Expand my Community achievements bar.

Announcing the launch of new sub-community for Campaign Web UI to cater specifically to the needs of Campaign Web UI users!
SOLVED

Two tracking servers?

Avatar

Level 3

Hi guys,

We have a big client where we will use two different trackings servers (for URL & availability/redundancy issues).

My question is, can I use two different tracking servers for the same marketing instance?

As I know we can only set up one tracking server per marketing instance? not?

Should I create two different marketing instance and point each one for each one of the tracking servers?

Have anyone design any architecture like that before?

Best,

-Rick

1 Accepted Solution

Avatar

Correct answer by
Community Advisor

If you opt for two parallel marketing instances, I think you also need a loadbalancer in place..

You can do something more tricky and instead of having two marketing instances in parallel, you opt to configure a failover instance. I haven't done it before, but I would start by  creating an active and passive instance. On the passive instance, all services shutdown, and create a shell/powershell script on a cron job to monitor the active instance through various means (i.e - /nl/jsp/ping.jsp?) alerting, etc. and upon failure or connectivity error then activate the passive instance whilst continuing to monitor when the default instance is back online to failover to it again.

You can also opt make your default instance really robust, in my case I developed a few scripts scheduled on a cron job to monitor my services and perform troubleshooting automation, you can try out what works best for you and contact Adobe support to find out how they design/configure failover.

Amit has given a good explanation on a related subject.

Multiple WfServers Marketing servers for Hybrid Model

And more information on enterprise deployment.

Enterprise deployment

Amit_Kumar​ perhaps can advise henrique on this subject.

View solution in original post

5 Replies

Avatar

Community Advisor

Yes you can have two tracking servers per instance, you would need architecture knowledge in order to set this up, also a network administrator to open up connectivity between the two instances as you also need to setup a load balancer on top of them.

Avatar

Level 3

David @david_garcia,

Awesome,

But in your opinion, the client is worry about the availability/redundancy, I was thinking to go with two marketing instance, each one point for each tracking server, or do you think one marketing instance pointing to both tracking server with the load balance on the tracking server will have the same result?

Best,

Rick

Avatar

Community Advisor

Depends really, I would first start by doing 1 marketing instance with two frontal servers. then you can spin up another marketing server connected to the same database, in such case, if mkt1 goes down, then mkt2 is up and operators can resume.

It is also important on the database side, if the  default database goes down then both mkt1/2 and frontal servers will not be able to access the database, you need to configure a failover. if db1 goes down then services failover to db2, but in this case you also need a sql administrator to design and configure this for you.

Avatar

Level 3

That what I had in mind.

Two mkts, each one pointing to one tracking but same DB, (leaving the main one (mkt1) with the wfserver auto start) and the other one without, so in case fails I will make mkt2 assume it. (I can't have two wfserver in parallel, right?).

Cool, so I think I'm going to the right direction.

best,

Avatar

Correct answer by
Community Advisor

If you opt for two parallel marketing instances, I think you also need a loadbalancer in place..

You can do something more tricky and instead of having two marketing instances in parallel, you opt to configure a failover instance. I haven't done it before, but I would start by  creating an active and passive instance. On the passive instance, all services shutdown, and create a shell/powershell script on a cron job to monitor the active instance through various means (i.e - /nl/jsp/ping.jsp?) alerting, etc. and upon failure or connectivity error then activate the passive instance whilst continuing to monitor when the default instance is back online to failover to it again.

You can also opt make your default instance really robust, in my case I developed a few scripts scheduled on a cron job to monitor my services and perform troubleshooting automation, you can try out what works best for you and contact Adobe support to find out how they design/configure failover.

Amit has given a good explanation on a related subject.

Multiple WfServers Marketing servers for Hybrid Model

And more information on enterprise deployment.

Enterprise deployment

Amit_Kumar​ perhaps can advise henrique on this subject.