Rodos Tracker

Re: VCB and Symantec Backup Exec 12.5 Questions

Rodos — Tue, 10 Mar 2009 22:28:17 GMT

Must agree with Dave on this.

I find backup in VMware one of the most misunderstood areas, especially when VCB is involved.

Backup design starts with understanding restore, what do you need to restore and how often? Some environments will want to restore individual mailboxes a number of times a week, if this is the case an application aware agent that can restore directly back into the system is going to be required. Other systems are rarely restored and usually only of a DR test, here a full machine image works just great. I find that most designs end up with VCB taking lots of the static and system data and then agents for a few key applications where extra functionality is required. Backup is all about restore.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: View/quickprep breaking Windows XP SP3 domain membership?

Rodos — Sun, 01 Mar 2009 07:21:02 GMT

Kam, you get this resolved yet?

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: How to estimate the man-days needed for VI3 deployment?

Rodos — Thu, 19 Feb 2009 13:25:16 GMT

You know the answer depends. What is the scope. How many hosts, VMs, what SAN, Networking, Migration, Training works are need in conjunction? What type of SAN? Multiple or single site?

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: Capacity Planner, VMware official security related document

Rodos — Wed, 11 Feb 2009 01:28:34 GMT

There is a security document in the dashboard. If its a real big issue simply use PlateSpin Recon. It does all the same, you only pay per machine for the time period you monitor (say one month) and you can then run and analyse the data at no further cost for as long as you want. All of the data stays within the customer site.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: Best practice for the number of luns per esx host?

Rodos — Tue, 10 Feb 2009 13:05:39 GMT

10 is low, Chad from EMC mentioned they are seeing around 20. Its going to depend on the sizes of your systems too.

40 is an okay number of LUNs to zone to the host, your limit is 256.

You may find this post helpful .

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: Age Old Question

Rodos — Tue, 10 Feb 2009 04:23:53 GMT

Well it depends.

First are both sufficient to meet your needs, because XenApp is different to XenDeskop is different to VDM. Lets assume that XenApp and VDM both meet your needs.

The question of which has a better ROI, which is different to your up front capital cost will be influenced by:

If you already have XenApp.
If you already have VMware.
The skill sets of your staff.
The existing management systems you have in place (deployment, inventory, monitoring).
If you have application packaging in place.
If you are going to virtualise your applications with ThinApp.
The ratio of your fleet between XenApp/VDM and physical desktops and laptops.
The number of applications, their amount of isolation requirements and how often you need to apply updates.

When I have costed up a base install of VDM for X number of desktops with X applications compared to building the same environment with XenApp the VDM comes out cheaper. The solution also scales easier. But your mileage may vary.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: Recommended hardware for VMware Scenario

Rodos — Mon, 09 Feb 2009 22:01:47 GMT

Man Ed is fast, I was going to say start with the HCL and as usual he beat me to it. I think sleep must become optional.

When looking at your system consider the appropriate amount of RAM for the given number of processors, especially if you are wanting to save some money. How many VMs are you going to run and how much RAM are they doing to need. You want to balance your RAM and CPU.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: Defragmenation of vm's on ESX.

Rodos — Mon, 09 Feb 2009 21:57:31 GMT

Good timing. Eric Siebert did a write up of this yesterday.

Have a read at

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: Moving Virtual Environment to another physical location - Any gotchas?

Rodos — Mon, 09 Feb 2009 21:54:50 GMT

You did not mention backup.

Is everything going in one truck, what if it crashes?

What teams do you need to have on call for support.

Who makes the decision to pull the plug and abort and by when do they have to make it to give you enough time to get things going again, or do you just have to move and keep working till its all fine?

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: Roll up, roll up! 60-point deployment blueprint

Rodos — Sun, 08 Feb 2009 07:42:00 GMT

Due by the end of March. Steve good to see you are being courageous!

Not so sure about all the bribing. I think the people you want will do it anyway and last the distance. Cheap rewards may bring cheap work. Just an opinion.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Operation Procedures

Rodos — Thu, 05 Feb 2009 07:24:25 GMT

G'day,

I did a blog post on operation procedures at but I figured people here would both find it useful but also be able to contribute to the discussion.

