VIP vs IP in RAC

Question

10.222.57.118 node4-nfs.us.oracle.com node4-nfs
148.87.199.6 node4.oracle.com node4
148.87.199.10 node1-vip.oracle.com node1-vip
148.87.199.11 node2-vip.oracle.com node2-vip
148.87.199.12 node3-vip.oracle.com node3-vip
148.87.199.13 node4-vip.oracle.com node4-vip
10.222.48.145 node1-rac.us.oracle.com node1-rac
10.222.48.146 node2-rac.us.oracle.com node2-rac
10.222.48.147 node3-rac.us.oracle.com node3-rac
10.222.48.148 node4-rac.us.oracle.com node4-rac
let for node 1 ip is 148.87.199.10 and VIP is 10.222.48.145
Here clinets used VIP 10.222.48.145 to connect to instance 1 on node 1.
if the node 1 fails VIP of node 1 : 10.222.48.145 diverts all connections to other nodes depends on load balancing.
Am I correct ??
And what this bellow node 4 entry corresponds to ??
10.222.57.118 node4-nfs.us.oracle.com node4-nfs
148.87.199.6 node4.oracle.com node4
ANd IP cannot failover as fast as VIPs ? Any reason for this ?
Thanks.
..

Answer 1

887400 wrote:
148.87.199.10 node1-vip.oracle.com node1-vip
10.222.48.145 node1-rac.us.oracle.com node1-rac
let for node 1 ip is 148.87.199.10 and VIP is 10.222.48.145I have no idea what you want to know or say. All you are giving us is a list of hostnames and IP adresses. My guess is that 10.222.48.145 is the IP for your interconnect interface on node 1 since both the VIP and public IP need to be on the same subnet.
Here clinets used VIP 10.222.48.145 to connect to instance 1 on node 1.
if the node 1 fails VIP of node 1 : 10.222.48.145 diverts all connections to other nodes depends on load balancing.
Am I correct ??No. Clients connect to listeners running on any of the VIPs (for 10g) or SCAN IPs (in 11g). Clients may connect to any listener randomly or in order and each listener will redirect new connections to an instance that is providing the service specified (or requested) by the client and connections may end up on another node depending on load-balancing and other parameters.
If a node fails, the VIP will be transferred (or brought up) on another node but it will not actually serve anything on that IP but rather reject all incoming requests. SCAN IPs together with an extra scan listener (in 11g) also fail over to other nodes but will actually serve connection requests.
And what this bellow node 4 entry corresponds to ??
10.222.57.118 node4-nfs.us.oracle.com node4-nfsI have no idea. I am not administrating your system. How would I know? From the name, I would guess it is an extra IP for an NFS server on node 4. But I have no idea what that IP or server does.
ANd IP cannot failover as fast as VIPs ? Any reason for this ?What is the question? The public IP never fails over, that is the one that an administrator would use to connect to the server OS (with ssh for example). And there is no point in having it fail over because every server still needs one "static" IP. The reason you need VIPs is that without them, clients could get stuck in a waiting state when a node goes down. When a node goes down during a certain phase of TCP connection establishment, the client will wait for a reply for as long as the TCP timeout which is usually around 15 minutes. To prevent this, the VIP is switched over to another node that will simply reject the connection as fast as possible so that the client can try again with another IP address.
Bjoern