Skip to content

Commit 30a37bb

Browse files
authored
Merge branch 'main' into jcg-ai-gen-ai-oic-email-analysis
2 parents 9bf4110 + 436ecc2 commit 30a37bb

File tree

10 files changed

+312
-49
lines changed

10 files changed

+312
-49
lines changed
Lines changed: 125 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,125 @@
1+
# Overview
2+
3+
This repository provides a step-by-step tutorial for deploying and using [Mistral 7B Instruct](https://mistral.ai/technology/#models) Large Language Model using the [vLLM](https://github.com/vllm-project/vllm?tab=readme-ov-file) library.
4+
5+
# Requirements
6+
7+
* An OCI tenancy with A10 GPU quota.
8+
* A [Huggingface](https://huggingface.co/) account with a valid Auth Token.
9+
10+
# Model Deployment
11+
12+
## Mistral models
13+
14+
[Mistral.ai](https://mistral.ai/) is a French AI startup that develop Large Language Models (LLMs). Mistral 7B is the small yet powerful open model that supports English and code. The Mistral 7B Instruct is a chat optimized version of Mistral 7B. Mixtral 8x7B is a 7B sparse Mixture-of-Experts that supports French, Italian, German and Spanish on top of English and code (stronger than Mistral 7B). It uses 12B parameters out of 45B total.
15+
16+
## vLLM Library
17+
18+
vLLM is an alternative model serving solution to NVIDIA Triton. It is easy to use as it comes as a preconfigured container.
19+
20+
## Instance Configuration
21+
22+
In this example a single A10 GPU VM shape, codename VM.GPU.A10.1, is used. The image is the NVIDIA GPU Cloud Machine image from the OCI marketplace. A boot volume of 200 GB is also recommended.
23+
24+
## Image Update
25+
26+
Since the latest NVIDIA GPU Cloud Machine image is almost 1 year old, it is recommended to update NVIDIA drivers and CUDA by running:
27+
28+
```
29+
sudo apt purge nvidia* libnvidia*
30+
sudo apt-get install -y cuda-drivers-545
31+
sudo apt-get install -y nvidia-kernel-open-545
32+
sudo apt-get install -y cuda-toolkit-12-3
33+
sudo reboot
34+
```
35+
36+
## System configuration
37+
38+
Once the NVIDIA packages are updated, it is necessary to reconfigure docker in order to make it GPU aware:
39+
40+
```
41+
sudo apt-get install -y nvidia-container-toolkit
42+
sudo nvidia-ctk runtime configure --runtime=docker
43+
sudo systemctl restart docker
44+
```
45+
46+
## Container Deployment
47+
48+
To deploy the model, simply run the vLLM container:
49+
50+
```
51+
docker run --gpus all \
52+
-e HF_TOKEN=$HF_TOKEN -p 8000:8000 \
53+
ghcr.io/mistralai/mistral-src/vllm:latest \
54+
--host 0.0.0.0 \
55+
--model mistralai/Mistral-7B-Instruct-v0.2
56+
```
57+
where `$HF_TOKEN` is the HuggingFace Auth Token set as an environment variable. Pulling the image for the container may take up to 20 minutes.
58+
59+
Once the deployment is finished, the model is available by default at http://0.0.0.0:8000.
60+
61+
# Model Calling
62+
63+
The Mistral model is available through a OpenAI compatible API. As a prerequisite you must have the curl package installed.
64+
65+
```
66+
sudo apt-get install curl
67+
```
68+
69+
Below are 3 examples of curl requests. The `json_pp` utility (JSON Pretty Printer) eases the model output reading by printing the JSON data in a legible, indented format.
70+
71+
* Check the model version available in the container:
72+
73+
```
74+
curl http://localhost:8000/v1/models | json_pp
75+
```
76+
77+
* Complete a sentence:
78+
79+
```
80+
curl http://localhost:8000/v1/completions \
81+
-H "Content-Type: application/json" \
82+
-d '{
83+
"model": "mistralai/Mistral-7B-Instruct-v0.2",
84+
"prompt": "A GPU is a",
85+
"max_tokens": 128,
86+
"temperature": 0.7
87+
}' | json_pp
88+
```
89+
90+
* Chat
91+
92+
```
93+
curl http://localhost:8000/v1/chat/completions \
94+
-H "Content-Type: application/json" \
95+
-d '{
96+
"model": "mistralai/Mistral-7B-Instruct-v0.2",
97+
"messages": [
98+
{"role": "user", "content": "Which GPU models are available on Oracle Cloud Infrastructure?"}
99+
]
100+
}' | json_pp
101+
```
102+
103+
# Notes
104+
105+
Mixtral8x7B is much more greedy that Mistral 7B and it will not fit in a single A10 GPU VM, nor a quad A10 GPU BM. Therefore it is necessary to either:
106+
* Increase the size of the shape to a BM.GPU4.8 (8 x A100 40 GB GPUs).
107+
* Use a quantized version such as [TheBloke/mixtral-8x7b-v0.1-AWQ](https://huggingface.co/TheBloke/mixtral-8x7b-v0.1-AWQ). However, AWQ quantization on vLLM is not fully optimized yet so speed might be lower than the original model.
108+
109+
```
110+
docker run --gpus all \
111+
-e HF_TOKEN=$HF_TOKEN -p 8000:8000 \
112+
vllm/vllm-openai:latest \
113+
--host 0.0.0.0 \
114+
--port 8000 \
115+
--model TheBloke/mixtral-8x7b-v0.1-AWQ \
116+
--quantization awq \
117+
--tensor-parallel-size 4 \
118+
--gpu-memory-utilization 0.95 \
119+
--enforce-eager
120+
```
121+
122+
# Resources
123+
124+
* [vLLM Documentation](https://docs.vllm.ai/en/latest/)
125+
* [Mistral Documentation](https://docs.mistral.ai/)

