Skip to content
Graphlogic.ai documentation
Docs guide
How to create a bot
FAQ
What to prepare before creating your Agent
What is Graphlogic.ai
Releases history
How to get started with the Platform
How to create your Agent
How to use Chat Context in your Agent
How to succesfully train your Agent
How to deploy your Agent to channels
Agent usage
How to check Analytics for yur Agents
Additional information
Channel Integration
Resources & infrastructure requirements copy
Onboarding task

icon picker
Resources & infrastructure requirements copy

Required application software

To install the platform, Kubernetes must be deployed.
Requirements for Kubernetes:
Version 1.27.x
StorageClass that supports RWX (ReadWriteMany) —
https://kubernetes.io/docs/concepts/storage/persistent-volumes/
⁠
⁠
Software Required for the Platform to Operate:
Postgres 16.x
Redis 7.x.x
S3 Compatible Storage, ex. Minio
Rabbitmq 3.12.x
Maximum message size 512MB
ClickHouse 23.8
If fault tolerance and high availability are required, the following software should be deployed outside of Kubernetes (see Recommended System Requirements for Deployment). Network accessibility between all necessary software and Kubernetes is necessary.
If fault tolerance and high availability are not required, all services can be hosted in a Kubernetes cluster. In this case, only Kubernetes will be needed from the necessary software to deploy the platform, and the application software will be deployed automatically during the platform installation process (see Minimum System Requirements for Deployment in Kubernetes Cluster).

Environment requirements

Minimum system requirements for deployment in a Kubernetes Cluster

Server Group Name
Purpose
Description
CPU
RAM, GB
Disk Space, GB
Number of VMs
1
masters
Kubernetes orchestrator master nodes
​
4 cores, performance per core not less than Intel® Xeon® Gold 6230
4
20
1/2/2024
2
workers
Kubernetes orchestrator managed nodes
platform services, except NLU, LLM
8 threads, performance per core not less than Intel® Xeon® Gold 6230
32
250
3+
3
nlu-worker
Kubernetes orchestrator managed nodes
NLU services, NER services
not less than 16 cores, performance per core not less than Intel® Xeon® Gold 6230
64
150
1+
4
llm-worker
Kubernetes orchestrator managed nodes
LLM&RAG services
not less than 16 cores, performance per core not less than Intel® Xeon® Gold 6230
32
150
1+
There are no rows in this table
⁠

Recommended system requirements for deployment

Kubernetes Cluster

Server Group Name
Purpose
Description
CPU
RAM, GB
Disk Space, GB
Number of VMs
GPU
1
masters
Kubernetes orchestrator master nodes
​
4 cores, per-core performance not less than Intel® Xeon® Gold 6230
4
20
1-3
​
2
workers
Kubernetes orchestrator managed nodes
Platform services, excluding NLU, LLM
8 threads, per-core performance not less than Intel® Xeon® Gold 6230
32
250
3+
​
3
nlu-worker
Kubernetes orchestrator managed nodes
NLU services NER services
not less than 16 cores, per-core performance not less than Intel® Xeon® Gold 6230
64
150
1+
​
4
llm-worker
Kubernetes orchestrator managed nodes
LLM&RAG services
not less than 8 cores, per-core performance not less than Intel® Xeon® Gold 6230
64
150
1+
GPU graphics card performance not less than NVIDIA A10 16GB
5
voice-workers
Kubernetes orchestrator managed nodes
Voice services: ASR service TTS service
not less than 8 cores, per-core performance not less than Intel® Xeon® Gold 6230
64
150
1+
GPU graphics card performance not less than Nvidia A10 GPU 16GB
There are no rows in this table
⁠

Databases

Server Group Name
CPU, Cores
RAM, GB
Disk Space, GB
Number of VMs
1
Postgres
16
64
1,000
2
2
Rabbitmq
8
16
300
3
3
Clickhouse
8
16
1,000
2+
4
Redis
8
16
300
3
There are no rows in this table
⁠

GPU requirement calculation

ASR TTS operation

Up to 5 languages
Type
200 sessions
300 sessions
400 sessions
ASR+TTS
Nvidia A10 16GB
2xNvidia A10 16GB
2xNvidia A10 16GB
ASR only
Nvidia A10 16GB
Nvidia A10 16GB
2xNvidia A10 16GB
TTS only
Nvidia A10 16GB
Nvidia A10 16GB
2xNvidia A10 16GB
There are no rows in this table
⁠

Monitoring and logging recommendations

The following versions of monitoring and logging tools will be deployed within the Platform:
Elastic Stack 8.5.1
Kube Prometheus Stack 61.3.1
If these tools are already in use by the client, they can be utilized for monitoring and logging the Platform.

Environment Preparation Results

Resources allocated and software configured according to the selected option for fault tolerance
Graphlogic team provided with
Access to Kubernetes and Kubeconfig
IP addresses and administrator accounts in G-specific software if the fault-tolerant delivery option is chosen, when the application is located outside the Kubernetes Cluster
Environment availability organized for the Graphlogic team
Connection data issued for connecting to client monitoring and logging systems if their use is anticipated
Domain name created for accessing the UI platform, optionally, a domain name can be created for the S3 object storage service
Since the platform operates with external environments over https protocol, SSL certificates need to be issued for the above-mentioned domain names

Test Environment

It is assumed that the test environment will be deployed on the same environment prepared for the production environment. As part of environment preparation, only DNS names and certificates for the https protocol for access to the UI (and s3 optionally) will be required
Required application software
Environment requirements
Minimum system requirements for deployment in a Kubernetes Cluster
Recommended system requirements for deployment
Kubernetes Cluster
Databases
GPU requirement calculation
ASR TTS operation
Monitoring and logging recommendations
Environment Preparation Results
Test Environment
Want to print your doc?
This is not the way.
Try clicking the ⋯ next to your doc name or using a keyboard shortcut (
CtrlP
) instead.