need help in work case.

Hello,
I need you help about the following case:

  • We have 2 remote site and both of the has very heavy app (Oracle,SAP...).

-We need high availability though 2 IP (link aggregation)
i.e 2 internet card for each server.

*Hope this is understandable:

What We want is:
in case one of the IP/links is down, the process and the transactions in re-directed to the 2nd live IP/link?!
How to make this redirection (MUST maintain the process and transactions online)?by switch or by severer or HOW?

Kindly answer ASAP:

Regards
Ahmed Amer
Cairo,Egypt

Not sure I understood:
HA through 2 IP , Ok, but what has link aggregation to do with?
When you aggregate lan interfaces you hope having a better throughput (bandwidth...)
Not the same thing as to assign separate IP to two lan intefaces (2 IPs)

What are you trying?

all what i care about is HA and in case there is link failure between the 2 site (let`s say site in Paris and the other in London),that dose not affect the transactions!
The employees in London use the server in London and we don`t want any failure in the connectivity!

I said link aggregation so 1 link/IP is active and the other in standby!
in this case,link aggregations,I am afraid if the duration that the backup/standby link/ip is long and affect the transactions!

Regards
Ahmed Amer
Cairo,Egypt

Well...
When I mounted HA clusters, I used to have not 2 but more than 3 lan cards or like in HP a card with 4 ports and even then I added 2 extras...
For to be safe you would need 2 (3 better) separate complete LAN infrastructures.
The idea is you have a "private" LAN for your heartbeat, thats 1-2 lan connections, 2 connections to your production LAN, one in use the other on standby etc...

I'm afraid that if you are to build HA you will need HA software... like HP MC/Serviceguard or IBM HACMP or oracle...
Since you have not specified your platform it will be difficult to say more...
Now as an altenative (good...) as you seem to be interested by only not loosing connection, then link/port aggregation is the way to go look at HP's HP-UX APA documentation to see what I'm talking about and how it works...
I know you have the same thing for AIX, SOLARIS, and quite sure LINUX (we use them on AIX and SOLARIS but just cannot remember the name...)

We use HACMP!
you highlighted many points :slight_smile:

BTW,a colleague of mine just told me about something called "bonding"!
If it is acceptable to post the website of this solution so may be someone gets the info,it will be great!
Kindly,let me know if it is acceptable to post the site here!

Regards
Ahmed

So you are on IBM AIX platform..
If you agree tomorrow I will move the thread there so my AIX peers will be more aware than myself... last time i was in charge of a HACMP cluster was some time ago on a SP2...

Yes you can post (or try to...) the website the more information on your solution the more the better for us to understand and try to help you...

Hello vbe,
move it to there if you like,it is OK to me if you do :slight_smile:

After a lot of searching and reading,i think the most reliable solutions are:

-GLBP - Gateway Load Balancing Protocol

as far as i think,we will use one of those 2 solutions.

Regards
Ahmed Amer
Cairo,Egypt