deploy pre-configured TrustGraph packages
83+
- Set up Intel-optimized AI inference servers
84+
- Configure port forwarding for secure access
85+
- Initialize monitoring and web interfaces
86+
87+
## Quick Process Overview
88+
89+
1. **Obtain access** to Intel Tiber Cloud instance
90+
2. **Create HuggingFace token** and accept model licenses
91+
3. **Choose deployment type** (Gaudi, GPU, or GR)
92+
4. **Deploy via script** with SSH parameters
93+
5. **Connect via SSH** with port forwarding
94+
6. **Access services** through forwarded ports
95+
96+
## Access Configuration
97+
98+
Intel Tiber Cloud uses SSH jump host access:
99+
100+
```bash
101+
# SSH connection format
102+
ssh -J guest@[jump-host] sdp@[target-host]
103+
104+
# Deployment command
105+
./deploy-tiber guest@[jump-host] sdp@[target-host] [deployment-package]
106+
107+
# Port forwarding
108+
./port-forward guest@[jump-host] sdp@[target-host]
109+
```
110+
111+
## Access Points
112+
113+
Once deployed, you'll have access to:
114+
115+
- **TrustGraph API**: Port 8089 (forwarded from 8088)
116+
- **Web Workbench**: Port 8889 (forwarded from 8888)
117+
- **Grafana Monitoring**: Port 3001 (forwarded from 3000)
118+
119+
## Model Support
120+
121+
**Large Language Models**: Llama 3.3 70B and other HuggingFace models
122+
**License Requirements**: HuggingFace account with model access
123+
**Hardware Optimization**: Intel-specific optimizations for inference
124+
**Inference Engines**: vLLM (Gaudi) and TGI (GPU/GR)
125+
126+
## Complete Documentation
127+
128+
For detailed step-by-step instructions, deployment packages, and troubleshooting, visit:
129+
130+
**[TrustGraph Intel Tiber Cloud Guide](https://github.com/trustgraph-ai/trustgraph-tiber-cloud)**
131+
132+
The repository contains:
133+
- Deployment automation scripts
134+
- Intel platform-specific configurations
135+
- Pre-built deployment packages
136+
- SSH connection utilities
137+
- Port forwarding setup
138+
- Detailed setup instructions
139+
- Intel hardware optimization guides
140+
141+
## Prerequisites
142+
143+
**Intel Tiber Cloud Access**: Account and instance allocation
144+
**HuggingFace Token**: For model downloads and licensing
145+
**SSH Access**: Familiarity with SSH jump host connections
146+
**Model Licenses**: Acceptance of required model licenses (e.g., Llama)
147+
148+
## Performance Benefits
149+
150+
**Hardware Acceleration**: Native Intel AI acceleration
151+
**Large Model Efficiency**: Optimized for 70B+ parameter models
152+
**Reduced Infrastructure Costs**: Efficient hardware utilization
153+
**Custom Optimization**: Direct access to hardware-level tuning
154+
155+
## Use Cases
156+
157+
**Research & Development**: Access to cutting-edge Intel AI hardware
158+
**High-Performance Inference**: Large model deployment with optimal performance
159+
**Self-Hosted AI**: Complete control over AI infrastructure
160+
**Intel Ecosystem Integration**: Leverage Intel's AI software stack
161+
162+
## Next Steps
163+
164+
After deployment, you can:
165+
- Load documents through the web workbench
166+
- Test Graph RAG queries with Llama 3.3 70B
167+
- Monitor Intel hardware performance through Grafana
168+
- Experiment with Intel-optimized AI configurations
169+
- Benchmark performance against other platforms
170+
- Integrate with Intel's AI development tools

‎deployment/scaleway.md

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
---
22
title: Scaleway
33
layout: default
4-
nav_order: 5
4+
nav_order: 3.2
55
parent: Deployment
66
grand_parent: TrustGraph Documentation
77
---
@@ -142,4 +142,4 @@ After deployment, you can:
142142
- Monitor processing through Grafana
143143
- Scale the cluster as needed
144144
- Integrate with other Scaleway services
145-
- Ensure GDPR compliance for your AI workflows
145+
- Ensure GDPR compliance for your AI workflows

0 commit comments

Comments
 (0)