Download as pdf or txt
Download as pdf or txt
You are on page 1of 46

Week 1: Planning an SAP HANA SR HA Deployment

Unit 1: Overview of the SAP HANA architecture


Overview of the SAP HANA architecture
Overview of SAP HANA

• SAP HANA is a database and a platform


• The SAP HANA database component stores the data in
memory
o This eliminates I/O bottlenecks
o Data is persisted to disk at regular intervals
• The data in memory is compressed
• Everything is calculated on-demand

open.sap.com Slide 2
Overview of the SAP HANA architecture
SAP HANA database terminology

• Host
o The hardware and operating environment on which the SAP HANA
database runs
− The host provides all the required resources and services such
as CPU, memory, network, and storage
• System
o A system is one or more instances with the same SAP system ID
and instance number
o The term system is interchangeable with the term SAP HANA
database
o An SAP HANA system with more than one instance is distributed
over several hosts

open.sap.com Slide 3
Overview of the SAP HANA architecture
SAP HANA database terminology

• Instance
o A set of SAP HANA system components installed on one host are
called an instance
• SAP HANA <SID>
o The SAP HANA <SID> is the name of the SAP HANA system
o The <SID> is assigned during the installation
o The <SID> is unique throughout an organization
o The <SID> consists of exactly three alphanumeric characters
− Only uppercase letters are allowed
− The first character must be a letter

open.sap.com Slide 4
Overview of the SAP HANA architecture
SAP HANA is a “multitenant database container system”

• A single SAP HANA system can host multiple


databases
• A database lives in what is referred to as a “container”
within the SAP HANA system
o The container enables complete isolation of a database
from any other databases in an SAP HANA system
• SAP HANA 2.0 SPS01 and above is installed as a
multiple database container system
• A minimal SAP HANA system contains:
o System database
− Only one system database is required for an SAP HANA
system
o One tenant database

open.sap.com Slide 5
Overview of the SAP HANA architecture
Identifying a database

• The SAP HANA system is identified using a system identifier


o SID
• An SAP HANA database is identified using two pieces of
information
o SID
o Database name

open.sap.com Slide 6
Overview of the SAP HANA architecture
SAP HANA deployment options

• SAP HANA can be deployed in either of the following


configurations:
o Scale Up
o Scale Out
• Scale Up
o When more capacity is required, add more hardware to the
existing host
o Limited by factors including the hosts CPU capacity, memory
capacity, and IO throughput
Scale Up
• Scale Out
o When more capacity is required, add more systems

Scale Out

open.sap.com Slide 7
Overview of the SAP HANA architecture
Information sources

• In this course we will focus on the deployment of SAP


HANA with system replication, performance optimized, in
a scale up configuration
• SUSE product and best practices documentation is
available from
o www.suse.com/documentation
• This training material is based on the SUSE Best Practice
Document:
o SAP HANA System Replication Scale-Up Performance
Optimized Scenario
− The latest version of this document is available from the
SUSE documentation web site
• The cluster nodes in the demonstrations are running
SLES for SAP Applications 15 SP2

open.sap.com Slide 8
Thank You!

Contact Information:
open@sap.com
Week 1: Planning an SAP HANA SR HA Deployment
Unit 2: Native SAP HANA HA features
Native SAP HANA HA features
Fault tolerance and high availability

• Fault tolerance aims to eliminate all single points of failure


o Eliminating all single point of failure can be very costly
• High availability
o Accept some downtime
− Reduces the implementation costs
o Aim to recover quickly
− For example, a 20 second delay during the recovery process
may be acceptable
• Implementing high availability for SAP HANA is a
combination of:
o SLE High Availability
− The HA extension is included with a SLES for SAP Applications
subscription
o SAP HANA's native fault tolerance features
o Fault tolerant hardware components

open.sap.com Slide 2
Native SAP HANA HA features
Assessing single points of failure

• Providing a completely fault-tolerant system is expensive


• The key to achieving fault tolerance is to eliminate all single
points of failure
• Assess everything:
o Power
o Hardware
o Networking
o Data centers
o .....
• Put fault tolerance in place where the cost vs return is
acceptable
o e.g. dual PSU in a server

open.sap.com Slide 3
Native SAP HANA HA features
Overview of the available SAP HANA native HA features

