SAP HANA on FSx for NetApp ONTAP Estimator Rules
Scope of the Estimator
The tool is used to size the FSx for ONTAP file systems for an SAP
landscape. The sizing is derived by capacity and performance
requirements of the SAP systems. The SAP landscape can consist of
HANA production, non-production systems and SAP application servers.
The sizing includes the backup solution for the HANA databases and
all other SAP file systems. The backup solution is using Snapshot
backup operations combined with block-delta replications to a
dedicated FSx for ONTAP filesystem used as backup target.
The estimator calculations consider FSx for ONTAP features, like
storage efficiency and capacity pool tiering to provide the most
cost-efficient solution. The estimator provides capacity and
performance calculations for all FSx for ONTAP file systems which
have been defined in a reference architecture for SAP landscapes on
FSx for ONTAP.
Reference Architecture
The reference architecture consists of the following FSx for ONTAP
file systems.
-
One or multiple FSx for ONTAP Single-AZ file systems in AZ1 and/or
AZ2. The number of required file systems is determined by the
total performance requirements of the HANA systems which are
placed in AZ1 or AZ2. The estimator also calculates the total
required capacity in each availability zone.
-
One FSx for ONTAP Multi-AZ file systems which is used for log
backups of the HANA systems and for SAP application server in a
high availability configuration. The Multi-AZ file system is
spread across AZ1 and AZ2 and provides high availability in case
of a zonal failure of AZ1 or AZ2.
-
One FSx for ONTAP Single-AZ file system, which is used as a backup
target for all HANA systems and SAP file systems. The backup FSx
for ONTAP file system is placed in a third availability zone.
The figure below illustrates the distribution of the different FSx
for ONTAP file systems to the different availability zones.
Rules for placement of SAP and SAP HANA volumes
Each HANA system consists of a data volume, a log volume, a shared
volume and a log backup volume. Data, log, and shared volumes are
placed on a primary FSx for ONTAP system, either in AZ1 or AZ2
depending on the user selection. The example landscape in the figure
below shows a HANA production system using HANA System Replication.
In this case the HANA primary node would be placed in AZ1 and the
secondary node in AZ2.
HANA log backup is always stored on the Multi-AZ file system, which
ensures that log backups are available in case of a zonal failure.
Snapshot backup operations are considered for HANA data and shared
volumes. These Snapshot backups are replicated to the backup FSx for
ONTAP file system using SnapVault.
SAP application server DI (dialog instances) are placed on the
primary FSx for ONTAP system. SAP application server ASCS (central
services) are placed on the multi-AZ FSx for ONTAP file system
assuming that they are configured with zonal high availability. All
SAP application server filesystems are backed-up using Snapshots and
SnapVault replication as well.
Capacity Sizing
Generic capacity sizing rules
Storage efficiency savings are calculated for HANA shared and SAP
app server binaries only.
Snapshot backups are calculated for HANA data, HANA shared, and SAP
app server volumes.
Tiering to FSx for ONTAP capacity pool:
-
Tiering policy "All“ for backups at the backup FSx for ONTAP file
system.
-
Log backups are tiered depending on the number of snapshots kept
at the primary file system. If Snapshots are kept e.g. for 3 days
the tiering policy should be configured with three days as well to
ensure that local snapshots can be recovered with log backups from
the SSD tier.
Block change rate per day is individual to each HANA system. Typical
values are:
- ~30% for production
- ~5-10% for non-production HANA systems
-
1-3% for binary volumes (HANA shared, app server).
Capacity sizing parameters
The table below lists all parameters which are used for the capacity
calculation. The first section includes the input parameters which
are individual to each HANA system or SAP application server. The
other two sections are global parameters, which are used for all
systems.
Primary FSx for ONTAP file system(s) capacity calculation
Data, log and shared volumes per HANA system
The calculation is based on SAP best practices recommendations.
