Day 2 Started with another Networking session, this one I felt was a little better than the last and I will post a link to the Slides when they appear but some of the highlights are :

  • VM’s that need to talk to each other should be on the same Vswitch so the external network is never touched.
  • A typical esx server with 4 network cards should have the service Console and vmkernel on the first vswitch connected to two physical Nics ie one onboard NIC and one PCI NIC. Vmtraffic should be on the second vswitch which also has two physical NICS attached.
  • Use separate portgroups for different VLAN’s
  • If you have 5 to 10 network cards use network trunking
  • Virtual port ID teaming is faster than MAC based teaming and is the preferred option. It load balances all VM’s across the switch not load balanced per vm.
  • IP Hash load balancing does per vm teaming but apparently you have to channel the ports so its not recommended unless you have a VM with a really high load.
  • There is a Link state tracking KB article coming soon.
  • NIC members in a team must be in the same layer 2 domain.
  • If you use channelling then use IP Hash teaming but make sure you disable channel negotiations  as this is not supported.
  • Avoid using Native VLAN’s.
  • TSO/LSO (TCP segment off load) is not supported
  • VMware` are working on pass through where the guest can access the physical NIC directly for near native performance
  • Infiniband is supported
  • 10 GB Ethernet copper is coming soon.

 

Next was a session called Top support issues another good download when the slides arrive but highlights were.

All the support guys are Irish (based out of Ireland, I’m still not to sure how good the VMware support is).

They did an example of how to recover if you loose access to the console which is exactly what happened to me in the past so it is obviously a common problem also chapter 7 in the following document is good.

Google : serviceconsole-guide.pdf

  • Issues with STP (spamming Tree Protocol) can cause HA to fail over.
  • Don’t expand a VMDK  disk which has snapshots on it as it will buggered
  • Then he went on to explain how snapshots can fill the entire disk and corrupt the VM so you have to restore from backup. Which is exactly what happened to me when working for NSK last week. Hence you should all have read the email I sent out. (when snapshots go bad).
  • You can crop snapshot delta files but you loose the data.
  • You can have a maximum of 32 snapshots.
  • Don’t use extents but if you have to add one to an ESX server make sure you then manually rescan the LUNS all the other ESX servers so they know about it.
  • 2TB is the limit of VMFS
  • There was a section on how to recover after deleting a VMFS partition.

Another Support session was next with another invaluable slide set.

VI3 won’t let you present the same LUN to two ESX servers with a different LUN ID as it will presume it is a snapshot LUN and automatically hide it. This happened to one of our customers.

  • You can see the LUN being hidden in /var/log/vmkernal

There are two ways to deal with it :

·         Resignature – used when presenting same lun to the original ESX server you have to reregister the VM’s

·         Disallow snaphot LUNS can be changed to 0 so LUNs are visable.

 

There is a driver that allows writes to VMFS2 volume when doing an upgrade.

VCB issues

o   Disable automount

o   The same LUN id’s must be used on the proxy as our being used on the ESX server

o   Multipathing is not supported on the proxy

 

 

Did the VI3 VCP exam and passed .

 

 

Next I went to the ACE session which is the VMware product that no one has ever heard about as it is niche and VMware still can’t get their act straight when positioning it in the market. (Also there probably isn’t a Sales person alive that could understand/sell it – Ok well not one at Ultima anyway – the gauntlet has been laid down...).

 

Basically its VM running on workstation  6 that is wrapped in a security layer.

You can set

o   When it will expire

o   How many times it will run

o   Who can use it A.D. integrated

o   Password protect it

o   Restrict its networking

o   Block keyloggers

o   Manage it with ACE Server

o   And loads more

 

So it might be great for outside consultants coming in to your company where you give them this VM on a USB key (oh yes its portable) and its a totally locked down VM.

 

Marketing give you all this rubbish about other scenarios like training and demo’s so I pointed out to the presenter that none of the LABS or other DEMO’s were using it at TSX.

 

I was really looking forward to this session as Gareth told me that ACE 2 was going to have some great new features like VDI integration. Well it doesn’t  its just a polished ACE v1 with USB stick compatibility. So thanks for wasting an hour of my life Gareth! In your defence though, they did say at VMworld that they might do some of that stuff but never did. I think this product is basically like a Virtual Appliance and will probably merge with the VA’s.

 

Highlights from the trade show : --

            Got two highlighters, three pens, a knife, a polo shirt a nice rucksack and a USB hub. – RESULT!!!

           

            SCRIPT LOGIC have a stall and are pushing all their normal stuff which I won’t go into as the Microsoft guys will already know about that.

 

            Double Take where there – they have all their normal “in-vm” replication but more excitingly and ESX level replication engine to compete with vizioncore and ESXreplicator – its agent less and called Double Take for VMware Infrastructure. – DEMO from website.

 

            Aexia – Never really heard of them before but they also have a replication engine at the esx level Virtual Solution Box (VSB) which is a Virtual Appliance and has similar functionality to esxRanger. – FREE Demo from website.

 

All in all a very good day can’t wait to get back and try out some of these new products.