The attached document contains a sample architecture diagram that is based upon a real implementation.
Hi Indrajit
We are in the planning of an architecture for DR with all the components you mentioned. I am worried about DR DB activation in case it's needed. We could use DR for high availability and Load balancing, but in that case, the DR Database wouldn't have any use until a full DR activation. Am I right? Any toughs about that?. Do you have that model in use?. Any other experience about the DR activation process? Thank you.
Hi,
The main intention here was to use the advanced availability capability of Service Desk Manager to build a DR solution. This is focused to SDM only with minimum DR components. The idea is to use the DR Standby Server to switch role to Background Server in case of a failure at the main site. In case of a complete site disaster, the DR database role should become active so that the new DR BG server would connect to it. The DR SB server and the DR App server are always part of the configuration. In a larger scenario, we can make this design more intelligent by using multiple servers at the DR App level and have a HW load balancer to segregate user traffic based on locations (example: users closer to the DR location get routed to the App servers in the DR site). That way you make use of the existing DR systems rather than keeping a set of standby resources. However, the pipe between the two sites must be fat enough to support acceptable performance.
To consider other components, this design has to be extended further - this is just a starting guideline. This does not show the other dependent components (outside of the CA SM components) like LDAP/AD or Email Servers or other CA SM components like Service Catalog / PAM / EEM, etc. Ideally, you should also plan for DR architecture for these too. You are correct - the file server should also be replicated to make sure attachments are available in a DR situation.
Hope that helps.
Warm regards,
Indrajit Banerjee.
Thank you for this example! Could you answer some questions on assumptions of the design?
What is the expected recovery time? Since core SDM and its data are the only functions being replicated, I assume this is for immediate Incident and Problem management around the DR event. What are their priorities on restoration of the other functions?
Do we assume that the SDM authentication is external via AD or similar and that the AD is also part of the DR but not shown here for clarity?
Similarly, is the file server also replicated or does the use case assume that attachments are not essential for the recovery period?
Regards,
J.W.