ARCHITECTURE CONCEPT

Where should data preparation happen?

There is no single execution environment that is right for every data path. Data preparation can run on conventional server infrastructure or, where the workload and platform justify it, on programmable infrastructure closer to the source.

The architectural role can remain the same even when the execution environment changes.

Move preparation to the point where it creates value.

In many environments, preparation happens inside downstream applications because that is where the data first becomes useful.

Moving common preparation earlier can reduce repeated work when several downstream systems need the same underlying data.

That does not mean preparation must always happen at the earliest physically possible point. Placement should reflect the infrastructure, workload, operational requirements, supported platforms, and the value created by moving the work.

Conventional server infrastructure

For many environments, existing or newly deployed servers are the practical place to perform shared data preparation.

Software can receive supported incoming data near its point of origin and transform it in flight without requiring specialized programmable infrastructure.

This approach allows the preparation role to be introduced into infrastructure organizations already operate.

ClaraStream Core performs Listen + Transform on conventional server infrastructure.

Programmable infrastructure closer to the source

Some environments may benefit from moving the same preparation role onto programmable infrastructure closer to where data enters the system.

This can be relevant when processing location, performance requirements, deployment architecture, or the capabilities of the underlying platform make acceleration valuable.

Using programmable infrastructure does not change the role of the preparation layer. It changes where and how that role executes.

ClaraStream Edge is software being developed for accelerated data preparation on programmable infrastructure closer to the data source.

The initial ClaraStream Edge implementation is being developed for the Napatech F2070X DPU. Additional platform implementations would be individually ported, validated, and supported.

The same role. Different execution environments.

CORE

Listen + Transform

Conventional server infrastructure

Software-based preparation using standard server environments

EDGE

Listen + Transform

Programmable infrastructure closer to the source

Software designed to use supported acceleration resources

Core and Edge are alternatives for the preparation role. Data does not need to pass through Core and then Edge.

What should influence placement?

SOURCE PROXIMITY

Where does the data originate, and is there value in preparing it before it moves farther through the environment?

PERFORMANCE REQUIREMENTS

Does the workload justify acceleration or a different execution environment?

DEPLOYMENT CONTROL

What infrastructure can the organization operate, support, and secure effectively?

PLATFORM SUPPORT

Is the target execution platform specifically supported, ported, and validated for the preparation software?

OPERATIONAL FIT

Does moving preparation change deployment complexity in a way that creates more value than burden?

Preparation and distribution remain separate decisions.

Choosing where preparation runs does not determine where the data must ultimately go.

Core or Edge performs the preparation role. Stream provides continuous distribution of the prepared result to downstream subscribers.

Edge prepares. Stream distributes.

Placement inside the Data Movement Plane™

The Data Movement Plane™ architecture separates the role of preparing data from the systems that ultimately use it. That allows preparation to be placed according to infrastructure requirements rather than being permanently tied to each downstream application.

Compare data preparation and distribution Explore ClaraStream Core Explore ClaraStream Edge