- Data volume: 1 x RAM
- Log volume: IF(RAM <= 512; RAM/2; 512)
-
Shared volume: MIN(RAM; 1024) x
(100-STORAGE-EFFICIENCY-SAVINGS-BINARIES) / 100
SAP application server binaries volume
FILE-SYSTEM-SIZE-APP-SERVER x
(100-STORAGE-EFFICIENCY-SAVINGS-BINARIES) / 100
Snapshot backups
Snapshot backup capacities are calculated with the formula:
-
FULL-BACKUP x DAILY-SNAPSHOTS-KEPT x CHANGE-RATE
HANA data volume Snapshots are calculated with :
-
RAM/2 x SNAPSHOTS-KEPT-AT-PRIMARY x CHANGE-RATE/100
HANA shared volume Snapshots are calculated with :
-
MIN(RAM;1024) x (100-STORAGE-EFFICIENCY-SAVINGS-BINARIES)/100 x
SNAPSHOTS-KEPT-AT-PRIMARY x CHANGE-RATE-BINARIES/100
SAP application server binaries Snapshots are calculated with :
-
FILE-SYSTEM-SIZE-APP-SERVER x
(100-STORAGE-EFFICIENCY-SAVINGS-BINARIES)/100 x
SNAPSHOTS-KEPT-AT-PRIMARY x CHANGE-RATE-BINARIES/100
Backup FSx file system capacity calculations
All backup capacities are calculated with the formula :
- Initial transfer + N x delta transfer
-
(full backup) + (daily backups kept) x (full backup) x (block
change rate)
Backup HANA data volume capacity is calculated with :
- RAM/2 + (RETENTION x RAM/2 x CHANGE-RATE/100)
Backup HANA shared volume capacity is calculated with :
-
MIN(RAM;1024) x (100-STORAGE-EFFICIENCY-SAVINGS-BINARIES)/100 +
RETENTION x MIN(RAM;1024) x
(100-STORAGE-EFFICIENCY-SAVINGS-BINARIES)/100 x
CHANGE-RATE-BINARIES/100
Backup SAP app server binaries capacity is calculated with :
-
FILE-SYSTEM-SIZE-APP-SERVER x
(100-STORAGE-EFFICIENCY-SAVINGS-BINARIES)/100 + RETENTION x
FILE-SYSTEM-SIZE-APP-SERVER x
(100-STORAGE-EFFICIENCY-SAVINGS-BINARIES)/100 x
CHANGE-RATE-BINARIES/100
Total backup capacity is calculated with :
-
SSD tier : (Backup HANA + Backup Share + Backup Binaries) x
MIN-ON-SDD/100
-
Capacity pool tier : (Backup HANA + Backup Share + Backup
Binaries) x (1-MIN-ON-SDD)/100
Multi-AZ file system capacity calculations
Log backups capacity is calculated with :
-
SSD tier: (RAM/2 x LOG-BACKUP-PERCENTAGE/100 x RETENTION) x
SNAPSHOTS-KEPT-AT-PRIMAY
-
Capacity pool tier: (RAM/2 x LOG-BACKUP-PERCENTAGE/100 x
RETENTION) x (RETENTION - SNAPSHOTS-KEPT-AT-PRIMARY)
SAP app server binaries capacity is calculated with :
-
FILE-SYSTEM-SIZE-APP-SERVER x
(100-STORAGE-EFFICIENCY-SAVINGS-BINARIES)/100
Performance Sizing
Sizing rules for SAP HANA production systems
Performance sizing for SAP HANA and FSx for ONTAP is based on
throughput capacity requirements. If an existing HANA system is
migrated to FSx for ONTAP, throughput capacity should be determined
from the existing HANA system.
For new HANA systems, throughput capacity can be determined based on
KPI‘s defined by SAP
- KPI for writes: 250 MB/s per HANA node
- KPI for reads: 400 MB/s per HANA node
File systems can be shared by multiple HANA systems as long as the
total performance requirements are met. HANA systems with high read
or write throughput requirements may require higher throughput
capacities or even multiple file systems to scale beyond the
throughput capacity of a single file system.
-
High read throughput requirements. Example: Fast data loading at
HANA startup.
-
High write throughput requirements. Example: Very large systems
with a very high volume of data loads or changes.
Sizing rules for SAP HANA non-production systems
Non-production HANA systems do not require the SAP performance KPIs
and are sized with lower values. Typical values are 30-50% for
test/QA systems and 10-20% for development/sandbox systems, 25% in
average.
FSx for ONTAP quality of service (QoS) can be used to limit
throughput for non-production systems to guarantee performance for
production systems running on a shared FSx file system.
FSx for ONTAP performance
For details on the availability of FSx for ONTAP throughput capacity
options in the different AWS regions, please check
Amazon FSx for NetApp ONTAP performance - FSx for ONTAP
Using the Estimator
The following chapters show the different screens and operations of
the estimator.
HANA system input screen
HANA system input screen
SAP landscape example
Capacity sizing results
Performance sizing results
Sizing summary and pricing
Feature set and known limitations
Feature set
-
Sizing of SAP HANA systems, SAP application servers and non-SAP
file systems (e.g. transport or interface directory)
-
Capacity sizing for primary and Multi-AZ FSx for ONTAP file
systems according to reference architecture
-
Backup solution sizing including local Snapshot and backup
replication to FSx for ONTAP backup file system
-
Capacity calculation includes tiering to capacity pool tier and
storage efficiency for non-HANA volumes
-
Performance sizing for HANA data and log at primary FSx for ONTAP
file systems
- Cost calculation depending on regin select
- Import and export of SAP landscape data
Known limitations
-
The non-availability of specific FSx for ONTAP throughput capacity
options in specific regions is not considered
-
Cost calculation is done depending on region selection, but is
based on price information from December 2023
-
Costs for cross-zonal SnapVault replication is not part of the
calculation
About the Authors
Concept :
-
Nils Bauer, SAP Competence Center Manager,
Solutions Engineering, NetApp
Contributors :
-
Niyaz Mohamed (NiMo), Principal Architect, Cloud
& Hybrid Cloud Solutions, NetApp
-
Pradeep Kumar, Technical Marketing Engineer and
Automation Specialist, Cloud & Hybrid Cloud Solutions, NetApp
-
Vedant Sethia, Technical Marketing Engineer and
Automation Specialist, Cloud & Hybrid Cloud Solutions, NetApp