cloud-infrastructure/multicloud/README.md

Lines changed: 29 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -2,36 +2,49 @@
22

33
Multicloud architectures leverage the coordinated use of cloud services from two or more public cloud vendors. Organizations use multicloud environments to distribute computing resources and minimize the risk of downtime and data loss. Organizations may also adopt two or more public cloud providers for their unique capabilities.
44

5+
Reviewed: 08.04.2024
56

6-
## Team Publications
7+
# Team Publications
8+
9+
## Video Content
710

8-
- [Learn About Multicloud Architecture Framework](https://docs.oracle.com/en/solutions/learn-about-multicloud-arch-framework/index.html)
911
- [Multicloud Is The New Norm - Part I](https://www.youtube.com/watch?v=WzyJiAXldDM)
1012
- [Multicloud Is The New Norm - Part II](https://www.youtube.com/watch?v=27L_ZeF1o9Q)
1113
- [Multicloud Is The New Norm - Part III](https://www.youtube.com/watch?v=qesCk_nIvY8)
12-
- [Multicloud in the Mainstream - Discovery Report](https://www.oracle.com/uk/cloud/multicloud/industry-mainstream/)
13-
- [Google Cloud and OCI making the most of multicloud](https://services.google.com/fh/files/misc/google-cloud-oci-guide.pdf)
14-
- [Use AWS endpoint service to securely connect applications to Oracle Autonomous Database](https://docs.oracle.com/en/solutions/adb-endpoint-in-aws/index.html)
14+
15+
## Blog Posts
16+
17+
- [DNS in multicloud disaster recovery architectures](https://blogs.oracle.com/cloud-infrastructure/post/dns-in-multicloud-disaster-recovery-architectures)
18+
- [Oracle OCI — encrypted connection to AWS, Azure, GCP](https://aviatrix.com/blog/oracle-oci-encrypted-connection/)
19+
- [Oracle and Microsoft Simplify the Path to Multicloud Choice](https://blogs.oracle.com/cloud-infrastructure/post/simplify-path-to-multicloud-choice)
1520
- [ODSA versus the OCI-Azure Interconnect](https://blogs.oracle.com/cloud-infrastructure/post/odsa-versus-oci-azure-interconnect)
1621
- [Network latency using OCI-Azure Interconnect and best practices](https://blogs.oracle.com/cloud-infrastructure/post/network-latency-oci-azure-best-practices)
1722
- [Transitive cloud routing in multicloud architectures](https://blogs.oracle.com/cloud-infrastructure/post/transitive-routing-in-multicloud-architectures)
18-
- [Connect Oracle Data Safe to Oracle databases on multicloud and hybrid cloud environments](https://docs.oracle.com/en/solutions/data-safe-multicloud-ods-hybrid/index.html)
19-
- [Extend your high availability database architecture to multicloud using OCI GoldenGate replication](https://docs.oracle.com/en/solutions/oci-multicloud-db-replication-goldengate/index.html)
20-
- [Deploy multicloud Oracle Database Service for Microsoft Azure in a hub and spoke topology](https://docs.oracle.com/en/solutions/odsa-azure-hub-spoke/index.html)
2123
- [Implementing Oracle Database for Azure with a hub-and-spoke network](https://blogs.oracle.com/cloud-infrastructure/post/implementing-oracledb-azure-hubandspoke-network)
2224
- [Connecting OCI to GCP with Equinix Network Edge](https://blogs.oracle.com/cloud-infrastructure/connecting-oracle-cloud-infrastructure-to-google-cloud-platform-with-equinix-network-edge-cloud-router)
2325
- [Connecting OCI to AWS with Megaport](https://blogs.oracle.com/cloud-infrastructure/connecting-oracle-cloud-infrastructure-to-amazon-vpc-with-megaport-cloud-router)
24-
- [Connecting on prem network to OCI with Colt on demand](https://blogs.oracle.com/cloud-infrastructure/connecting-your-on-premises-network-to-oracle-cloud-with-colt-on-demand)
25-
- [Deploy a Multicloud split-stack solution across OCI, AWS and GCP](https://docs.oracle.com/en/solutions/oci-aws-gcp-multicloud/index)