• SAP HANA includes several native high availability features


• These SAP HANA features are:
o Service auto-restart
o SAP HANA auto-restart
o Host auto-failover
− Only available when using a scale out configuration
o SAP HANA system replication
• These native SAP HANA features reduce the time to recover
from a failure
• A fully automated HA solution cannot be provided using only
the SAP HANA HA features
• To fully automate an HA solution for SAP HANA requires the
specialized components provided in SLES for SAP
Applications

open.sap.com Slide 4
Native SAP HANA HA features
Native SAP HANA HA features – service auto-restart

• Service auto-restart use case:


o Local software failure
• Local faults such as software failures can often be
handled using the same hardware
• A simple first step is to restart the failed service on the
same server
o Advantages:
− SAP HANA provides this function at no cost
− No loss of database integrity
o Disadvantage:
− Transactions not committed to the logs are lost
− May take some time to recover
• The SAP HANA auto-restart watchdog provides this
function
o This service is provided by the SAP HANA daemon
• In a clustered environment be aware failing SAP HANA
over to a node with SAP HANA active on it can lead to a
much faster recovery than restarting SAP HANA on the
same node

open.sap.com Slide 5
Native SAP HANA HA features
Native SAP HANA HA features – SAP HANA auto-restart

• When the system hosting SAP HANA is started, SAP HANA


will automatically startup
• Use cases:
o After a power failure
o Powering on the server after a shutdown
• When power returns after a power failure the SAP HANA
database system automatically performs a startup and
recovery
• Configuration required:
o SAP HANA database system is configured with auto-restart mode
on
• In a clustered environment SAP HANA is under cluster
control, therefore we do not want auto-restart enabled

open.sap.com Slide 6
Native SAP HANA HA features
Native SAP HANA HA features – system replication

• SAP HANA system replication – SAP HANA SR


• SR is provided with SAP HANA at no extra cost
• Consists of a primary and a secondary system
• The primary and secondary systems can be:
o In the same data center
o In remote data centers
• The secondary can be either passive or active
o Passive – Client requests are not accepted
o Active – Read-only client requests are accepted

open.sap.com Slide 7
Native SAP HANA HA features
Native SAP HANA HA features – system replication

• System replication modes:


o Synchronous in-memory (SYNCMEM)
− The secondary system sends an acknowledgment back to the
primary system as soon as the data is received in memory
o Synchronous (SYNC)
− The secondary system sends an acknowledgment back to the
primary system as soon as the data is received and persisted to
the log volumes on disk
o Asynchronous (ASYNC)
− The primary system sends redo log buffers to the secondary
system asynchronously. The primary does not wait for
confirmation from the secondary system
• The SYNC mode is used when configuring SAP HANA SR
scale up performance optimized on SLES for SAP
Applications with HA

open.sap.com Slide 8
Native SAP HANA HA features
Native SAP HANA HA features – operation modes

• There are three operation modes for SAP HANA system


replication:
o Delta Data Shipping
− System replication where delta data is transferred in addition to
continuous log transfer
− Default delta log transfer interval = 10 minutes
o Continuous Log Replay
− The transferred redo log is continuously replayed at the
secondary site
o Continuous Log Replay with Active/Active
− This is the same as the Continuous Log Replay option but read
only access for clients is permitted
• Continuous Log Replay is used when deploying SAP HANA SR
on SLES for SAP Applications with HA

open.sap.com Slide 9
Native SAP HANA HA features
SAP HANA system replication – how it works

• On the primary:
o SAP HANA makes updates in RAM on the
primary
o The updates are written to the
transaction log
o At the next savepoint, the data is
written to disk
• On the Secondary:
o When the primary makes updates in
RAM they are synchronously written to
the secondary
o The updates are written to the
transaction log
o At the next savepoint, the data is
written to the disk
• Result: The SAP HANA databases on
the primary and secondary are
synchronized

open.sap.com Slide 10
Thank You!

Contact Information:
open@sap.com
Week 1: Planning an SAP HANA SR HA Deployment
Unit 3: SAP HANA Deployment Options
SAP HANA deployment options
SAP HANA deployment scenarios

• This training is focused on deploying SAP HANA with


system replication, scale up in a performance-optimized
scenario
• This deployment scenario is not the only deployment
scenario, and in this unit we will make you aware of other
available scale up deployment scenarios
• Important: SUSE may have a supported HA solution for the
following deployment scenarios, but you must check the
SUSE Best Practices document for the particular
deployment scenario that suites your requirements

open.sap.com Slide 2
SAP HANA deployment options
SAP HANA SR with HA deployment configurations

