AutoSys Workload Automation

Here is the HA configuration on AWS EC2 instance.  Please review and let me know if any changes or suggestion.

All EC2 instances with Win 2012 R2

Primary - EC2 instance (8vCPU,32GB) - EEM1, WCC1, AE1 

Shadow - EC2 instance (8vCPU,32GB)  - EEM2, AE2 

Tie-breaker - EC2 instance (8vCPU,32GB) - WCC2, AE3 

Meta-Store - EC2 instances with SQL-Server 2016 Standard Edition - 2vCPU,8GB

    --> AEDB1 

    --> WCCDB, AEDB2

Version details

AE: r11.3.6 SP6 (in HA with a tie-breaker)
WCC: r11.4 SP5
EEM: r12.51 CR04

Hello

Here are a few useful links to help you to design the architecture:

AWS ports

WAAE Best Practices

Security Best Practices with EEM

CA WCC Best Practices

In addition, to balance the load on the machines, I would put EEM1 and WCC1 on the Tie-breaker machine and EEM2 with WCC2 on the Shadow machine.

The tie-breaker scheduler is just going to update its heartbeats in the database and check if the two other schedulers are active. Therefore it won't be very busy

Same for the Shadow scheduler.

Let's see what the others think about it as well

Regards

Jean Paul

Hi Jean Paul - Please review the revised configuration and provide your suggestion.  Thanks Soundar.  

Hi Sender ,

Just to inform you .

Database platforms such as SQL Server and others are not currently supported with AWS but will be supported with the AE 11.3.6 SP7 maintenance. 

The support of WAAE with AWS is under certification for 11.3.6 SP7.

Best regards,

Faouzia

AWS Linux instances much cheaper...  - just sayin.  

Not really EC2-specific, but I would not recommend installing EEM and WCC on the same host as AE.  The combination of AE, EEM, WCC, Command Sponsor, Common Components, PMUX, PKI etc all on the same host can sometimes tie you in a knot, for example when upgrading one or more things in place. 

I'd also suggest 32GB RAM for AE servers is overkill, escpecially as per Jean-Paul, for the TieBreaker.  Obviously its not stated how many AE instences are being run, but if its multiple ones then I'd spread Primary and Shadow Schedulers across hosts.  Other folks may see this differently.  Good AE performance still depends greatly on fast databse I/O, so I'd be inclined to increase the spec of the SQL Server machine a little (although SQL Server is usually plenty fast), consider scaling back on the spec of the AE servers, and have 2 separate servers hosting EEM and WCC.

Hi Antony - Please review the revised configuration and provide your suggestion. Thanks Soundar

As per previous post, I would not install AE, EEM and WCC together on the same hosts.  A few reasons: 

• Possible software incompatibilities with everything together.
• In place upgrades are difficult or impossible, especially if they require a reboot.
• Suppose your Primary Scheduler fails and the Shadow takes over.  Its now competing for resources with WCC and EEM.  Your Shadow Scheduler machine should be same as the Primary and deliver the same performance in the event of a Scheduler failover.

Why not have:

• AEP and AES on medium size hosts
• AET on a small host
• EEM1 and EEM22 on medium hosts
• WCC1 and WCC2 on large hosts

Also - would not recommend installing on the D drive.  You will finish up with some stuff on C no matter what you choose.

Hi Antony, 

Thank you once again, can you clarify on the small medium and large hosts. 

Is the below one looks good?

Zone1 - EC2 (4vCPU,16GB) - AEP, EEMP

Zone2 - EC2 (4vCPU,16GB) - AES, EEMS

Zone 2 - EC2 (4vCPU,8GB)  - AET 

Zone 1 - EC2 (8vCPU,32GB) - WCCP

Zone 2 - EC2 (8vCPU,32GB) - WCCS

Meta-Store - EC2 instances with SQL-Server 2016 Standard Edition - 2vCPU,8GB

Zone 1 - AEDB instance (primary)

Zone 2 - AEDBS instance (Shadow), WCCDB

Did you face any issue installing in the non-system drive (D:\)? 

Thanks

Soundar

Hi Jean Paul / Antony / Faouzia / Linda

Thank you all for the suggestion.

Here is some additional information which I missed at first place: 

1. There will be only one instance and EEM policies are created to separate the tenants based on the job names.

2. The availability zone also to be mentioned in the EC2 machine, will mention as Zone1, Zone2, etc.,

3. The primary, shadow and tie-breaker will be updated with suffix like AE1 to AEP (primary) AES (Shadow) AET (Tie-br)

4. The installation will be on the non-system drive (D:) like D:\CA\ 

Here is the updated configuration:

Zone1 - EC2 instance (8vCPU,32GB) - AEP

Zone 2 - EC2 instance (8vCPU,32GB)  - EEMS, WCCS, AES 

Zone 1 - EC2 instance (8vCPU,32GB) - EEMP, WCCP, AET 

Meta-Store - EC2 instances with SQL-Server 2016 Standard Edition - 2vCPU,8GB

Zone 1 - AEDB instance (primary)

Zone 2 - AEDBS instance (Shadow), WCCDB

Please review the above and let me know if any changes.  

Regarding the Faouzia comments - 

CA Workload Automation AE Release 11.3.6 and later and CA WCC Release 11.4 and later (including the agents) are supported on the cloud with the following database limitation:
➢ Amazon RDS and Microsoft Azure PaaS databases are not supported.

I assume the EC2 instance with SQL-Server will be supported for AE 11.3.6. SP6, please confirm.

Regarding Linda comments - Thanks for the suggestion, currently the plan to use AE windows environment.

Thanks

Soundar

The instance name needs to be the same for Primary, Shadow and Tie Breaker. Not sure if you are implying you’ll setup 3 AutOSys instances in HA or if you are thinking you can have different instance names for the Shadow and Tie-Breaker.

Hi daddepalli - I know the instance name should be same.  Thanks for the update. 

Do you see any comments on the above AWS configuration?

I cant comment on the AWS/EC2 configuration as I am no expert on that side, but like Anthony suggested, break out WCC, AutoSys and EEM on separate servers.