Here is the contents to save you jumping. Post follow up comments with any of your own idea or items.

VMware Operations

What are your operational procedures for your VMware environment. I often get asked, "Rod, now that I have my new VMware environment, what do I need to do to run it on an ongoing basis?" To me this comes down to two things.

Monitoring

Your monitoring system provides the following functions for you.

Ensures that you are alerted to any pending problems
Allows you to investigate the current and historical state of your environment to assist in trouble shooting
Provides uptime and usage information for management reporting
Provides capacity management projections

Here is an example list of elements for monitoring.

Free space of Datastores
Free space of Service Consoles.
List of orphaned snapshots
List of long running snapshots
Size of VC database
Monitor CPU READY (ms) or CPU %READY per VM per host
Monitor %CPU BUSY percentages per VM per host
Monitor network and disk I/O usage per VM per Host
Monitor service console memory swap usage
Monitor VM balloon memory and swap usage
Host downtime reporting
Server hardware faults (power supplies, fans, IO cards, disks, CPUs, RAM)
SAN hardware faults (disks and vendor specific)

Your monitoring will certainly consist of VMware vCenter Server and also your hardware monitoring platform. Often these are supplemented by a VMware specific product like Vizioncore vFoglight, Veeam Monitor or Nimsoft.

Management

Your management processes and procedures provide the following functions for you.

A list of maintenance activities to perform on a periodic basis.
A list of operational procedures on how to perform standard maintenance and trouble shooting tasks.
A change management impact matrix to detail the potential impact and risk of a particular type of change.

Here is an example list of operational procedures.

The procedure to create a new virtual machine
The procedure to place a new virtual machine within the virtual infrastructure into a Production state. This may be identical to the physical server commissioning procedure.
The procedure to place an ESX server into and then out of maintenance mode, migrating the guests onto other ESX Server hosts.
The procedure used to contact VMware for support. It should include contact information and specify contact methods as well as means of collecting information.
The procedure to add a LUN to an existing ESX server cluster.
The procedure to patch a template used for creating virtual machines.
The procedure to create a snapshot of a virtual machine.
The procedure to restore the virtual machine state to its previous state at the start of the snapshot.
The procedure for investigating user reported virtual machine performance issues. What to check and how to respond.
The procedure to add a disk to an existing virtual machine.
The procedure to expand the size of an existing disk for a virtual machine.
The procedure to shrink a disk used by a virtual machine.
The procedure to remove a disk from a virtual machine.
The procedure to decommission a virtual machine.
The procedure to migrate (VMotion) a virtual machine between ESX Server hosts in the same ESX cluster.
The procedure to build an ESX server.
The procedure to add an ESX server into an existing ESX cluster.
The procedure to migrate a virtual machine between ESX Server hosts in the different ESX clusters (i.e. between datacenters).
The procedure to confirm that a SAN link is active, to be used after a SAN link has failed and been restored.
The procedure to confirm that a network link is active, to be used after a network link has failed and been restored.
The procedure to enable the network group to troubleshoot user reported network / performance issues.
The procedure for backing up/restoring VMs (VM-level and file-level).
The procedure for backing up/restoring VirtualCenter database.
The procedure for backing up/restoring license server files (or keys).
The procedure for restoring VirtualCenter Server.
The procedure for restoring ESX hosts.

Rodos

Blog: http://rodos.haywood.org/

Re: vmotion between SRM clusters

Rodos — Tue, 03 Feb 2009 01:39:08 GMT

You could make it work but you may have results you don't want.

If you vmotion it to the other cluster you will need to zone the LUNs to both ESX clusters. Only one SAN can be primary for the LUN so if you move the VM to the other site all of the IO will still go back to the previous site, you could then storage vmotion the disks to a LUN which is primary on the same site as the VM. However this is going to make your SRM really weird and difficult to setup if its covering those same VMs. Also to vmotion across clusters do they need to be controlled by the same vCenter Server, thats going to be very hard if you want to use SRM and have a separate vCenter Server manage the ESX hosts at its own site.

