Disaster Recovery approach edge-gateway

[#1]

All Architects

Please suggest the approach for DR on edge gateway recommended by API Gateway .

I have following in mind ,but would like to have apigee supportablility and recommendations first .

• Replicate the DB [ Cassandra ] of one gateway with other gateway's Cassandra in different geography with different end point
• Frequently migrate data using management server APIs/ migrate tool to sync data from vulnerable gateway to a secure geography gateway server [ different end point ]

[#2]

@Arpit Sharma - Take a look at http://docs.apigee.com/private-cloud/latest/installation-topologies

This uses a multi-dc approach (or multi AZ etc)

The cassandra installation spans both DCs, you would run active/passive, failing over to the secondary location in the event of a disaster.

Data-sync would be in almost real-time (eventual consistency via Cassandra)

[sjm2000]

@Andy Trickett

It sounds great ha not a good disaster recovery strategy. Example if one location have malware attack, it's not good to be replicated in real time. Let me know if I am wrong . We all have similar issues for disaster recovery strategy when it comes to Apigee.

The reason being 5 node installation is HA but never mentioned in docs, how to make it a proper DR enabled system.

@arpit you have a solid idea. I wish it can be made in some script format.

[#4]

I have a question ,for real time data replication [Replicate continuously the 3-node Cassandra ring to the remote 3-node ring] . Is there a possibility that remote CS DB gets some I.P. references of source DB .If yes ,does apigee provide any script to update the references with target DB I.P. ?

[#5]

Did you check the For configurations with multiple Cassandra nodes section in the http://docs.apigee.com/private-cloud/latest/moving-apigee-servers? @Arpit Sharma

[#6]

@Maruti Chand i am not moving components .im thinking about just copying data from an established server CS node ring to another established target CS node ring . So is the link relevant ?

[#7]

@Arpit Sharma, okay I understand the context after reading your original question, I thought you were talking about changing the IPs.

BTW when you say [Replicate continuously the 3-node Cassandra ring to the remote 3-node ring} why do you want to do this as I see it is same as they being in a ring right wrt to the replication?

Using APIs to retrieve data and write to a different ring sounds like a complicated process as this is not one time job.

I think if you want to go with the design that you described, you should consider relying on the backup data but your IPs, org and environment names should be same.

I don't see IPs being tied to the data/keyspaces in C* but looks like restore process needs the IPs to be same as per http://docs.apigee.com/private-cloud/latest/how-reinstall-and-restore-components

I have never tried the above though.