Deployment Options

Platform as a Service

Updated: October 16, 2020

This document explains how the Nuxeo Platform (and can be used to provide an application factory system that allows to instantiate on demand applications for a given configuration.


Target Use Cases

SaaS Deployment

When building an application you want to deliver as a service, there are several problems you need to address:

  • Scalability: You need to be able to address quickly increasing load and your customers want good performance
  • Price efficiency: Your customers want good performance for a good price
  • Customization: Each customer wants to be able to slightly tweak the application to their own requirements
  • Customization maintenance This point can quickly become a major issue

Subsidiaries / Regional Deployment

Inside a large company, especially when it is international, it is very common to require the same application deployed for several departments or several countries.

Actually, even if at the beginning all departments or countries want the same application at some point they usually identify some "small differences" that justify that they need some custom configuration and possibly their own deployment too.

Usually the most critical points to handle are:

  • Urbanization: You don't want each department to deploy its own software solution.
  • Maintenance and upgrade: You don't want to multiply maintenance price by the number of logical deployments.
  • Deployment speed: Deploying a new "instance" should be fast and painless.

Projects with Common Requirements

For similar reasons, inside a large company you can easily have several projects that would benefit to share the same common ground. Typically if you have several Case Management projects for different departments or several DAM projects for different countries.

Even if these are separated projects with potentially different life cycles, it can make sense to define a kind of "group wide" common ground:

  • To avoid redoing the same customization several times
    • Authentication and security requirements
    • Company branding integration
    • Baseline workflow
  • To mutualize maintenance
  • To speed up projects deployment

Why Nuxeo Platform Does Help

All the use cases above have in common that:

  • You first need to build a common application, ideally without starting from the ground.
  • You want then to be able to customize it for a project or client, without compromising upgrade possibilities.

The Nuxeo Platform provides the tools for achieving that:

  • A component model and packaging model so that you can have three cleanly separated layers
    • The components provided by the Nuxeo Platform that handle all the infrastructure issue
    • The components you want to build for your custom application
    • The components each project / client may want to build
  • Tools to make configuration and extension building easy
    • Nuxeo Studio: to manage all your business configuration
    • Nuxeo IDE / Nuxeo Codenvy: to build extensions
  • A packaging model and deployment system

Single Application with Multi-Tenants Approach

Share Everything Approach

The classical approach, used by several solutions, is to build a multi-tenant application: There is only one big application, which provides isolation features so each tenant of the application feels "as if it was their application"_._

This approach exists for a very long time and has the big advantage to provide a complete sharing of all resources which may seem to be the most economic option.


The nuxeo-multi-tenants add-on provides a multi-tenant isolation system that adds to the Nuxeo Platform:

  • A configurable notion of tenant: By default a tenant is a domain.
  • A user isolation at tenant level: Except for main administrators, users and groups and defined and managed at tenant level.
  • A data isolation at tenant level: Documents are isolated a on per tenant basis using a security policy and ACLs.

Depending on the way you define a tenant, you could also use completely separated repositories for each tenants (allowing to have physically separated Databases and BinaryStores).

Limits of the Multi-Tenant Model

The multi-tenant model has some intrinsic limits.

All Resources Are Indeed Shared

The first drawback is that since all resources are shared inside the same application, one tenant can create issues by consuming too much resources or creating errors.

The application layer can manage some quotas (the platform provides that for the storage for example), but as long as the tenant isolation remains at application level, it is not possible to prohibit one tenant to use too much CPU, memory or bandwidth.

Configuration and Customization Has to Be Limited

As long as all tenants run inside the same application, the configuration and customization must be limited.

  • Contributing too much code may endanger the other tenants
    • There is no classloading isolation enforced
    • See previous remark about resource consumption
  • Only a very limited number of extension points are tenant aware.
  • Having each tenant declare 10 custom document types can result in a huge number of document types at the repository level
    • Creating hundreds of tables is usually not a scalable path

Availability and Projects Constraints

Concentrating the different projects/customers/tenants on the same application automatically makes it highly critical.

Hosting three non critical applications for three separated continents is easy. But if you run them as three tenants on the same application the maintenance window automatically become very small and the application needs to be HA 24/24 7/7.

In terms of project constraints, it also means that all projects needs to agree on the upgrade schedule: Sadly, when the number of projects increases, getting that agreement can end up being much more complicated than the actual technical upgrade itself.

