Self Hosted Server Installation Requirements

Prev Next

Server Installation Requirements

Setting up the server is a critical first step in any KYP.ai implementation. Before starting, decide whether to proceed with a cloud-based or on-premise setup:

  • Cloud servers are deployed and maintained by KYP.ai or an authorized partner.

  • On-premise servers are installed and maintained by your internal IT team.

All system components run in Docker containers on a single machine. The specs below apply to virtualized environments. For physical (bare metal) setups, contact KYP.ai for tailored requirements.


Infrastructure Requirements

💡 Key Notes:

  • The setup assumes server virtualization.

  • Resource needs to depend on data volume, number of users, and enabled features.

  • For environments with 1000+ users, consult the KYP.ai Team.


🖥️ Configuration for OPS Monitoring (No Screenshot Processing)

Users

< 30

30 – 300

300 – 1000

1000 – 5000

CPU

4 vCPU

8 vCPU

8 vCPU

16 vCPU

Memory

32 GB

32 GB

64 GB

64 GB

Storage*

100 GB / IOPS ~1000**

250 GB / IOPS ~1500**

600 GB / IOPS ~3000**

1.5 TB / IOPS ~3000**

📸 Configuration for Process Discovery / PoC (With Screenshot Processing)

Users

< 30

30 – 300

300 – 1000

1000 – 5000

CPU

4 vCPU

8 vCPU

16 vCPU

32 vCPU

Memory

32 GB

64 GB

128 GB

192 GB

Storage*

400 GB / IOPS ~3000**

1.5 TB / IOPS ~4500**

4 TB / IOPS ~6000**

10 TB / IOPS ~6000**

* Fast, attached disk as the additional partition. File system: ext4 or xfs with ftype=1 (d_type support enabled).

** Storage calculated based on: 30 days screenshots retention, 20% usage of screenshot apps and a 12 month contract.


☁️ Public Cloud – Recommended Instance Types

The following tables provide cloud instance recommendations for different user ranges and use cases. Prices may vary depending on region and provider.

🔧 Operations Steering (Excluding Screenshot Processing)

Users

AWS

GCP

OVH

Azure

OCI

< 30 users

r5.xlarge (GP3)

n2-highmem-4 (pd-standard)

r3-32 (high speed storage)

Standard_E4as_v5 (premium SSD)

VM.Standard3.Flex (2 OCPU, 32 GB)
with lower cost volume

30 – 300 users

m4.2xlarge (GP3)

n2-standard-8 (pd-standard)

b3-32 (high speed storage)

D8as v4 (premium SSD)

VM.Standard3.Flex (4 OCPU, 32 GB)
with balanced volume

300 – 1000 users

r5.2xlarge (GP3)

n2-highmem-8 (pd-standard)

r3-64 (high speed storage)

Standard_E8as_v5 (premium SSD)

VM.Standard3.Flex (4 OCPU, 64 GB)
with balanced volume

1000 – 5000 users

m5.4xlarge (GP3)

n2-standard-16 (pd-ssd)

b3-64 (high speed storage)

D16as v4 (premium SSD)

VM.Standard3.Flex (8 OCPU, 64 GB)
with balanced volume

📸 Process Discovery (Including Screenshot Processing)

Users

AWS

GCP

OVH

Azure

OCI

< 30 users

r5.xlarge (GP3)

n2-highmem-4 (pd-standard)

r3-32 (high speed storage)

Standard_E4as_v5 (premium SSD)

VM.Standard3.Flex (4 OCPU, 32 GB)
with lower cost volume

30 – 300 users

r5.2xlarge (GP3)

n2-highmem-8 (pd-standard)

r3-64 (high speed Gen2 storage)

Standard_E8as_v5 (premium SSD)

VM.Standard3.Flex (4 OCPU, 64 GB)
with balanced volume

300 – 1000 users

r5.4xlarge (GP3)

n2-highmem-16 (pd-standard)

r3-128 (high speed Gen2 storage)

Standard_E16as_v5 (premium SSD)

VM.Standard3.Flex (16 OCPU, 128 GB)
with balanced volume

1000 – 5000 users

m5.12xlarge (GP3)

n2-standard-48 (pd-ssd)

r3-256 (high speed Gen2 storage)

