PI Disaster Plan

[Former Member]

Hi Experts,

currently we are having 150+ interfaces in production system and few are very critical and time sensitive for business. Recently we encountered a production issue which lasted for 8 hours to bring the system back up and running. How is anyone doing disaster plan for production PI system outages ?

Options I can think of is,

- Route time critical interfaces via quality region during outage time period and rever it back

- Maintain another production like PI system and keep your interfaces ready in it

- Other sender/receiver applications can develop interface with another middleware

Please suggest how in your organisation are doing for this.

Thanks,

Giridhar

[Former Member · Accepted solution]

Hello

A company I was working for had a MS Cluster with 2 active nodes running on XI 3.0 and later on XI 7.0.

Another company was running a virtuell environment and could switch the entire SAP system to another virtuell host within a few hours "only".

If you have that many interfaces you should defenitly have a DR concept in place. I guess money should not be the king for such a large environment.

Nesimi

[anupam_ghosh2]

Hi Giridhar,

The customer for whom I am working has a DR system in place . As you already pointed out we have hundreds of interfaces and production cannot be down for more than 2-3 hours. We maintain copy of same interfaces in more than 3 different servers. One definite server other than production is made a production-DR. We need to make necessary interface changes. On a specific day production is brought down and the production DR is up in its place. We check if the data is flowing through it. During the process all users are locked up except one. once testing is successful the production is up again and working. This ensures that a server is up and ready to work in place of production.

Hope this in short explains DR activity in short.

regards

Anupam

[Former Member]

Thanks Nesimi and Anupam for your answers.

I have few more questions,

- Do the DR system lying on the same host or IP address ?

- In case of SOAP Sender/Receiver Interfaces, we need to exchange the URL to other application system...so do we need to make this change in other application systems ?

- For Idoc/RFC, we need to change port configurations in Sender R/3 systems. Do we need to make these changes as well.

- Do external partner change the FTP address to send to new DR system.

Sorry for putting some questions like this, basically I want to understand how much effort will be there for PI technical team and other application teams to do when we are moving production interfaces to DR system.

Thanks,

Giridhar

[Former Member]

if you go with a MS cluster you will have basically one virtuell adress (hostname, ip) which will be mapped to 2 different Nodes (A,B). Node A and B both have their individuel hostname and IP adress. The only hostname and IP adress that you should use in all your config/external partners etc will be your virtuell name which will never be changed in case of a switch to another cluster node.

[Former Member]

Hi Nesimi

Thanks for your answer.

In my organization also PI production system is running on cluster environment with 2 nodes (A and B). The problem is when we are encountering any PI issue, the automatic failover of services and messages in the Outbound queues arenot happening between nodes bringing the whole system down for more than 4hours. So I am asking you about disaster plan when the entrire Production system is down.

Thanks,

Giridhar