The base layer of SAP HANA provides two file I/O interfaces:
You can configure the SAP HANA file I/O layer with configuration parameters to optimize file I/O for a given storage array and file system. The Linux XFS file system is used on all Dell EMC storage LUNs for the HANA persistence.
After the SAP HANA persistence is installed on PowerMax LUNs, set the following file I/O layer parameters for optimal I/O processing:
After the initial SAP HANA installation is complete, set the parameters by running the SAP HANA hdbparam command as <sid>adm in the Linux shell:
su - <sid>adm
hdbparam –p # lists current parameter setting
hdbparam –-paramset fileio [DATA].max_parallel_io_requests=256
hdbparam –-paramset fileio [LOG].max_parallel_io_requests=256
hdbparam –-paramset fileio [DATA].async_read_submit=on
hdbparam –-paramset fileio [LOG].async_read_submit=on
hdbparam –-paramset fileio [DATA].async_write_submit_blocks=all
hdbparam –-paramset fileio [LOG].async_write_submit_blocks=all
Starting with SAP HANA 2.0, the hdbparam command-line tool was deprecated. Instead, the parameters are defined in global.ini > [fileio].
Using the SQL console in SAP HANA studio, set the max_parallel_io_requests=256 parameter in the global.ini file by running the following commands:
ALTER SYSTEM ALTER CONFIGURATION ('global.ini', 'SYSTEM') SET ('fileio','max_parallel_io_requests[DATA]') = '256';
ALTER SYSTEM ALTER CONFIGURATION ('global.ini', 'SYSTEM') SET ('fileio','max_parallel_io_requests[LOG]') = '256' WITH RECONFIGURE;
The following figure shows what the fileio section of global.ini looks after the parameters are set:
Figure 5. fileio section of the global.ini file in SAP HANA studio
Both async_read_submit=on and async_write_submit_blocks=all are set by default during the installation. For more information, see SAP Note 2399079: Elimination of hdbparam in SAP HANA 2 (access requires an SAP username and password).
SAP HANA NSE uses the data volume that the main database is using. While no special configuration steps are necessary for the NSE persistence layer, sizing must take into account the additional capacity for the feature.
When you use the SAP HANA NSE feature, a portion of DRAM is used as a buffer cache to dynamically load paged data from the persistence (data volume). By default, the size of this buffer cache is 10 percent of the total main memory of the system. For more information about the SAP HANA buffer cache, see the SAP HANA Administration Guide.
SAP recommends not exceeding a ratio of 1:8 for the buffer cache and the total amount of warm data handled by SAP HANA NSE. When using SAP HANA NSE, you can store data in the warm tier in the following granularities:
Data location handling is built into the SAP HANA Data Definition Language (DDL). Manage the configuration by using the SAP HANA CLI SQL client hdbsql or the SQL editor in SAP HANA studio or SAP HANA cockpit.
To create a table using SAP HANA NSE (the warm tier), run the following DDL command:
CREATE T_DELL_NSE (id INT, name VARCHAR(100)) PAGE LOADABLE;
To create a column in the warm tier, run:
CREATE T_DELL_NSE (id INT PAGE LOADABLE, name VARCHAR(100));
To move an entire existing table to the warm tier, run:
ALTER TABLE T_DELL_NSE PAGE LOADABLE CASCADE
To move a column to the warm tier, run:
ALTER TABLET_DELL_NSE ALTER (name VARCHAR(100) PAGE LOADABLE);
SAP HANA NSE Advisor, which is based on real-time statistics from an existing SAP HANA database, provides recommendations for which data to move from the hot (in-memory) tier to the warm tier (SAP HANA NSE). Use the SAP HANA NSE Advisor information for guidance regarding the amount of data to move to the warm tier on an existing system.
For more information about SAP HANA NSE, SAP HANA NSE data sizing, and related topics, go to the SAP Help Portal.