• The following deployment configurations are available in a


scale up deployment:
o Performance-optimized
o Cost-optimized
o Multi-tier chain topology
o Multi-tier multiple-target topology

open.sap.com Slide 3
SAP HANA deployment options
Overview of the performance-optimized deployment scenario

• SAP HANA with system replication


• The secondary system is dedicated
• SAP HANA data is pre-loaded into memory on the secondary
• Advantages:
o Fully automated failover with SLE HA can be configured
o Fast recovery from the failure of the primary
− The data is already in memory on the secondary
• Disadvantages:
o Higher cost because secondary system is dedicated and is only
used when a failover occurs
• The performance-optimized deployment enables a fully
automated HA solution for SAP HANA to be deployed using
SLES for SAP Applications

open.sap.com Slide 4
SAP HANA deployment options
Overview of the cost-optimized deployment scenario

• SAP HANA with system replication


• The secondary system is not dedicated
• SAP HANA data is not pre-loaded into memory on the
secondary
o SAP HANA therefore does not require memory for loading the
data
− Data preload must be turned off
o SAP HANA memory size configuration is adapted for this scenario
• SAP HANA is running on the secondary system in an SR
configuration but only persists the logs and data to
persistent storage
• Other software can be run on the system, for example, an
Oracle development system

open.sap.com Slide 5
SAP HANA deployment options
Overview of the cost-optimized deployment scenario

• Advantage:
o Lower cost because the secondary system can be used for other
purposes, e.g. testing or QA
• Disadvantage:
o Slow recovery because running services need to be stopped, SAP
HANA started, and then there is a delay while the SAP HANA data
loads into memory

open.sap.com Slide 6
SAP HANA deployment options
The failover process – performance vs cost-optimized

• Performance-optimized
o The detection of a failure and the failover process can be fully
automated with SLE HA
• Cost-optimized
o The detection of a failure and the failover process is manual
therefore the failover process cannot be automated with SLE HA
o When failover is required:
− First the failover requirement must be detected, this is a
manual process
− Any non-SAP HANA services hosted on the secondary must be
stopped, for example, an Oracle development system
− The SAP HANA memory must be reconfigured to use all the
available memory
− The data must be loaded into memory; this can take an
extended period of time

open.sap.com Slide 7
SAP HANA deployment options
Multi-tier chain topology

• SAP HANA with synchronous system replication from the


primary to the secondary
• Add a third node with asynchronous system replication from
the secondary to the third node
• The SLE HA cluster only handles the primary and the first
secondary system

open.sap.com Slide 8
SAP HANA deployment options
Multi-tier multiple-target topology

• SAP HANA with synchronous system replication from the


primary to the secondary
• Add a third node with asynchronous system replication from
the primary to the third node
• The SLE HA cluster only handles the primary and the
secondary system that are using synchronous
replication between them for SR

open.sap.com Slide 9
Thank You!

Contact Information:
open@sap.com
Week 1: Planning an SAP HANA SR HA Deployment
Unit 4: Planning the Required Infrastructure
Planning the required infrastructure
Setting the context for the plan

• What are we aiming to achieve?


o Deploy SAP HANA scale up performance optimized with system
replication on SLES for SAP Applications with High Availability
• To support SAP HANA SR with HA there are some
prerequisites
• This training material follows the SUSE best practice
document:
o SAP HANA System Replication Scale-Up - Performance Optimized
Scenario

open.sap.com Slide 2
Planning the required infrastructure
Planning – cluster components

• The best practice document this training is based on


has been tested and certified on a two-node cluster
o Three nodes can be deployed but the third node is only used
as a decision maker for fencing
− For example, if using diskless SBD
o The third node must not run any SAP HANA resources
• A valid STONITH method must be defined
• Both nodes are on the same layer-two network
segment
• Name resolution is available locally to each cluster
node
• Time synchronization is configured and active
• The Linux users and group required by SAP HANA are
defined locally on each cluster node
o Each user must have the same UID on each node
o The group must have the same GID on each node

open.sap.com Slide 3
Planning the required infrastructure
Planning – SAP HANA components

• Cluster nodes hosted in geographically remote data centers


must match the requirements of SLE HA extension
recommendations
• Both primary and secondary SAP HANA instances must have
the same SID (SAP identifier)
• Configure AUTOMATED_REGISTER=<"true | false">
o If "true" the previously failed primary is automatically registered
as a secondary to the running primary
o If "false" the previously failed primary must be manually
registered as a secondary to the running primary
• SAP HANA auto-restart must be disabled
o The SAP HANA instance will be under cluster control
− SAP HANA must not be managed by systemd or sapstartsrv

open.sap.com Slide 4
Planning the required infrastructure
Multi-tenant SAP HANA databases and HA

• Multi-tenant databases are supported using HA


• The SAP HANA RDBMS is treated as a single system by HA
o Therefore, it includes all database containers in a single HA
solution
• With an MDC implementation cluster, failover decisions are
based on the complete RDBMS
o Individual containers are not assessed during the failover decision
process
• Designing test procedures for MDC systems with HA requires
careful planning

open.sap.com Slide 5
Planning the required infrastructure
Overview of sizing for SAP HANA

• Sizing SAP HANA is a complex process and requires the


skills of an experienced SAP consultant
• The key components to consider when sizing SAP
HANA include:
o Memory
o Storage
o CPU
o Network
• A good place to start is:
• https://www.sap.com/about/benchmark/sizing.quick-
sizer.html#quick-sizer

open.sap.com Slide 6
Planning the required infrastructure
SAP HANA network planning

• SAP HANA network requirements need to be assessed on a


case-by-case basis
• Always refer to SAP documentation for the latest best
practice configuration options
o https://www.sap.com/documents/2016/08/1cd2c2fb-807c-0010-
82c7-eda71af511fa.html
• Communication is over the following logical network zones:
o Client
o Internal
o Storage

open.sap.com Slide 7
Planning the required infrastructure
Logical network zones for SAP HANA

• The client zone includes:


o SQL clients running on SAP application servers
o Browser applications
− Using HTTP/S
o Data sources
− For example, BI
• The internal zone includes:
o The communication between hosts in a distributed SAP HANA
o system
− This would be a scale out scenario; this training is focused on
− a scale up scenario
o The communication used by SAP HANA system replication
• The storage zone:
o SAP HANA persists the data and transaction log data to persistent
storage
o If storage is on the SAN, the SAN infrastructure must be planned
o Storage can be local to an SAP HANA node

open.sap.com Slide 8
Thank You!

Contact Information:
open@sap.com
Week 1: Planning an SAP HANA SR HA Deployment
Unit 5: Deploying the required infrastructure
Deploying the required infrastructure
Demonstration – Deploy the Required Infrastructure

• The infrastructure has been deployed and in this


demonstration we will take a tour of that infrastructure
o This module contains useful reference material for later units
• The services systems networking, repository and hosts file
configuration will be viewed
• The storage systems network configuration and iscsi targets
will be viewed
• The networking configuration and storage
available on node01 will be viewed
• The networking configuration and storage available
on node02 will be viewed
• The network configuration on the nfs system will be viewed
o The SAP HANA installation archive file has been previously
downloaded and copied onto a file system on the nfs system

open.sap.com Slide 2
Deploying the required infrastructure
Demonstration Summary – Deploy the Required Infrastructure

• The infrastructure has been deployed and in this


demonstration we went on a tour of the infrastructure that
has been deployed
o This module will be a useful reference module if you need to
check the configuration during later units
• The networking, repository and hosts file configuration for
the services system was viewed
o The services system provides access to the repositories required
by the systems related to the cluster
• The storage system was checked
o There were many local disks available
o All the disks except the disk containing the operating system are
shared using iscsi

open.sap.com Slide 3
Deploying the required infrastructure
Demonstration Summary – Deploy the Required Infrastructure

• The storage systems network configuration was viewed and


was found to be on the following networks 192.168.17.0,
10.10.11.0, 10.10.12.0 and 10.10.13.0
• The iscsi devices were then listed

o There are 6 devices on the 192.168.17.0 network and the other 3


devices are each on their own 10.10.x.0 network

open.sap.com Slide 4
Deploying the required infrastructure
Demonstration Summary – Deploy the Required Infrastructure

• The networking configuration on node01 was checked


• The locally available storage was viewed
o Some of the devices are iscsi targets
o Some of the iscsi targets are shared with node02 and some are
only local to node01
• The networking configuration on node02 was checked
• The locally available storage was viewed
o Some of the devices are iscsi targets
o Some of the iscsi targets are shared with node01 and some are
only local to node02
• The network configuration on the nfs system was viewed
• The SAP HANA installation archive file has been previously
download and copied onto a file system on the nfs system
• The status of the nfs service was checked and found to be
not running
o The nfs service will be configured in a later unit
open.sap.com Slide 5
Deploying the required infrastructure
Demonstration System Networking Reference

open.sap.com Slide 6
Thank You!

Contact Information:
open@sap.com

You might also like