Lefthand Networks provide a great solution here by virtualising the local storage on the ESX box and Network RAID across to the other ESX box. You need to run the Lefthand VSA on each server, but does provide that storage redundancy and vMotion capabilities for Virtual machines.

Hope this helps :)

If you have access to another server that you can run Windows 2008 on - then you can use this server as an NFS store. Then implement HA across your ESX hosts to ensure you have hardware based failover capacity on each ESX host. You can also VMotion across the hosts with your shared NFS store.

VMWare itself doesn't have any storage failover capabilities yet - only hardware based failover with HA. Hope this helps :)

No worries- moe than happy to help out. That's what communities are all about - best place to learn and help others...

Windows 2003 can do the NFS data store if that's all you have so that should be sufficient. Your ESX servers (depending on your license model - I am going to assume Enterprise) will have HA, VMotion and DRS. This means that you have the required failover you are looking for. You are spot on - with two ESX 3.5 servers and a central NFS data store - then you have yourself a pretty standard, redundant and highly available ESX cluster. One other thing with your NFS store - just keep this under 2TB and you will be fine :)

Cheers. Feel free to ask as many questions as you need...

For the Windows 2003 NFS side of things, check out the following blog - it describes really well on how to create an NFS share and then give it to an ESX host. They are primarily doing for ISO storage, but no reason why your Virtual Machines can't be stored here (if anything NFS is a great datastore location because of speed, thin provisioning, etc)...http://vmetc.com/2008/02/19/create-a-nfs-share-for-vm-iso-files-with-windows-200 3-server-r2/

Because you have Enterprise licensing then - yep, you definately get VMotion - so thats a plus. With the High Availability, there is downtime should one of the ESX hosts go down. This is because the Virtual Machines are restarted on the other host. There is normally a delay of approx 60 secs (which is tuneable - but I wouldn't change the settings) and then the time it takes for the Operating system to start.

Sounds like you have some fun times ahead during the next week or two - good luck and feel free to ask any questions if you like...

Hey - No worries, glad to help out. Sounds like you are moving forward which is great.

With the vSwitch setup, there shouldn't be any reason why you can't configure a single vSwitch and use Port Groups to segment the traffic. You should then have two VMKernel Port Groups (SC and vMotion) and then another for your VM's. With your NIC setup, you can then configure all as Active and then use IP Hash for Load Balancing. This ensures traffic load balancing to and from the ESX servers, even when the VM gets vMotioned.

With your Physical switch (am assuming you are only having 1, but would naturally recommend 2), I would recommend trunking the four ports that your NFS server is plugging into via LACP (oops - I assumed your NFS server was a HP Server?? if not - then disregard that last comment). Within your vSwitch all you need is a seperate VLAN for the vMotion and align the VLAN ID with the Port Group ID within your vSwitch. It is recommended that your SC is on a seperate Network as well, but to be honest it really doesn't make any difference and 99% of customers don't seperate this one out. If you do have 2 physical switches then definately make sure your Load Balancing is set to IP Hash - this means that if you vMotion across to another host or a NIC goes down, the packet doesn't get lost between the VM and workstation talking to it.

Other than that stuff - have fun with it - you'll find its all pretty straight forward and your environment will be really strong. How many VM's and what sort are you putting on (service wise - Exchange, SQL, etc) this will determine your DRS strategy as well.

Have fun

Cool - yeah there is a lot to this stuff, but nce you get through it, its a great feeling.

Your on the right track for the loopback, but its in the ESX Server itself. You need to modify the /etc/hosts file and add a loopback address : 127.0.0.1 localhost

Hope this helps...Enjoy..Yell out if you need anything else.

For the error :

Install the P2V client tool on Windows 2003 physical server manually and try to follow some best practice for P2V. I have updated few on my blog http://www.google.com/search?q=P2V&sitesearch=vikashkumarroy.blogspot.com

For this kind of scenario you must use stand alone P2V tool.

Coming to what you are looking at

• 1. Install ESX host on both DL380.
• 2. I guess when you talks about HA/DRS and fail over then you may have VC license. So add both the ESX host into VC.
• 3. Create a new VM with windows 2003 or use the one which you are doing P2V.
• 4. Now create NFS share on win2k3 using blog http://vmetc.com/2008/02/19/create-a-nfs-share-for-vm-iso-files-with-windows-200 3-server-r2/
• 5. When you are installing NFS service on windows , you have like to install it on both ESX host . The reason when one of the ESX host is down the VM will go down which means your shared storage is down. But if you create VM on both the ESX host then you can provide redundancy for each ESX host.
• 6. Once you are shared NFS folder on WIN2K3, create VMKernal port on each ESX host and mount NFS store on it . You then create datastore and then put VM inside.
• 7. These NFS can be shared on both the ESX host so that you can enjoy benefit of vmotion /HA

Let us know how it goes

Vikash

My Blog