Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

We are looking at a Dell M1000e enclosure and appropriate Blades with 4 NICs each. We are planning on running Linux/Oracle 11g RAC on two blades, storage will be handled on an iSCSI SAN for which two NICs (via passthrough) will be connected leaving us with two NICs (via blade centre switches).

We would like to have an interconnect (obviously) , an external IP and an internal IP.

Would best practice be to:

  • bond the remaining two interfaces and VLAN as appropriate to provide three virtual interfaces?
  • run the interconnect on one interface and VLAN the external/internal interfaces?
  • purchase a blade with more NICs as the above is a terrible idea?
  • Another option?

Please feel free to point out the blindingly obvious or to relevant documentation on

I am specifically interested in supported configurations and best practices.


share|improve this question
up vote 1 down vote accepted

To your question, I am a fan of separate interfaces for the OCFS2 Interconnect, but from your setup it looks like the best solution would be NIC Bond for two interfaces and VLAN out the virtual ip's for the ext and int (public and private) hostnames. The separation for the interconnect traffic is just for latency, not bandwidth. If you are certain there will be no latency to affect your interconnect there should be no problem in your BOND + TRUNK solution above.

share|improve this answer

I'm no fan of using iSCSI at 1Gbps for anything that requires decent performance such as a database box. I'd switch to 8Gbps FC myself but if that's not an option then I'd think about using a pair of 10Gbps NICs per server and using VLAN tagging to create virtual interfaces rather than using more 1Gbps NICs. This way it'll perform as well as let you chop-and-change how your network is setup on a server-by-server basis.

I'm not a Dell guy but it appears they don't have any half-height blades with dual 10Gbps NICs on the system board, which is a bit old-school, but you can add the Broadcom 57711 adapter pretty easily and then use their M8024 10Gbps switch, ignoring the built in 1Gbps NICs.

No idea what you're iSCSI SAN is but obviously it would make a lot of sense to 10Gbps-enable that, or stuff it full of 1Gbps adapters.

That's what I'd do anyway.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.