How to install Microsoft App-V 5.0 SP1

I never expected it took me whole afternoon on just installing a App-V, well, I do admit I did went out for other stuff, but it is little ridiculous how complicated and lack of correct information the whole process is.

I write this post to help myself and others for future reference. If you do run into similar issue, hope this post can help.

Test Environment:

Windows 2012 runs HYPER-V on it

SQL 2008 R2 runs on Hyper-V

 

SQL Preparation:

This is the part which kills everyone. If you think you just jump on the SQL and create couple of DB and jump back and run installation wizard, then I can tell you will see following errors.

This warning is telling you that you didn’t prepare DB as what Microsoft wanted and go back now!

Funny thing is it only has one button “OK”, no cancel?

 

If you continue to install, you will run into The error was : Invalid object name ‘dbo.RoleAssignments’ after you finish installation.

The correct installation method is following kirxblog and prepare the DB.

Essentially, there are two methods to prepare DB. You can either bring installation disk on SQL server and run through installation wizard which requires you to install whole bunch of craps or you can manually install it as what this post will tell you.

1. get DB scripts ready

In terms of extract DB script, you need to find installation exe file and run it with /layout as parameter.

Once you have done that, you will following structure

2. Create 3 AD groups

Now, according to installation instruct of script, you need two groups which are read and write. I also create 3rd group called AppVAdmin for managing.

You need to add App-V server computer account into AppvAdmin group. Also domain admins as well.

3. Modify DB scripts with Powershell script

The original DB script which you extracted from installation exe file requires SID of AD groups in terms of making it work! It’s not easy to find those SID and not mention other nasty stuff.

Thank for Kirx beautiful powershell script, we are able to prepare those DB scripts with Powershell script.

Instead of all other nasty information, all what you need to do is to provide name of groups and name of DBs. Powershell script will prepare the DB script and help you to finish the work.

download Kirx script and edit.

Found those lines and replace it

 

If you never run powershell script on this serve, you may want to unlock it.

Make backup for your DB scripts and copy them to c:\temp folder like this.

c:\temp\ManagementDatabase and also report database script.

Copy kirx script to c:\temp

Run it, if it works, it should return with no result.

However, it will change the file name under those DB scripts.

Original folder.

Changed

 

3. run the Scripts on DB

Copy your new script to SQL server and open management studio

Run 1-1 Database script which will create new DB on SQL

Run all other script base on DB you just created

 

4. Add AppvAdmin group to db.Ower on new DB

Your AppvAdmin group should contain App-V server account by now

 

Install Prerequisites for App-V

You need to download Microsoft C++ 2010 x86 version and install on App-V server. Yes, it is x86 not x64 version!

Otherwise, you will get this one.

 

Install App-V

Now, we can start to see those fish eggs , oh, no, orange Apps in the Box

 

Now, Unless you have SQL on your App-V box, otherwise, DO NOT tick those boxes. It will change your wizard steps.

 

 

Once you finish installation, I strongly recommend you to restart your App-V and most likely, you just add computer account into AppvAdmin group and requires restart to take effect.

Otherwise, you may see this.

After restart App-V, you should be able to this no error on console.

 

 

 

 

 

 

 

Reference:

http://kirxblog.wordpress.com/2012/11/01/remote-database-preparation-for-app-v-5/

Using MVMC convert VMDK to VHD

After poor marketing design vRAM past, Vmware has made another mistake on restricting of free license of ESXi. The free version of ESXi makes you only run VMs on host which has no more than 32GB memory.

This change (in my opinion, it’s also a wrong decision again!) pushes all customers from Vmware side to Microsoft side and also with no free licensing on Test and DR environment (which means you have to buy Essential Kits to cover Test and DR Hosts), we are facing no options but start to use Hyper-v in all tests environment.

One of most important elements of using Hyper-V is to convert existing VMWARE VMs to Hyper-V server.

MVMC (Microsoft Virtual Machine Converter) is one of tools you will use during the process of converting from Vmware vSphere to Microsoft HYPER-V.

Now, you will face some common issues and I hope this post will help you save some time.

My test environment:

1 Hyper-v Windows 2012 server

1 Vmware ESXi 4.1 with Windows 2008 R2 VMs

The goal is to convert VM from vmdk to VHD and up running in the Hyper-V

MVMC components:

MVMC actually has two parts.

1. MVMC to move VM with both CLI and GUI.

2. MVDC (virtual Disk Converter) to only convert disks from vmdk to VHD.

Start MVMC

Now, if you run the MVMC Gui, you will get this interface,

You are required to type ESXi or vCenter information here to get access to VM list. Behind the scene, What it does is to access ESXi web API to check VM lists on that ESXi.

As you can see, here is where you select your VM to be converted. The BLUE exclamation mark indicates that VM can’t be converted. There are quite few reasons why a VM can’t be converted, but on this picture, State is “OFF” and another VM without Vmware Tools can’t be converted.

