SAP HANA – Blog
SAP HANA Architecture
HARDWARE SIZING AND CAPACITY PLANNING FOR SAP HANA
Physical node (failover and failsafe)
# Of CUP
Memory Size –
Actual memory requirements in HANA strongly depend on operational conditions, which cannot be taken into account without analyzing the HANA system itself
Size your BW on SAP HANA System Determining needed hardware resources is usually done by the SAP hardware partner based on order information you submit.
This table offers rough estimates of system hardware needs.
|Source Data footprint||HANA Appliance RAM required||HANA Appliance Disk required|
|0.5 TB||315 GB||1.53 TB|
|1 TB||577 GB||2.81 TB|
|5 TB||2.56 TB||12.8 TB|
|10 TB||5.06 TB||25.31 TB|
You can also calculate your own HANA RAM needs using one of these formulas:
RAM = (Source Data footprint – 50 GB) * 2 / 4 + 90 GB
RAM = Column Store Tables footprint * 2 / 4 + Row Store Tables footprint / 1.5 + 50 GB
Though the HANA database uses in-memory technology, disk space is required to store persistently data used in memory. The disk space must be large enough to hold:
Before- and after-image versions of data (2xRAM)
Database log (1x)
Operational space (1x)
At least one process image (in case of software failure) (1x)
One data export (1x)
So, the required size of disk space is 5x the size of the data in memory (RAM).
SAP Note 1637145, SAP BW on HANA: Sizing SAP In-Memory Database.
SAP Note 1702409, HANA DB: Optimal number of scale out nodes for BW on HANA.
Physical Node -The administration console of the SAP HANA studio provides an all-in-one support environment for system monitoring, backup and recovery, and user provisioning.
In case of scenarios like data center failures due to fire, power outages, earthquake, and so on, or hardware failures, such as the failure of a node, SAP HANA supports the hot-standby concept using synchronous mirroring with a redundant data center concept, including redundant SAP HANA databases
Data Center 1 Switch over Data Center 2
To provide an idea about sizes and access speeds of a current memory hierarchy, the table below compares the different layers in this memory hierarchy (CPU characteristics for Intel’s Nehalem architecture
Type of Memory Size Latency
L1 cache 64 KB ~4 cycles [2 ns]
L2 cache 256 KB ~10 cycles [5 ns]
L3 cache (shared) 8 MB 35–40+ cycles [20 ns]
Main memory GBs upto terabytes 100–400 cycles
Solid-state memory GBs upto terabytes 5,000 cycles
Disk Up to petabytes 1,000,000 cycles
Hardware Architecture : current and past performance Bottlenecks
What is a Schema in HANA ?
It’s nothing but a collection of database objects , such as
http://www.CloneSkills.com , Info@CloneSkills.com , 800.836.5696 CloneSkills, Inc.
Procedures , etc…