26-
- [Best practices for hybrid and Multi Cloud  OCI networking design](https://docs.oracle.com/en/solutions/oci-best-practices-networking/index.html)
27-
- [Multi cloud patterns to deploy Oracle Applications (EBS, PeopleSoft, JDEdwards) to Azure and OCI](https://learn.microsoft.com/en-us/azure/virtual-machines/workloads/oracle/oracle-oci-applications)
26+
- [Connecting on-premises network to OCI with Colt on demand](https://blogs.oracle.com/cloud-infrastructure/connecting-your-on-premises-network-to-oracle-cloud-with-colt-on-demand)
27+
- [Oracle and Microsoft bring Oracle Database@Azure to Europe](https://blogs.oracle.com/cloud-infrastructure/post/oracle-microsoft-databaseazure-europe?source=:so:ch:or:awr::::&SC=:so:ch:or:awr::::&pcode=)
28+
29+
## Reference Architectures
30+
31+
- [Learn About Multicloud Architecture Framework](https://docs.oracle.com/en/solutions/learn-about-multicloud-arch-framework/index.html)
32+
- [Use AWS endpoint service to securely connect applications to Oracle Autonomous Database](https://docs.oracle.com/en/solutions/adb-endpoint-in-aws/index.html)
33+
- [Resolve DNS records seamlessly in OCI multicloud architectures](https://docs.oracle.com/en/solutions/resolve-dns-oci/index.html#GUID-84375E55-F207-4A72-84E8-C17CE0CE6BF3)
34+
- [Connect Oracle Data Safe to Oracle databases on multicloud and hybrid cloud environments](https://docs.oracle.com/en/solutions/data-safe-multicloud-ods-hybrid/index.html)
35+
- [Extend your high availability database architecture to multicloud using OCI GoldenGate replication](https://docs.oracle.com/en/solutions/oci-multicloud-db-replication-goldengate/index.html)
36+
- [Deploy multicloud Oracle Database Service for Microsoft Azure in a hub and spoke topology](https://docs.oracle.com/en/solutions/odsa-azure-hub-spoke/index.html)
37+
- [Deploy a Multicloud split-stack solution across OCI, AWS, and GCP](https://docs.oracle.com/en/solutions/oci-aws-gcp-multicloud/index)
38+
- [Best practices for hybrid and multicloud OCI networking design](https://docs.oracle.com/en/solutions/oci-best-practices-networking/index.html)
39+
- [Multi cloud patterns to deploy Oracle Applications (EBS, PeopleSoft, JD Edwards) to Azure and OCI](https://learn.microsoft.com/en-us/azure/virtual-machines/workloads/oracle/oracle-oci-applications)
2840
- [Extend Active Directory integration in Hybrid Cloud](https://docs.oracle.com/en/solutions/integrate-oci-msft-ad/)
29-
- [Oracle OCI — encrypted connection to AWS, Azure, GCP](https://aviatrix.com/blog/oracle-oci-encrypted-connection/)
30-
- [DNS in multicloud disaster recovery architectures](https://blogs.oracle.com/cloud-infrastructure/post/dns-in-multicloud-disaster-recovery-architectures)
31-
- [Oracle and Microsoft Simplify the Path to Multicloud Choice](https://blogs.oracle.com/cloud-infrastructure/post/simplify-path-to-multicloud-choice)
3241
- [Establish a multi-cloud private network connectivity through Oracle Integration Cloud](https://docs.oracle.com/en/solutions/multi-cloud-with-oic/index.html)
33-
- [Multicloud split-stack architecture with OCI, GCP and AWS](https://docs.oracle.com/en/solutions/oci-aws-gcp-multicloud/index.html)
42+
- [Multicloud split-stack architecture with OCI, GCP, and AWS](https://docs.oracle.com/en/solutions/oci-aws-gcp-multicloud/index.html)
43+
44+
## Certifications
45+
3446
- [Become an OCI Multicloud Architect Associate](https://mylearn.oracle.com/ou/learning-path/become-an-oci-multicloud-architect/120606)
47+
- [Become an Oracle Database@Azure Specialist](https://mylearn.oracle.com/ou/learning-path/become-an-oracle-databaseazure-specialist/135857)
3548