Scalability Issues

With multi-tenants the easiest solution is to use a single and big database server for all the tenants.

The drawback of the approach is that when you need to scale because you need more performances or simply because you are adding a lot of tenants then you can be stuck because Database servers don't easily scale out.

With Nuxeo, you have the option of using a per-tenant repository, but in this case you need to automate the Database provisioning. This is actually the first step to switch to the other approach described below (Platform as a Service)

BackUp / Restore

As long as the data isolation is handled at application level, you can not rely on system level backup restore tool, at least, not if you want to restore a data from a particular tenant without restoring the data for all tenants. Platform as a Service

How Containers Can Help

As seen earlier, for the classical multi-tenants approach to be really efficient, we would need the tenant isolation to be enforced at system level so that:

  • Tenant resource are really reserved and not accessible to other tenants, including CPU, Memory, I/O.
  • Each tenant should use a different runtime so the configuration from a tenant can never impact another tenant.

Basically the idea is that for each client a set of dedicated containers running Nuxeo is used.

With the container based approach all the problems of a single application with multi-tenants disappear:

  • Each project / client is really isolated
    • At data level
    • At runtime level
    • At resource level
  • Each project can have its own life cycle for upgrades
  • Project can be as custom as needed

That's because we are convinced that this approach has much more value that "standard multi-tenants" that we started investing on the container based approach. A Container Factory is the result of the work we have been doing for on the platform as a service approach. It is an application factory relying on a container-based execution grid.

Basically, is an infrastructure that provides tools to automate the whole process:

  • To define the application configuration by selecting a target platform and the packages to be deployed
  • To provision Nuxeo server containers and DB servers containers
  • To start the containers by deploying the target packages and binding to a URL

Actually also provides a web portal that can be used to configure and provision Nuxeo instances associated to an application configuration.

High Density Containers use the Docker lightweight container technology to multiplex several application containers on a single host.

This approach allows to optimize the run infrastructure, even when running on virtualized environment like AWS:

  • Multiple containers can be run on the same VM
  • Containers can be passivated when not needed anymore
  • Containers can be restarted very quickly when needed

Because lightweight containers are cheaper to create or to shutdown, this 2-level container architecture allows to have a very reactive provision policy so that we can quickly scale the application up or down.

About Internals

To manage this container factory is based on several key components:

  • Docker: as the lightweight container system
  • CoreOS: as system host for running the Docker container
    • Fleet: as manager to distribute services across the cluster
    • etcd: as distributed registry
  • Gogeta: as dynamic load balancer
  • manager: the Nuxeo based application used to control the whole system can run on AWS leveraging RDS and S3 or on Vagrant / VMWare. The full technology stack can be available for on premises installation., IDM and Other Services

We are currently working on adding more services to so that applications running on can be even easier to deploy.

The first service we will introduce (Q4 2014/Q1 2015) is Nuxeo IDM for having a central User management service. The idea is to allow to manage all the users from one central location and simply assign them to the different based applications.

We also plan to provide a service gateway so that based applications can easily leverage Cloud based services like Push, Conversion.

Examples of Usage


One of the use cases we designed for is also to be able to host the Nuxeo Trial service where users must be able to:

  • Select their target configuration, including an associated Studio project
  • Start a Nuxeo instance running this configuration, and possibly in several environments

On the Nuxeo side, we must ensure that:

  • We don't run too many containers unless really needed
    • Because we do pay the containers to AWS !
  • We provide a simple UI to manage all that provides this service and can be used as is by SaaS providers for similar purpose.

Developments and Test Environment

We realized that in some context / company, building the required development environment can be challenging or at least be time consuming:

  • Having servers or VMs so that developers can test their work
  • Having servers or VMs to run the acceptance tests

Here again provides a clean solution, we just need to have the underlying IaaS provider: either AWS, Vagrant or VMWare.

Subsidiaries / Regional Deployment

Let's consider a use case where in a large international company each region must have it's own "flavor" of the corporate DAM.

However, when looking in more details, each region should have access to two repositories:

  • A corporate repository, shared across all regions
  • A local repository, specific to one region

The Nuxeo Platform and provides the required infrastructure:

  • Platform as a Service
  • Multi-repositories support

We'd love to hear your thoughts!

All fields required