One of my customer needs a Disaster Recovery Solution for following scenario:
- Primary Data Center is at SITE A, and the Backup Data Center is at SITE B.
- Clients are located at multiple other branch offices.
- Interconnectivity b/w all sites is on MPLS VPN. All sites have point-to-point connection to MPLS cloud.
- SITE A connects to MPLS cloud using dual 7206VXR routers. Each 7206 has dual E3 links. SITE B also has similary scenario.
- Branch Offices connect to MPLS cloud using single 2800 router. Each 2800 has single E1 link.
- Traffic is IP only.
The customer wants to do this:
1. If SITE A is up and SERVER A is down, the client to server traffic should be redirected towards SERVER B at SITE B. When Server A becomes available again, traffic should be redirected back to Server A.
2. If SITE A is down (obviously SERVER A will also become unavailable) the traffic should again be redirected towards SERVER B at SITE B. When SITE A becomes available again, traffic should be redirected back to Server A.
request 1) is about application redundancy, i.e. OSI layer4 or above; request 2) is about IP redundancy and routing.
So lets first look at request 2): this can be achieved through normal IP routing. In an MPLS VPN routes will be learned dynamically. So the redundancy and desaster recovery should not be a problem with respect to IP routing.
One question remaining is about the proposed IP addressing scheme: will SITE A and SITE B have identical IP addresses?
Assuming that this is not the case you need to investigate your clients. Are they capable of using primary/backup IPs of servers?
Are they using DNS solely? In case you have only the option of setting a single server IP in the clients, no backup server, then you need clustering options for servers across the WAN.
In case your clients use DNS you could have two IPs per DNS name pointing to primary/backup. This might also help with request 1). But understand, that in an IP solution like MPLS VPN you can not simply add "application awareness". There are new features like OER (Optimized Edge Routing) and objekt tracking, where you basically use SAA to adjust IP routing. I am however not sure if this will solve all your problems.
I am in a similar situation where I have an application that requires high availability and the application uses a heartbeat between the two servers. This heartbeat requires the two servers to be on the same subnet yet the client wants these servers at different data centers. So the dilemma is that the servers must reside at different data centers but be on the same IP subnet. I have posted a thread regarding this. Does one use GRE or L2TP?
Topology & Design:
Two ACI fabrics
Stretching VLANs using OTV
Both fabrics are advertising BD subnets into same routing domain
Some BDs(or say VLANs) are stretched, but some are not.
Endpoints can move betwee...
VMware Trunk Port Group is supported from ACI version 2.1
VMM integration must be configured properly
ASA device package must be uploaded to APIC
ASAv version must be compatible with ACI and device package version
Topology &Design:Traffic flow within same fabric:Endpoint moves to Fabric-2Bounce Entry Times OutTraffic Black-holedSummarySolutionAppendix:
In the Previous articles of ACI Automation, we are using Postman/Newman a...