3649
# License
3750

cloud-infrastructure/multicloud/azure-interconnect/README.md

Lines changed: 4 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -2,12 +2,14 @@
22

33
Oracle Interconnect for Microsoft Azure provides organizations with a simple migration path to a multicloud environment that includes Oracle Database capabilities such as Oracle Exadata Database Service, Autonomous Database, and MySQL HeatWave. Customers can innovate using the best of Oracle Cloud Infrastructure (OCI) and Microsoft Azure with seamless interoperability.
44

5+
Reviewed: 08.04.2024
6+
57
# Table of Contents
68

79
- [Team Publications](#team-publications)
810
- [Useful Links](#useful-links)
911

10-
## Team Publications
12+
# Team Publications
1113

1214
- [Simplify OCI and Azure Interconnect encryption and observation for multicloud using Aviatrix](https://blogs.oracle.com/cloud-infrastructure/post/simplify-ociazure-interconnect-observation-using-aviatrix)
1315
- [Enabling redundancy between 2 Azure interconnect connections](https://www.ateam-oracle.com/post/oci-azure-interconnect-advanced-scenarios---part2)
@@ -25,7 +27,7 @@ Oracle Interconnect for Microsoft Azure provides organizations with a simple mig
2527
- [Deploy the Oracle Cloud and Microsoft Azure Interconnect Using Hub-and-Spoke Topology](https://docs.oracle.com/en-us/iaas/Content/Resources/Assets/whitepapers/deploy_oracle_cloud_and_microsoft_azure_interconnect_using_hub_spoke_topology.pdf)
2628

2729

28-
## Useful Links
30+
# Useful Links
2931

3032
- [Original Service Press release](https://www.oracle.com/corporate/pressrelease/microsoft-and-oracle-to-interconnect-microsoft-azure-and-oracle-cloud-060519.html)
3133
- [Service Landing Page](https://www.oracle.com/cloud/azure/interconnect/)

cloud-infrastructure/multicloud/connectivity-partners/README.md

Lines changed: 3 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -2,12 +2,9 @@
22

33
Oracle has integrated the FastConnect service with a geographically diverse set of IP, VPN, and Ethernet network providers and cloud exchanges to make it easy for you to establish a connection to Oracle Cloud services. You can establish a FastConnect connection from your on-premise or other cloud service providers to the data center where your Oracle Cloud resources are provisioned by requesting cloud connectivity services from any of Oracle's FastConnect partners.
44

5-
# Table of Contents
6-
7-
1. [Useful Links](#useful-uinks)
8-
9-
10-
## Useful Links
5+
Reviewed: 08.04.2024
6+
7+
# Useful Links
118

