Comments on: Single initiator zoning

By: kevin

kevin — Fri, 05 Mar 2010 06:45:07 +0000

Pretty good article.
Helps a lot.

By: RusselCorey

RusselCorey — Fri, 11 Dec 2009 03:58:26 +0000

I just wanted to point out that you can zone by WWN and still get hardware enforcement. Most modern Brocades will allow you to zone by WWN and get hardware enforced zoning. Pretty sure Cisco supports this as well.

Zoning by WWN is nice since the SAN administrator can prepare each fabric ahead of time without necessarily knowing which switchports will be used.

That said, single initiator with either multiple target or single initiator single target is my preference.

(I just realized this post is a year old…)

By: Todd

Todd — Thu, 01 Jan 2009 22:36:00 +0000

Wow, I’m glad I found this — great post.

Now, my question is…what do you use for a naming scheme for all those two member zones?

I currently use the following.

HOSTNAME_HBA1_SPA
HOSTNAME_HBA1_SPB

HOSTNAME_HBA2_SPA
HOSTNAME_HBA2_SPB

By: Titaniumlegs

Titaniumlegs — Mon, 01 Dec 2008 20:09:22 +0000

I see both sides of soft vs hard, but I lean heavily toward soft since people do move things around.
I also strongly recommend single initiator zones, but multiple targets is usually fine, and leaves less chances for mistakes. If you replace an HBA, you change one or two zones (if using dual-port HBAs), or just the alias(es).

The things that misbehave the most or do stuff like bus resets are initiators and tape devices (or the traffic to them). So, keep tapes (or anything non-disk/LUN) off the disk zones, and use separate HBAs for tape connections, if you’re even using FC tape right off the ESX servers.

By: JeremyinNC

JeremyinNC — Mon, 03 Nov 2008 17:57:02 +0000

I agree that single initiator is the way to go, but I prefer soft zoning since it’s trivial to take care of a bad port or moving a host to another spot in your data center (and possibly a different switch in the same fabric).

As far as how to visualize single initiator zones just think pure old school and keep virtual 1-1 connections in mind.

By: Duncan Epping

Duncan Epping — Fri, 31 Oct 2008 07:13:34 +0000

That’s why I said “prefer” and not “it’s a best practice” or “you must”… I prefer to, but it depends.

By: Dan

Dan — Fri, 31 Oct 2008 03:02:37 +0000

Hi Duncan,
I agree with the SI-ST zoning methodology 100%, but I really can’t agree with the Hard Zoning approach. In environments I’ve worked in, if we used hard zoning we’d be forever re-zoning.
Cheers
Dan

By: Jenny Mackintosh

Jenny Mackintosh — Thu, 30 Oct 2008 21:08:56 +0000

Good afternoon Duncan,

I would like to speak with you about a potential partnership between ITKnowledgeExchange.com and Yellow-Bricks.com. I wasn’t sure where to e-mail, so I thought I’d post a comment. Could you e-mail me or give me a call at your earliest convenience?

Thank you,

Jenny

. . . . . . . . . . . . . . . . .

Jenny Mackintosh
Community Manager
ITKnowledgeExchange.com
jenny@itknowledgeexchange.com
781.657.1681 – desk

By: william bishop

william bishop — Thu, 30 Oct 2008 12:27:53 +0000

Here also. Yes, a bit more to set up, but well worth it. On the zoning, I’ll go with the 20% of the group that uses soft zoning, it’s just easier if something goes wrong to move it without rezoning.

By: Bolderbast

Bolderbast — Tue, 28 Oct 2008 21:41:10 +0000

I agree with Jason on the Soft Zoning part. With soft zoning, replacing a defective HBA in a server would require a re-zone. But with hard zoning this would be nescessary when putting a server on a different fabic port. In the hard zoning configurations though, customers can run into trouble when tidying up their cabling and connecting servers to wrong ports on the fabric. I know, those customers should stay clear of this, but believe me: I know a few who have done this… So for me: Soft Zoning it is.

As for the Single Initiator, Single Target scenario. Couldn’t agree more: This is the way to go!