AEM Clustering

Are you talking about Author clustering or Publish clustering ?? You can read [1] in detail to understand how it was done in 5.6.1 and prior version 

[1] https://docs.adobe.com/docs/en/cq/5-6-1/core/administering/cluster.html

Using repository.xml, we could have set up active/active or active/passive shared data stores in the previous versions where as from 6.x, OAK doesnt support the same and hence for publish clustering we need to go with mongo DB and for author only cold stand by is supported. You really cant have active/active clustering.

Hi,

I understand that in AEM 6.0, for achieving clustering, we need to necessarily use MongoDBMK and not TarMK.

Please confirm the following two points from my side.

1. AEM 6.0 Author: Only Cold Standby is supported. It is called Active-Passive Clustering right?

2. AEM 6.0 Publisher: Active-Passive clustering as well as Active/Active clustering are supported?

Appreciate your responses.

Thanks,

Rama.

nsvsrk wrote...

Hi,

I understand that in AEM 6.0, for achieving clustering, we need to necessarily use MongoDBMK and not TarMK.

Please confirm the following two points from my side.

1. AEM 6.0 Author: Only Cold Standby is supported. It is called Active-Passive Clustering right?

2. AEM 6.0 Publisher: Active-Passive clustering as well as Active/Active clustering are supported?

Hello Rama,

as I said in your other thread, Publish cluster is not supported. Please go over the documentation on recommended deployments[0] 

Cold standby if not active/passive. As AP implies the ability to immediately cutover to the passive i.e. both instances are up but only one is receiving traffic. With the cold standby, the standby instance is only partially running and it needs to be shutdown and restarted in order to make it active.

For the publish tier, you can have multiple publish instances in a TarMK farm, which provides the resilience required. Why do you feel you need clustering on the publish tier?

Regards,

Opkar

[0]https://docs.adobe.com/docs/en/aem/6-0/deploy/recommended-deploys.html

Hi Opkar and all,

Q: Why do you feel you need clustering on the publish tier?

I read the following in the URL:https://www.terrabeata.com/2015/01/clustering-in-aem-6/

"Clustering should not be used in most cases for publish instances. For publish servers, speed is very important. The TarMK repository is the quickest type at the moment. Publish servers can be scaled by adding individual servers to a farm in which each receives replications from the authoring environment. There are edge cases in which this rule does not hold. One example where a cluster may be needed for publish servers is where there is a great deal of user generated content. Another case for clustering is for sites with massive amounts of data."

In my case, my assumption is that there is a huge volume of UGC and it has to support growing and massive amounts of data.

As you suggested, for Publish, probably I should use MongoDBMK + MSRP.

Thanks,

Rama.

The AEM Communities team moved to a common store[0] approach precisely because it offered the best performance level combined with TarMK. I wouldn't recommend going down the route of MongoDB, if your use case is purely UGC. 

Also, the article you posted is quite an old article, at the time clustering was considered for the publish tier, but the advice has since been revised. I'd get in touch with the sales team from Adobe that you are in contact with to decide the best approach, as they can get in touch with engineering to advise you.

Is this a new project? As you seem to be referencing AEM6.0.

Regards,

Opkar

[0]https://docs.adobe.com/docs/en/aem/6-1/administer/communities/overview.html#Community Content Persistence

Hi Opkar and all,

It is not a project. I am creating a reference proposal to frame the basis for all upcoming ones for AEM is 6.1.

I will qualify my solution with all assumptions like huge UGC, growing volumes, etc.

It looks like a good proposition for my description is the following. Please offer your inputs.

     Publish: Multiple publish instances in a TarMK farm + ASRP 

     Author:   Multiple publish instances in a TarMK farm + ASRP,  as ASRP is common to Author and Publish

1. Prior to 6.1, we used to reverse replicate UGC to author and after careful moderation replicate to all other Publish instances.

    With 6.1, using SRP,  If the UGC is not moderated, how is the security ensured? Whatever one user enters will be seen by the whole world!

2. Is there any diagrammatic representation of Author and Publish with TarMK farm and ASRP, which I could use in my document?

Planning to complete my document today and so appreciate your quick response and patience.

Request to share your contact, so that I could have a 2 minute call with you and close this.

Thanks,

Rama.

Hi Opkar,

I am able to get some URLs on this content moderation.

Appreciate your help.

Thanks,

Rama.