129
- [FastConnect Connectivity Partners ](https://www.oracle.com/uk/cloud/networking/fastconnect/connectivity-partners/)
1310
- [FastConnect Partners by Region](https://www.oracle.com/uk/cloud/networking/fastconnect/providers/#emea)

cloud-infrastructure/multicloud/federation-azure/README.md

Lines changed: 7 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -1,15 +1,16 @@
11
# OCI Implementation Series: Multicloud Federation Setup (Azure) and Solution Efficiency with JIT
2-
3-
# Table of Contents
42

5-
1. [Team Publications](#team-publications)
3+
Multicloud is a cloud computing strategy that incorporates the best services from more than one public cloud provider to deploy a solution. This enables customers to meet diverse business requirements and ensure they are choosing the best cloud provider for each workload, based on capabilities, requirements, and performance.
4+
5+
With Oracle Cloud Infrastructure (OCI) being one of the leading cloud platforms, organizations often find themselves in the position of utilizing OCI services along with at least one other cloud provider to meet their business requirements.
6+
7+
Multicloud IAM solution simplifies administration and plays a vital role in ensuring secure access to resources, controlling permissions, and maintaining compliance. Oracle Cloud Lift Services has implemented solutions that address these challenges for various enterprise customers in multicloud environment setup.
8+
9+
Reviewed: 08.04.2024
610

711
# Team Publications
812

913
- [OCI Implementation Series: Multicloud Federation Setup (Azure) and Solution Efficiency with JIT](https://blogs.oracle.com/futurestate/post/oci-implementation-series-multicloud-federation-setup-azure-solution-efficiency-with-jit). <!-- LIFT -->
10-
- Multicloud is a cloud computing strategy that incorporates the best services from more than one public cloud provider to deploy a solution. This enables customers to meet diverse business requirements and ensure they are choosing the best cloud provider for each workload, based on capabilities, requirements, and performance.
11-
- With Oracle Cloud Infrastructure (OCI) being one of the leading cloud platforms, organizations often find themselves in the position of utilizing OCI services along with at least one other cloud provider to meet their business requirements.
12-
- Multicloud IAM solution simplifies administration and plays a vital role in ensuring secure access to resources, controlling permissions, and maintaining compliance. Oracle Cloud Lift Services has implemented solutions that address these challenges for various enterprise customers in multicloud environment setup.
1314

1415
# License
1516

cloud-infrastructure/multicloud/integrate-okta-iam/README.md

Lines changed: 4 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -1,15 +1,14 @@
11
# OCI Implementation Series: Integrate Okta IAM with OCI Identity Domain in Hybrid Cloud Environment Setup
22

3-
# Table of Contents
3+
As organizations increasingly adopt multi-cloud and hybrid cloud infrastructures, managing identity and access across multiple environments can become a complex and challenging task. Traditional Identity and Access Management (IAM) solutions often struggle to keep up with the evolving cloud landscape, leaving organizations vulnerable to security breaches and compliance issues.
44

5-
1. [Team Publications](#team-publications)
5+
A multi-cloud/hybrid cloud IAM solution offers a comprehensive approach to managing identities, access, and security across multiple environments. By providing a centralized platform that can manage access to multiple cloud services and applications, a multi-cloud/hybrid cloud IAM solution simplifies IAM administration and strengthens security. Oracle Cloud Lift Services has implemented solutions that address these challenges.
6+
7+
Reviewed: 08.04.2024
68

79
# Team Publications
810

911
- [OCI Implementation Series: Integrate Okta IAM with OCI Identity Domain in Hybrid Cloud Environment Setup](https://blogs.oracle.com/futurestate/post/oci-implementation-series-integrate-okta-iam-with-oci-identity-domain-in-hybrid-cloud-environment-setup). <!-- LIFT -->
10-
- As organizations increasingly adopt multi-cloud and hybrid cloud infrastructures, managing identity and access across multiple environments can become a complex and challenging task. Traditional Identity and Access Management (IAM) solutions often struggle to keep up with the evolving cloud landscape, leaving organizations vulnerable to security breaches and compliance issues.
11-
- A multi-cloud/hybrid cloud IAM solution offers a comprehensive approach to managing identities, access, and security across multiple environments. By providing a centralized platform that can manage access to multiple cloud services and applications, a multi-cloud/hybrid cloud IAM solution simplifies IAM administration and strengthens security. Oracle Cloud Lift Services has implemented solutions that address these challenges.
12-
1312

1413
# License
1514

0 commit comments

Comments
 (0)