Standard_E32as_v5 (premium SSD)

VM.Standard3.Flex (48 OCPU, 192 GB)
with balanced volume


Network Requirements

🔓 Required Open Ports

  • 80 – Redirects to 443

  • 443 – Data transfer and frontend access (HTTPS)

  • 22 – SSH access for maintenance

🌐 External Access

  • Allow traffic to:

    • CentOS repositories

    • Docker Hub (https://download.docker.com)

    • *.kyp.ai for installation & monitoring

  • Customer workstations → Server on port 443

  • Register DNS entry for the server and workstations

  • Generate Valid SSL certificate for the server domain name (FQDN) and provide to KYP.ai

  • HTTPS traffic cannot be filtered for specific phrases or conditions

  • Minimum bandwidth: 2 Mb/s per 50 users

  • OPTIONAL: If Concierge extension is enabled with Open AI LLM, then outbound traffic on port 443 is required for Concierge to work https://api.openai.com/v1

🔄 Optional Network Configurations

  • In case of Load Balancer usage:

    • Register DNS entry for the load balancer

    • SSL certificate generated for the load balancer in desired domain added to trusted certificates store

    • Load Balancer traffic on port 443 should be redirected to the server on port 80

  • Data collection monitoring outbound communication port to email server on port 587/465

  • LDAP/Active Directory communication to LDAP server on port 389/636

  • SSO (Azure AD):

    • Access https://login.microsoftonline.com and https://graph.microsoft.com


Average daily traffic received for one user

If you want to check the average daily traffic generated by a single user, please refer to this article.


Backup Policy

Valuable data and configurations are encouraged to be backed up regularly of KYP.ai databases. While backups may take up some disk storage space, it will save you from future data loss. Protect your precious assets and ensure peace of mind.

  • Application backup is necessary to protect against data loss or corruption caused by hardware failure, software malfunction, cyber-attacks, or human error.

  • Losing applications and their data can result in downtime, financial losses, and reputational damage.

  • Regular backups allow businesses to recover quickly from disasters and minimize the impact of data loss.

  • Application backups help ensure compliance with regulatory requirements and avoid fines and legal penalties.

  • KYP.ai offers two methods for application backups: configuring backups in a specific directory or using customer-provided storage solutions like AWS S3.

  • Customers must specify the backup frequency and retention period from KYP.ai settings.

  • The best practice is to create automatic daily backups of the database and keep them stored for at least 7 days to enable easy restoration of functions in case of issues.


Software Requirements

🖥️ OS Compatibility

  • Recommended: Ubuntu 20.04.5+

  • Optional: RHEL 8.7+

🧰 Required Tools

Tool

Version

wget

1.14

curl

7.29.0

zip / unzip

3.0 / 6.0

htop / ctop

2.2.0 / 0.7.5

mc

4.8.7

docker

23.0.1

docker-compose

2.16.0

logrotate

3.8.6

vim

7.4

net-tools

latest

apache2-utils (for ubuntu)

latest

jq

latest

httpd-tools (for redhat)

latest

KYP.ai packages are downloaded during installation from the official repository.


Access Requirements (Optional) - only if support is requested

  • HTTPS for KYP.ai Support Team for troubleshooting and support.

  • SSH (root) access for KYP.ai DevOps Team for installation and daily maintenance

  • HTTPS access for KYP.ai Customer Success Team for configuration and data analysis.

  • (Optional) Provision a Windows 11 VM for testing the KYP Connect App


FAQ

Should I choose a cloud or on-premise setup for KYP.ai?

It depends on your organization's preferences:

  • Cloud: Hosted and managed by KYP.ai or an Implementation Partner.

  • On-Premise: Installed and managed by your internal IT team.

Is a backup policy included in the default setup?

Yes. By default, the KYP.ai server:

  • Retains the last 6 days of database backups on local storage.

  • Backup location and retention policy can be changed and need to be agreed upon before deployment.

  • Best practice is to store backups on external storage (e.g., AWS S3).

Who maintains the server after installation?

  • Cloud installations: Fully maintained by KYP.ai.

  • On-premise installations: Maintained by your IT team, with KYP.ai providing support for updates, troubleshooting, and configuration via remote access.