Exadata Architecture

Oracle Exatadata 

Features of exadata

• understanding exadata Performance
• Storage and compute node layout
• Configuring and Recovring Exadata

Exadata Architecture:

reference taken from nice and lovely blog written by 

(Nayak Vishwanath)

·    


                   LIBCELL

 LIBCELL, which is a library that is linked with the Oracle kernel. LIBCELL has the code that knows how to request data via iDB protocal.

CELLSRV

Cell Services (cellsrv) is the primary software that runs on the storage cells. It is a multi-threaded program that services I/O requests from a database server.

Management Server(MS)

·         MS is a Java program that provides the interface between cellsrv and the Cell Command Line Interface (cellcli) utility. MS also provides the interface between cellsrv and the Grid Control Exadata plug-in .

Restart Server(RS)

·         RS is actually a set of processes that is responsible for monitoring the other processes and restarting them if necessary

Input/Output Resource Manager(IORM)

·          IORM manages I/O at the storage cell by organizing incoming I/O requests into queues according the database name, category, or consumer group that initiated the request. It then services these queues according to the priority defined for them in the resource plan.

•Infiniband Switches 

 InfiniBand switches and cables to form a 40 Gb/second InfiniBand fabric for database server to Exadata Storage Server communication, and RAC internode communication.

•Cellinit.ora and Cellip.ora
After Oracle Exadata Storage Server is configured, the database server host must be configured with the cellinit.ora and the cellip.ora files to use the cell. 

The files are located in the /etc/oracle/cell/network-config directory of the database server host.These configuration files contain IP addresses, not host names.




cellinit.ora - This file contains the database IP addresses.

cellip.ora -    This file contains the storage cell IP addresses.

Example:

Quarter RAC Exadata machine contains 2 compute(DB) nodes  and 3 Cell(Storage) servers.Below picture shows the ip configurations. 

192.168.50.23 and 192.168.50.24 belongs to Compute Nodes.

192.168.51.27, 192.168.51.28 and 192.168.51.29 belongs to Cell Servers.

Startup Sequence of Exadata Storage Server:

/etc/init.d/cell.d  starts first when system starts

The exadata storage cells operate independently from each other. Cellsrv on each storage server processes I/O  requests via iDB message and independently transmits the return data/messages to the compute(DB) node. It is the responsibility of the compute server to sort, group and process the results returned from each storage server. 

The Management Server (MS) provides a java interface to the CellCLi command line interface, as well as providing an interface for Enterprise Manager(Grid Control) plugins. If MS is down you won't be able to use cellcli.

Process startup sequence  

•  - The cellrssrm process is the main Restart Server(RS) process. It launches three helper processes cellrsomt, cellrsbmt and cellrsmmt.
•  - The cellrsomt helper process is ultimately responsible for launching cellsrv.
•  - The cellrsbmt and cellrsmmt processes are additional helper processes responsible for launching cellrsbkm and the main Management Server(MS) Java process.
•  - Cellrssmt is called by cellrsbkm, and its ultimate goal is to ensure cell configuration files are valid, consistent and backed up.
•  - And of course, cellsrv is the main process that performs I/O on the exadata cell.

Exadata's Data Processing Layout :

Explained the data flow in below picture. Read the data point 1 to 9. 



Communication between DB and Cell server: