QEDD HA Concepts

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 32

Assure QuickEDD

HA
High Availability for IBM i
Precisely - Use only
July 2020
Introduction
Precisely - Use only

2
Business continuity in any situation
• Disaster (RPO/RTO)
• Human mistakes
• Hardware issues, Maintenance of the systems and OS
• Any read operations on target: Saves, Batches…
• Application upgrades
• Systems migration

Precisely - Use only

3
Designed for a successful switchover or failover
• Target is always available!

• Easy and fast to install (around 10 minutes per pair), upgrade, use, supervise
• Up to 2 versions of OS level difference between source and target (possible to go from V7R1 to V7R3 for instance)
• Transactions are applied in sequence
• Real-time replication with acknowledgements

Precisely - Use only


• Synchronization / Verification
• Target writes can be suspended for the whole flow, data and objects
• Jobs follow-up for failovers

4
QuickEDD at a glance
AUDIT Send
(QAUDJRN
DB )
Journal

Detect /
Journal Apply
Select
IFS

Acknowl.

Send the events IN SEQUENCE Real-time replication


thanks to one single virtual local
journal on the production box

Precisely - Use only


ANALYZE

DETECT
SRS

REPAIR
Target
Storage
of journal
Start first replication with full EDC, with no lock on production box, and any status of data on target
entries

Automated errors management


5 Audits / IOA can be performed any time
JFP – jobs follow-up
• Failovers management
• Which jobs were active on production box when it stopped
• Which jobs were in JOBQ
• Submit again in same JOBQ on target, with equal settings

Precisely - Use only


JOB2
JOB1

JOB3
BATCH2
BATCH1

BATCH3
USRJOB2
USRJOB1

USRJOB3

NIGHT1 QBATCH QPGMR

6
Replication scope
APPLICATIONS SYSTEM
- Database
JOBS - Profiles
FOLLOW UP - *AUTL
- Objects
- IFS - Configuration
- Sysval
SPECIAL SPECIAL
- Triggers - SQL SPECIAL

Precisely - Use only


- Job scheduler
- Constraints - RCAC
- MQ
- Blobs - Temporal Tables
SPOOLS - WAS
- Commit ctl - etc.
- NetServer

OS and microcode are not replicated and under client’s responsibility regarding their upgrades.
7
Concepts - general
Precisely - Use only

8
Concepts
• Settings
• Settings centralized on SOURCE system
• The environment contains everything needed for the settings and for the replications follow-up

• Replication
• SIP, Sequence Integrity Processing: unique engine to process all journal entries
• CCI, Core Cycle Information: synchronous replication loop, with real-time acknowledgements
• Automatic journaling management

Precisely - Use only

9
Concepts
• Controls and Synchronization
• EDC : Synchronization by comparison; only differences are sent to the target system
• Linked to replication acknowledgements
• Activated by IOA: Independent Object Audit

• Smart Remote Staging (SRS)


• Pause of the replications, thanks to temporary storage of the journal entries on target system: continuous data protection, for
example during backup made on the target system, instead of the production one

• Jobs Follow-Up (JFP) Precisely - Use only


• Replication of jobs information (status, SBMJOB details), mandatory in case of fail-over

10
Concepts
• Tools
• External audit, control and supervision tools
• Sources of programs are available to adapt to different systems

• Supervision UI
• Light client to monitor at a glance Quick-EDD/HA on all your systems and partitions
• Configure and modify replication

• Switch
Precisely - Use only
• Switch scenarios allow a simple and efficient adaptation to all sites and contexts
• Management of Switchover / Fail-over / Return on production system
• Systems migrations

11
Concepts - More details
Precisely - Use only

12
Settings
• The « Environment » Object
• A single entity to describe the entire scope
• Settings manage the 6 system domains
• User libraries (including QGPL, QUSRSYS, …)
• System objects (Profiles, configuration …)
• IFS, including QDLS
• Spools files
• System values

Precisely - Use only


• Jobs

• Following your selections, Quick-EDD/HA will automatically analyze your applications in order to build the objects
list and the list of journals which are necessary for the replication
• The environment is automatically replicated on the target system in order to access the information in case of
switch
• Settings are done only on SOURCE system

13
Local Journaling on Production Box
• Any journaling organization is supported
• Unlimited number of journals
• Quick-EDD/HA can manage journaling start for applications that are not already journaled
• Support of all journaling functions
• MINENTDTA
• Journal Cache and *STANDBY mode (SS1 Option 42)
• Before image – Commit/Roll-back

Precisely - Use only


• Receivers management can be entirely done by Quick-EDD/HA - for different external needs, receivers can be
totally or partially managed by other processes

• Not wished » journals entries can be ignored by the reading process to avoid communications overload

14
Replication
• A unique engine to guarantee the sequence: SIP
• All the journals are simultaneously handled to guarantee the operations sequence
• Audit journal events are integrated in the same process

• Synchronous communications: CCI


• Source system is informed in real time
• All the events are acknowledged in real time

Precisely - Use only

15
Replication: SIP

Simultaneously
reading all the journals
in the scope


Rebuild of the
operations sequence
Precisely - Use only 
before sending DB / IFS / Audit
events are all
integrated at the same
level

16
Replication: CCI
PRODUCTION RECOVERY
Send

Detect /
Apply
Select

Precisely - Use only


Acknowledge

Core Cycle Information (CCI)


 Detection from the journal  Direct application on TARGET object
 Sending on the target system  Real-time acknowledgement

SOURCE system is always informed about the replication status


17
Synchronization: EDC
• Synchronization by comparison; only differences are sent to the target system
• Used at the first start of the environment
• When an object gets created
• Linked to replication acknowledgements to repair target objects
• Can be also activated by IOA: Independent Object Audit

ANALYZE

Precisely - Use only DETECT

REPAIR

18
Replication ≠ synchronization
• An acknowledgement validates any replication operation
• Any replicated event is acknowledged in real time
• The SOURCE system manages the whole process

• In case of negative acknowledgement, AUTOMATIC EXECUTION of the object synchronization is made for an
immediate repair of the abnormality.

• Synchronization function
Precisely - Use only
• EDC module allows to compare two objects to detect the differences. Only these differences are sent on the network
• The synchronization does not use the SAV / RST (except for the envelop) function to avoid any locking of object on the source
system
• Once synchronized, the object comes back in the replication loop

19
IOA: independent object audit
• IOA function monitors the TARGET system
• Scheduled or on demand control of all or a part of the replication scope

• Check of the objects presence and correspondence (for the 5 domains of the system)

• Dynamic repair of the differences


• Each difference detected by IOA function triggers immediately a repair; either through fixing the properties or launching a

Precisely - Use only


synchronization.
• No manual action is required.
• NB: If only attributes differ between source and target systems, the objects is fixed, there is no need to synchronize

• A spool report is created after each execution

20
IOA -> VWA: verify while active
• VWA allows to launch a full data check even if the database is active all the time.
• VWA can be launched whenever you want on the database to check all the records, even if there is activity on the
files.
• VWA runs in several parallel processes, in order to simultaneously control different parts of a file
• VWA controls as many records as possible
• VWA stops in case of conflict with the application (update, new insert not yet applied,...)
• Current position is memorized allowing to restart the control later from this position

Precisely - Use only


• At the end, a report displays how many records have been checked (there are messages in the environment, a
dedicated display and a spool file report).

• This control can be launched several times until 99% of the records have been checked
• (1% missing linked to the current activity, will be checked during the next execution )

21
SRS: smart remote staging
• Used either for:
• Sending in real-time what’s been read and filtered from the source, waiting that the slower target system can apply the events;
automated behavior

• Perform remote backups while production is still up and running:


• Target Backup gets requested through a journal entry on production box
• When processed on target that journal entry launches
• End of all writing activity from QuickEDD
• *NOIO mode: Events are sent from production to target, then stored

Precisely - Use only


• When the backup completes, Events are destaged, then replication gets back to its CCI loop

22
Tools – additional functions

• Control functions
• Objects, data, replication scope control
• Supervision of the replication process

• Monitoring functions
• Interception of all severe messages (gravity > 30)
• Sending by e-mail; SNMP; MSGQ – JOBLOG is attached

Precisely - Use only


• Specific objects saves
• Scheduler, NetServer, TCP/IP, RDB entries, …
• WRKDIRE, Distribution list, …

• For a total adaptation, all tools are delivered with source files

23
Tools, example: specific objects save
System data

Precisely - Use only

Network data

24
Tools, example: email monitoring

Server address and port number

Sender

Up to 4 email recipients
Precisely - Use only
Authentication if required

25
UI : setup, supervision, switch
Active? Latency?

Precisely - Use only In Sync?


Monitor all
servers/Lpar’s on a
single screen

Auto refresh and access to


all environments
26
Switch
• Quick-EDD/HA switch
• Only one function to perform the Quick-EDD/HA role-swap, for the switch and for the return on production system.
• Transparent management of audit, journaling, triggers, constraints and identity fields
• In case of fail-over: Automated management of the « commit/rollback » transactions, jobs follow-up, stored journal entries

• Adapted tools

Precisely - Use only


• Switch scenarios allowing to adapt to all contexts
• Switchover and switch-back of specific objects

27
Syncsort Confidential - Do Not Copy or Distribute
Switch
• PMEDHSWI command
• Environment role-swap RCV  SND to prepare replication from former target to former production system
• Creation of new journals receivers – positioning in the environment to give the replication starting point
• Audit and journaling control – start if necessary
• Triggers and constraints activation
• Identity fields management

Precisely - Use only

28
Switch Scenarios
Each step is defined in an
independent scenario

Scenario can be executed in batch,


interactive or step by step mode

Precisely - Use only


Free commands script for a better
adaptation to all sites

29
Additional use cases
• Systems migration : Migration Kit (EDH_MIGKIT); Daisy chains; One source to Several targets; Several sources to
One target; etc.
• Assure QuickEDD Promoter
• Assure QuickEDD Anonymizer
• Assure QuickEDD limited to Data replication

Precisely - Use only

30
Questions
Question True False
1 Microcode is replicated
2 Quick-EDD/HA uses local journaling
3 Settings are done only on the source system
4 The option MINENTDTA of journaling is supported

5 A dedicated job handles QAUDJRN

6 The synchronization of the PF doesn’t use SAV/RST for data

7 Precisely - Use only


IOA can trigger some synchronizations

8 Replication must be stopped during backup performed on the target system

9 The command PMEDHSWI must be used on the source system

31
Thank you
Precisely - Use only

You might also like