SAP HANA in-Memory Computing Engine Architecture Overview


The memory computing engine is responsible for the server activities. It would take care of the memory management Read and Write which includes Row stores / Column stores, Delta memory merges.
It also takes care of the data persistence to the physical storage.
The regular DB activities like session management, memory management for the sessions, transaction consistency, transaction persistence are all taken care in the memory computing engine.
The SQL and other parsers would take client requests for the data retrieval, will define the execution plan, read consistency etc.

1. Repliacting data from Source systems either from SAP or Non-SAP , in case of SAP data is replicated to HANA via SLT incase of Non-SAP and SAP BI we use BODS , replication sever takes care of these activities.
2. Request processing trigeered by Business Object tools or any reporting tools supported by HANA here mainly we have teo types of row store and column store by default it is stored in column store to utilize the best performance of HANA it is better to store your table in Column store, if in case you mainly intrsted on record by record data then column store won’t be of much help, so it depends on your data and purpose of your data.HANA has its own language to interact with underlying tables i.e SQL script , the SQL parser and there are several other componenets for optimized query execution, for OLAP queries we have MDX. These are the several options.
3. Session & User Management , it mainly takes care of authorizations of user and role management and it is self explanatory.
4. Persistency to makse sure that data is consistent during all the times. As you know whole the data is In-Memory during power failures or system hault we need to make sure that consistent data is avialbale for every body.

This is a Source system from where you are getting the data and push them into HANA DB. In this,you can find Load Controller and Rep Agent which are Sybase Components (along with Rep Server which will sit in In Memory Computing Engine) which will be used for Data Provisioning from Source system to HANA DB.This is a Log based Data Provisionig
SAP Business Objects BI 4:
This is the place where the Data Services Server and Data Services Designer (Client version) installed. This also will be used for Data Provisioning from any source system to HANA DB. This is a Job based Data Provisioning, wherein you will create a job in Data Services Client and execute. This is not real time.
Here you will see SBO BI4 Information Design Tool (which will be used to create Universes on top of HANA Data) and it can be consumed in BO Tools like Dashboard, Crystal Reports,WebI, AO, Explorer, etc
SBO BI4 Servers are the one which stores the created reports in central place. BOE Server is kind of central repository
Other Source Systems:
Any systems (ERP/CRM/SAP BW or third party) which can be used as Source and data can be moved from them to HANA DB
In Memory Computing Studio:
This is the Eclipse based Tool with which the user can Create Views (Attribute,Analytic,Calc) on top of HANA Data. This tool can be used for Administration purpose also.
These are the supported reporting tools through which the user can create reports on top of HANA Data. There are different drivers available to connect to each reporting Tool.
In Memory Computing Engine:
This component resides under the Index Server of HANA DB
Session Management:
When the Client request for the data at runtime or when the user execute the report through reporting layer, Session Management will establish the connection to HANA database and maintain it for each different client.
Request Processing and Execution Control:
Once the session has been established, the SQL Parser will take a control and call the respective Engine accordingly. All the MDX queries will be processed by MDX Engine. There are OLAP Engine,Join Engine and Calc Engine to process different views.
Relational Engine:
This has Row and Column Store which stores the data in memory in Row and Columnar fashion
Persistence Layer: The persistence layer ensures that changes are durable and that the database can be restored to the most recent committed state after a restart.
Disk Storage:
This is the physical storage of Data and Log Volumes

To know in depth knowledge about these components,please visit