Problem with Xen Server setup

qplayed

Hi Guys,

I just came across a problem. attached is a simple diagram of what I am dealing with. As we all know, in our industry there are no IF's only WHEN..

Here's a quick explanation of the issue:

I have 3 Xen Server attached directly to our core switch...actually everything is attached here....no dist or access switches. Then we have all the virtual machines stored on a NAS box(NetApp S5xx). As you can see with a dead core switch my whole infrastructure is gone same goes for the NAS. One thing to note is that I wasn't the one that initially set this up. I just joined the company a couple of months ago. My question for the experts is, how do I fix this problem? Is there really a problem? I know there are other underlying issues with this but we can discuss that as we go along. Also how do I address management with this issue?

sprkymrk

Unfortunately many networks are burdened with "single points of failure". Many times it's more of a cost-prohibitive issue in setting up redundancy than it is a network design problem. Some networks may only have a single mail server, or 1 pipe out to the Internet, 1 firewall, whatever.

So, yes, there is a problem, but it is a common problem and not unique to your network. In your case the only thing I can think to do would be to purchase more equipment. I would just make management aware of the issue and see if they can budget for more equipment. A few things I would do:

1. Ask what the dollar value of each hour of down-time equals if the switch or NAS were to fail.
2. Offer a low-cost solution (maybe a cold-spare core switch and some replacement drives) that would at least minimize the amount of downtime a failure would involve.
3. Offer a "best" solution that would offer real redundancy to your network.

Good luck.

qplayed

Thanks for the input, I've already got a few things written up as far as minimizing downtime, extra hw etc...I just wanted to get a different perspective in the situation. The "IT" department consists of a programmer and a "project manager" who doesn't know squat. I'd like to present my case to management once I've exhausted all possible/feasible solutions. I don't want to bring in an issue w/o any viable solutions. thanks again!

astorrs

What is the core switch? If we're talking about a Catalyst 65xx or other enterprise class chassis switch there may be redundancy built into the switch itself that may partially mitigate any issue. The StoreVault S500 you have is a single controller NAS filer so you're right in that it has a single point of failure there. I'm not sure if the S550 is available with dual controllers or if there is an upgrade for your existing one, check with your NetApp VAR.

qplayed

nope the "core switch" is a 4506. I'll find out on monday about the netapp and see what can be done.

astorrs

qplayed wrote:

nope the "core switch" is a 4506. I'll find out on monday about the netapp and see what can be done.

Yeah if it was a 4507 you would be in luck because you could at least add an additional supervisor module for redundancy...

I took at look at the low-end NetApp offerings and it looks like all the StoreVault's are single controller. You have to move to a FAS 2020 to get redundant controllers.

Not your lucky day my friend.