Skip to main content

Data Management

Data Management in SEI is the strategy and tooling used to access, organize, and report on data across diverse systems and environments. Whether you’re working with a single ERP, consolidating multiple business systems, connecting to cloud‑hosted data, or analyzing high‑volume transactional records, SEI provides proven methods to ensure performance, scalability, and usability in every reporting scenario.

This section serves as your guide to choosing the right data management approach, by mapping common operational scenarios to the tools and configurations that fit them.

Data management tools​

ToolDescription
DataSyncSEI's data warehouse and ETL engine used for replication, consolidation, joining, and transformation. Ideal for cloud data, ERP offloading, and preparing datasets for analysis.
Data Model DesignerDesign, organize, and manage the structures that define your integrated data model. Allows you to combine and relate data from multiple sources, shaping it for reporting or OLAP analysis.
OLAP ManagerBuild and manage OLAP cubes for high‑performance analytical processing. Essential for large datasets (> 5M rows) and complex multi‑dimensional reporting needs.

By combining these tools, you can implement the right architecture for your reporting and analytics environment:

  • DataSync for securely copying and preparing data.
  • Data Model Designer for structuring and integrating.
  • OLAP Manager for optimizing speed and scalability at scale.

Scenarios​

This section outlines common data management scenarios — from ERP performance optimization to multi-source integration — and shows which combination of tools (DataSync and/or OLAP Manager) delivers the best results.

When choosing a scenario, check if your dataset exceeds 5 million rows. This threshold determines whether OLAP cubes are recommended for speed and scalability. Also, for multi-source or complex scenarios, check whether data transformation is needed to reconcile mismatched structures or identifiers.

Single data source​

Use this scenario when reporting relies on one source system such as a single ERP. It is best for straightforward reporting needs with low maintenance.

You should evaluate dataset size and performance requirements to choose the right setup. If the dataset is under 5 million rows, direct real‑time ERP access is typically sufficient, and no additional tools are needed. If it exceeds 5 million rows, implement OLAP cubes to boost query performance and reduce ERP load.

ScenarioVolumeData accessDataSyncOLAP
Small volume< 5M rowsReal-timeNoNo
Large volume> 5M rowsReal-timeNoYes

ERP Performance Challenges​

Use this scenario when reporting or data access slows down your ERP system. It is best for keeping ERP responsiveness high during heavy reporting loads.

You should replicate ERP data into a separate environment using DataSync to offload queries. For datasets under 5 million rows, DataSync replication alone is typically sufficient to maintain speed and stability. Over 5 million rows, combine DataSync with OLAP cubes to further improves query performance and reduces processing time.

ScenarioVolumeData accessDataSyncOLAP
Small volume< 5M rowsNear real-timeYesNo
Large volume> 5M rowsNear real-timeYesYes

Cloud data​

Use this scenario when reporting uses data from cloud-hosted sources such as SaaS applications or cloud ERPs. It is best for environments where local performance and scalability are required for analysis.

You must always replicate cloud data to a local environment using DataSync, as direct real-time analysis is not possible. Under 5 million rows, DataSync replication offers sufficient speed. For larger datasets, pair DataSync with OLAP cubes for faster queries and scalability.

ScenarioVolumeData accessDataSyncOLAP
Small volume< 5M rowsNear real-timeYesNo
Large volume> 5M rowsNear real-timeYesYes

Multiple data sources​

Use this scenario when reporting uses data from two or more different systems. It is best for organisations needing either independent source reporting or a unified historical view.

You should first determine whether the sources are compatible (for example, both SQL Server or Oracle) or different (for example, SQL Server and a cloud ERP).

  • For compatible sources with no merge required, OLAP cubes can be applied directly for speed.
  • For different source types, use DataSync to replicate and consolidate data before applying OLAP cubes.
  • For a unified historical view, merge the sources in DataSync, transform structures and identifiers as needed, and add OLAP cubes if the combined dataset exceeds 5 million rows.
ScenarioSource typeGoalDataSyncOLAPTransform
Independent reportingCompatibleNo mergeNoYesNo
Independent reportingDifferentNo mergeYesYesNo
Unified historical reportingAnyMerge and unifyYesYesYes

Transformation needs​

Use this scenario when datasets from different systems need reconciliation due to mismatched identifiers or formats. It is best for complex reporting where consistent data structures are required before analysis.

You should use DataSync as a staging and transformation platform to clean and standardise data. Transformation is mandatory in these cases. If the resulting dataset exceeds 5 million rows after transformation, OLAP cubes should be implemented to maintain speed and scalability.

ScenarioDataSyncOLAPTransform
Complex joins/mismatched IDsNoYes (if large volume)Yes

Consolidation of multiple databases​

Use this scenario when combining multiple databases into a unified reporting view. It is best for organisations needing centralised analysis of data stored across separate databases while maintaining consistency and performance.

You should use DataSync to consolidate all data sources into a staging environment before analysis. Avoid linking directly via SQL views for large datasets, as this can cause significant performance issues. If the consolidated dataset exceeds 5 million rows, add OLAP cubes to improve query speed and scalability.

ScenarioVolumeData accessDataSyncOLAP
Small volume< 5M rowsNear real-timeYesNo
Large volume> 5M rowsNear real-timeYesYes