Two different Lan configuration is possible....

[Accullise]

Hi freinds...

I have a doubt, can anyone please clarify that...

can we form a Two different Lan in a single system through ethernet....

Actually we have four alpha server connected through memory channel and LAN thorugh ethernet.Now i want to configure another Lan cluster with same IP through ethernet in the same system for failover.(when first lan getting failed state, at that time new lan taking care of that)can i configure another different Lan cluster through ethernet with individual switch. Is there any possiblities are there?

[Wim Van den Wyngaert]

VMS version ? Which LAN protocols and which versions ?

Wim

[Accullise]

VMS Version - 7.3-2
Lan protocol - Decnet

[Wim Van den Wyngaert]

Never used it but http://h71000.www7.hp.com/doc/732FINAL/aa-pv5nh-tk/00/01/119-con.html#cegejebd might help.

If both nodes are connected to the same network, you can simply enable decnet on both interfaces (we did it once with ncl scripts I no longer have but check ncl enable). The traffic will be balanced on the 2 cards.

Wim

[Hoff]

I'm going to try to answer each of two different potential questions I can infer here; some background on what you are doing might be useful to determining where you're actually headed with your question.

The first is related to IP and IP networking and setting up IP to operate in that environment, and the second is related to clustering and setting that environment to operate with a degree of redundant networking hardware.

Clustering and its associated underlying SCS protocol, the DECnet protocol, the IP protocol, and MOP are NOT particularly related protocols. I mention this strictly as you appear to be freely mixing "LAN clustering" and clustering hardware (Memory Channel) and IP in your question.

Recognize too that IP and OpenVMS clustering are entirely unrelated network protocols, so whatever you do with IP will NOT provide network fail-over for OpenVMS LAN clustering, save as a secondary aspect and byproduct of having configured a parallel LAN and the default behavior of clustering having detected that LAN.

--IP--

As for IP, there was a pointer provided to the OpenVMS Technical Journal (VTJ) article entitled "Configuring TCP/IP for High Availability" in a previous question you had posted on this same basic networking topic here:

http://forums12.itrc.hp.com/service/forums/questionanswer.do?threadId=1236410

Here's the direct path to the VTJ article:

http://h71000.www7.hp.com/openvms/journal/v2/articles/tcpip.html

Please take the time to skim this material, then work your way to the appropriate solution for your IP network.

There are other options toward achieving IP network uptime beyond what is listed in that VTJ article, including managed switches, IP BGP, and other such; IP network implementations and devices outboard of what you are working on and working with here. These approaches are intended to provide uptime against various classes of failures. (You might and would often use these IP networking techniques with a multisite data center, for instance.)

--Clustering--

With SCS and clustering, connecting up redundant and entirely parallel fast Ethernet or gigabit Ethernet paths is often enough to avoid (most) hardware failures. These parallel paths should not share switches. Clustering deals with this transparently, and (if desired) you can "cost" the SCS network traffic onto a particular LAN. (Setting up load balancing for SCS is fodder for a longer discussion; SCS works pretty well by default, so that task is not necessarily needed in any particular case.)

[Accullise]

thank friend...

actually i am going to add a new lan cluster with existing cluster through individual switch by running cluster_config_lan.com. On that procedure just i am going to change the interface name only. After that it works as a individual cluster? or i want to change any parameters?

Regards,
Saravanan.M