Have a quick read of one of my blog posts

http://rodos.haywood.org/2009/01/moving-workloads-split-cluster-or-cloud.html

which also points to an even better one to read on this topic

http://virtualgeek.typepad.com/virtual_geek/2008/06/the-case-for-an.html

Keep the discussion going.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: DR plan assistance, vreplicator, data domain etc

Rodos — Mon, 02 Feb 2009 22:57:50 GMT

Greg.

It may be out of your price range but check out Avamar from EMC and its competitors.

By the sounds of it what you need is a dedupe backup system. As you have already looked at DataDomain is a good fit for this but the size of your sites does not warrant a hardware solution but most likely a software one.

You may want to do whole machine states so you could use a local VM as a VCB proxy that did a single instance replication back to head office. The issue will be the space at the remote site to hold the backups.

Also don't be surprised if the existing VMware style 3rd party backup vendors such as Vizioncore and esXpress start to build dedupe into their coming versions.

VMware have dedupe in their yet to be released backup appliance which could work real nicely for you.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: 64-bit ESX?

Rodos — Mon, 02 Feb 2009 22:17:00 GMT

Further to what has been said note that their is public speculation about the next version (vSphere) and 64-bit.

http://lonesysadmin.net/2008/11/13/heads-up-vmware-esx-going-exclusively-64-bit/

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: vmotion between SRM clusters

Rodos — Mon, 02 Feb 2009 21:49:54 GMT

Rbirri what are your requirements driving this desire? Are you thinking of splitting a cluster across two sites? Knowing your requirements we may be able to shed some further light on what you are wanting to do.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: View Manager vs Lab Manager

Rodos — Sun, 01 Feb 2009 19:50:12 GMT

With view you can configure it either of the ways you have described.

For the second method where there is central master image with linked clones make sure the environment is capable of and built to be able to support a desktop recomposition.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: RDM or multiple large virtual disks?

Rodos — Thu, 29 Jan 2009 07:57:35 GMT

RDM will have higher I/O then a typically LUN mapping but this is generally used in cases where a IO intensive SQL server or Exchange server needs the extra I/O. If you file server is not having an I/O saturation problems you could probably stick with normal LUNS. As far as chopping it up you can stick with 2 2TB vdisks and then use windows Extend functionality in Windows, however I wouldn't recommend it.

This does not make much sense to me. First lets get some terms right. SANs have LUNS which are presented to ESX hosts. You can map those LUNs directed to a VM via an RDM, so the LUN gets used only by that VM and the LUN is formatted directly with the file system the VM uses. Or you can turn the LUN into a Datastore by formatting it with VMFS. As a datastore you can put multipe VMDKs for different VMs. You can take another LUN use it to extend an existing datastore.

So, back to your text. Are you saying that RDMs perform better than VMDKs on top of VMFS? This is really not true, the results are very close. What is sticking with normal LUNs, thats done by using an RDM, do you mean sticking with VMDKs. I can't understand why you would create 2TB VMDKs (vdisks) you don't really gain much, unless you use extents which just about everyone thinks is a bad idea.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: Running a legacy Human Machine Interface (HMI) app on new hardware?

Rodos — Tue, 27 Jan 2009 01:08:20 GMT

Oh, by the way, welcome to the forums.

In what I wrote I left a lot as an exercise for the reader. If you need further explanation on anything just pop a question and lots of people will assist.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: Running a legacy Human Machine Interface (HMI) app on new hardware?

Rodos — Tue, 27 Jan 2009 01:07:01 GMT

Start with some basic testing.

Determine if it has any specific hardware requirements such as serial devices or special interface cards. See if you can convert these to run over the network using Serial to IP converters etc.
Can you get an image of a machine, using Ghost or something. Run up the image on VMware Workstation and see if everything works.
Is the OS on the compatibility list. Not running any real time OS.

Once you are that far you are ready to start to work out what infrastructure to run it on, what VMware features you may need. How to handle the periferals, cabling, redundancy etc. You should be able to create a great solution and its a really good use case. Just be careful of any code in the systems that is dependent on clock cycles.

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/

Re: WAN traffic sensitivity

Rodos — Mon, 26 Jan 2009 21:48:59 GMT

Strange. Do your VPN connections terminate straight onto your firewall? How are the resources looking on the security server? Anything in log files?

Rodos

Consider the use of the helpful or correct buttons to award points. Blog: http://rodos.haywood.org/