Designing an efficient data loading architecture for your MBOs requires a good understanding of the data to be mobilized and the data sources that provide that data.
While you can use Unwired WorkSpace to quickly create a working prototype, developing a production environment that is scalable requires careful planning and detailed knowledge of the data movement between the CDB and the EIS.
You must understand the characteristics of the data that is to be mobilized:
- Read/Write ratio – read-only, read/write, mostly read, mostly write
- Sharing – private versus shared
- Change source – mobile only, EIS only, mobile and EIS
- Change frequency
- Freshness expectation
- Access pattern – peak/valley or distributed
- Data volume
Common Data Characteristics| Reference data |
- Mostly read or even read-only
- Usually shared between users
- Generally updated by EIS
- Infrequent or scheduled changes
- Able to tolerate stale data
- May be concentrated during initial deployment, occasional thereafter
- Large to very large data volume is possible
|
| Transactional data |
- Read and write
- Usually private but can share with other users
- Updated by both mobile application and back end possible
- Moderate change frequency
- High freshness expectation
- Access pattern varies depends on use case: morning/evening, or throughout the day
- Moderately low data volume (not including historic data which is considered as reference)
|
Data Sources
It is important to understand how, what, and when data can be obtained from the EIS to fill the CDB. What are the characteristics of the EIS to consider for data loading?
- Efficiency of the interface:
- Protocol – JCO, Web Services, JDBC
- API – number of interactions required to retrieve the data
- Push or pull
- Reaction to peak load
- Availability of internal cache for frequently accessed data