datavaccinator / dv-vault

DataVaccinator Vault Server Platform

DataVaccinator Vault

This is the DataVaccinator Vault platform. It has to be installed on some Linux environment in order to provide a REST API for pseudonymisation services.

The DataVaccinator Vault platform offers

• Multi tenant use (multiple service providers served on the same platform).

• Word based search using an encrypted index which ensures no readable information on server side.

• Can use Let’s Encrypt certificates out of the box.

• Supports both IPv4 and IPv6 and allows listening on multiple addresses (for the same domain).

• Fully automatic IP whitelisting by only serving API requests from dedicated IP addresses.

• DataVaccinator Vault supports working behind proxy servers and loadbalancers (like HAProxy).

• It supports CORS to avoid issues with cross origin requests.

• Installer-Script supports CentOS, Red Hat, Rocky Linux, Arch Linux, Suse Linux, Ubuntu and Debian systems (all x86_64).

• Installer-Script automatically generates systemd daemon, system user and group, database user and database structure.

What is it good for?

The DataVaccinator Vault protects your sensitive data and information against abuse. At the very moment when data is being generated, the service splits that data and uses advanced pseudonymisation techniques to separate content from identity information. Thus, the DataVaccinator Vault reduces cyber security risks in the health, industry, finance and any other sector and helps service providers, device manufacturers, data generating and data handling parties to manage sensitive data in a secure and GDPR-compliant manner. In contrast to other offerings, DataVaccinator industrialises pseudonymisation, thereby making pseudonymisation replicable and affordable.

Get more information, support and contact at https://www.datavaccinator.com.

It requires (for execution):

• Some Linux server (Intel/AMD, x86_64)

• CockroachDB database and drivers

It requires (for development):

• Some Linux server (Intel/AMD, x86_64)

  • min. 2 CPU cores (vCPU)

  • min. 2GB memory

  • min. 4GB disk

• CockroachDB database and drivers

• go compiler (1.16 or higher)

• php 7 (for migration tool and test scripts)

• Preferable some IDE (like VSCode, Atom, VIM-GO or LiteIDE)

What is missing:

• Currently no support for ARM systems.

• No support for Windows systems.

Get started

In order to run a DataVaccinator Vault, please follow this guide for installation on a dedicated Linux machine:

1. Install CockroachDB on the machine see our hints.

2. Download the latest DataVaccinator setup (…​setup.sh) from GitHub releases page to some temporary directory on your machine.

3. Run it like this:
   sudo sh dv-vault-1.0.0_x86_64_setup.sh

4. Follow the steps from the setup script.

5. Edit the conf.json file to your needs.

   • By default it is located at /opt/vaccinator/config.json folder.

   • Check the docs/vaccinator-config.adoc for details about the configuration options.

   • Remember to use sudo to edit this file.

6. Start the service using
   sudo systemctl start vaccinator

7. Control service start using
   journalctl -et vaccinator

8. Update the default entry with a password and allowed IPs like this (example):
   cd /opt/vaccinator/
./vaccinator -p -j='{"op": "update", "sid": 1, "name": "1st provider", "password": "<myPassword>", "ip": "127.0.0.1"}'

   • Check the docs/Commandline_Operations.adoc for details about the available options and syntax.

License information

DataVaccinator Vault is released as free software under the Affero GPL license (AGPL). You can redistribute it and/or modify it under the terms of this license which you can read by viewing the included LICENSE file or online at www.gnu.org/licenses/agpl.html

dv-vault's Issues

Add limits to check result

The check protocol call should return the limits for max payload, max searchword length and max publish duration.

SearchHash enhancement

Currently, the search hash is built like this:

• Hash the word character by character
• Add random bytes to the generated hash

The drawbacks are:

• False positive hits during search for words with the same beginning and random matches
• The entropy for the word length is only added by max 3 characters

As a suggestion, we might build the hash like this:

• Add a NULL byte to the word
• Add random bytes to the word until it reaches the size of 16 characters. Use 32 if the word is longer than 16 characters.
• Hash the word character by character

Benefits:

• Enhanced entropy for the words
• Impossible to predict anything by the length of the search hash (all 16 or 32 bytes)
• Eliminate possible false positives during search

Allow submitting own VID

Allow submitting own VID in ADD function. In this case, we need to verify uniqenes and return error in case it is not unique.

Add oAuth 2.0 support

Instead of IP whitelisting with ID and password, we might also support oAuth 2.0 for client authentication.

Installer may get some update mode

If the target file and some config file is already found, we may ask for a simple executable update.

If user says YES, we only stop the service, update the executable, run some optional update script and restart the service.

Add command line options

The client executable may get command line options to add/update/delete providers.

By using the same executable, all database functionality is already available and does not need any duplication.

Allow VID and searchword submission as array of strings

Today, the VIDs and searchwords are submitted as a space separated string. The protocol API should alternatively support an array of strings, too.

Affected protocol operations:

• getpublished
• get
• delete
• search

Allow additional authentication using sid and spwd form header values

Currently, authentication uses sid and spwd values from json form field. This generates overhead if a service provider has to supplement a call with those values (decode, add, encode).

Instead of this, we might accept sid and spwd as additional form key/value parameters. This will stop the service provider from having to decode and encode the json parameter.