Domain Specific Containers (DSC): Concept


Domain specific containers (DSC) are building blocks of domain platform model described in previous post. DSC allow applications to be executed in a safe way that guarantees data ownership will be retained by the platform.

Overview

The general idea of containers implies some form of isolation and control. Modern application containers provide means to control low level machine resources (CPU, RAM, disk, network). They are perfect tools for cloud infrastructure providers which allow them basically to rent out computing resources.

The core idea of domain platform model is ability to retain data ownership. It allows company to rent out user’s data. It means user’s data can be retrieved, processed and presented to end user – and since this is data rent it cannot be taken away. Domain specific containers (DSC) aimed to provide infrastructure that makes rent out data model possible.

DSC designs requires understanding of applications structure and their requirements. These aspects explained below in more detail.

Data flow models

In this section we analyze data flow between different components of system.
This is high level overview that creates basis for next steps.

One-way flow

The most simple model where application can only read data, process it somehow and present results to user.

One-way flow diagram

Blocks on this diagram.

  • Primary Data Source (DS) – contains original user data, applications can only read from it
  • Application – application code executed in DSC
  • UI – presents data to user

One-way flow with application data source

Same as previous model with additional Application Data Source (DS).

One-way flow with DS diagram

Application Data Source (DS) is fully accessible for Application code running in DSC. It supports full set of read and write operations.

Two-way flow with application data source

This model adds important application ability to receive user input from UI part.

Two-way flow with DS diagram

As result Application executed in DSC is able to:

  • Read data from Primary DS
  • Read data from Application DS
  • Write data to Application DS
  • Show results in UI
  • Receive input from UI

Deployment models

Here we need to understand how all application components will be deployed.
It’ll allow to define requirements for specific DSC implementations.

All models assume that Primary Data Source is deployed remotely (it’s server based).

Client only

Client only diagram

All components (excluding Primary DS) are deployed on the client device.

Client with remote DS

Client with remote DS diagram

Similar to previous model. The difference is that Application DS is deployed remotely. It introduces some latency and access reliability concerns but enables applications to:

  • persist data reliably,
  • share data between application instances (only in some special case because general data sharing enables potential data leaks).

Client with remote Application

Client with remote app diagram

This model assumes remote (server based) deployment of Application code running is DSC.

Client-server

Client-server diagram

Combines previous two models by allowing part of Application to be deployed remotely and part on the client side. Application DS is remote but we can also allow some persistence on device as well (not shown on the diagram).

From analysis of data flows and deployment models we can see that DSC implementation required on the client side and on the server side. More specifically for: cloud environments, mobile devices and web browsers.

Communication channels

Let’s review communication between system components in more detail.

Channels diagram

1) Data based communication (read only)

  • Application sends read request
  • Receives data response

2) Data based communication (read/write)

  • Application sends CRUD requests (create / read / update / delete)
  • Receives data response or status

3) UI based communication

  • Application sends UI components
  • Receives user input events (click, text change, key press, etc)

DSC implementations together with core platform should be able to support all these communication methods.

To maintain platform integrity and data safety all communications between components must be strictly validated at run time.

At this point DSC concept is described at a high level, some adjustments are still possible but they are not going to be dramatic. Further work need to be done on planning of actual DSC implementations and defining communication protocols / restrictions.