Now, here is the interesting part. A login user can be used here directly, and notice it says Final State of VM is ON or OFF?

Again, it’s Final State of VM. It’s not all time states. so the original VM will be shutdown no matter what and it’s up to you to make it start again at the end.

If you want to convert VM on the fly without shutdown a VM, this is not the tool you should use.

Behind the scene, MVMC talks to ESXi to get snapshot of original VM while it’s running, the snapshot should include memory status as well. so estimate free size of VM storage.

Once snapshot is done, it removes Vmware tools and shutdown the VM. Then, it will export VM as OVF template which essentially isolate VM from hypervisor hardware layer. If your VM is a Microsoft TMG, then you probably don’t want to do that since TMG remembers the MAC of it’s nic and stored it in it’s own database. Now, there is a way to “hack” TMG, but this post doesn’t include it.

But I strongly suggest to use a UNC with share to store VMs.

This is where VM will sit. Now, according to Microsoft, it requires at least double the size of VM. As matter of fact, it requires triple size space rather than double.

Behind the scene, Like what I mentioned above, a OVF template has been generated. Now, OVF template is a compress format, so a 50GB vm can be compressed around 13GB. so this 13GB is downloaded to \\testhyp02\vmdk folder and MVMC will start convert it to 50GB vhd. After that, 50GB vhd will be copy (or upload) to Hyper-V VM location so that’s another 50GB, compressed if you use Dynamic disk. finally occupied Size is 13+50+50=113GB.

Now, when you see this page, you are pretty happy that VM has been converted and insert into HYPER-V and up running.

Now, remember this is VM from OVF, so some clean up must be done.

install new version of Integrated Service.

If you don’t install this integrated service, those hardware can’t be identified.

It does warn you the OLD IP exists, but after overwritten IP, it seems working. If you want to show hidden device of that old NIC, you won’t be able to find it.

Now, What’s Wrong?

MVMC has used multiple Microsoft technology in terms of converting.

The most common error you will get is this.

You can find log file here.

Now, for Windows 2008 R2, I recommend you to do following steps.

1. Enable Winrm

Winrm allows Windows server (2012 in this case) to remotely manage VM in terms of removing Vmware Tools.

2. Enable Server manager Remote management which will create FW fules

3. Make sure FW rules exception is ticked

3. You need to enable WMI and File and Print rules on FW no matter FW is on or OFF.

Now, after you have enable all three tricks, you can run some simple test.

For WINRM, you can run “winrs /r:yourTestServer ipconfig” from your MVMC server. If it shows result, then WINRM is working.

for WMI, you can download WMI administrator Tools and remote access root folder of VM, if you can see content, then it means it’s working.

Restart MVMC wizard and try again.

What about convert VM on the fly without shutdown ?

I’m not quite sure whether any software can do that at this stage. But You DO can clone a VM to an isolated network. run VM and import it to Hyper-V with MVMC.

For large size disk VM, you may want to use MVDC to convert disk and build a VM in Hyper-V and insert disk.

please let me know any thoughts.

 

Add-on:

following is answer from Microsoft for converting V2V on the fly.

Answer back from some of our Hyper-V guys…

 

==============================

It’s “possible” to perform a P2V conversion on a VMware virtual machine using Virtual Machine Manager, which leverages VSS to take the required snapshots while keeping the target machine online. However, it’s required to remove the VMware client tools first, which will probably require a reboot anyhow.

 

References:

 

VMM P2V
http://technet.microsoft.com/en-us/library/cc764232.aspx
Troubleshooting P2V

http://social.technet.microsoft.com/wiki/contents/articles/439.vmm-troubleshooting-p2v-conversion-issues.aspx

 

Reference:

http://technet.microsoft.com/en-us/library/hh967435.aspx

Download:

http://www.microsoft.com/en-us/download/details.aspx?id=34591

How to configure vDS(virtual distribution Switch), rename Uplink, Create vMotion, Delete Uplink

First of all, Happy New Year of 2013!! I am happy the whole world didn’t blow up and my guess those Mayan dudes just running out of space on that piece of stone  so they think, why the hell I need to care about world in thousands years later?

Now, back to Vmware. With vSphere 5.1.0b released, I start to wonder whether it’s time to consider to use vDS (Virtual distribution switch) to replace VSS.

vDS has been around for years, only Enterprise plus license would actually use it. The concept of vDS is great, but the real world is not practical from my point of View to use vDS to complete replace VSS.

My suggestion is to have hybrid environment with vSS and vDS. As matter of fact, that , I’m afraid, is your only option. There will be time for you to failover VMs from broken vDS to something else, so between another vDS and vSS, which one you would go?

I did a little bit research regarding vDS and I would like to share some tricks and “how to” to everyone. Feel free to pop up question and correct my mistakes as usual.

vSphere Client or vSphere Web Client?

Now, with vSphere Web Client getting more and more popular, should we use Web Client and dump old one? The answer is No. The new Web Client is incompleted, slow but it does provide more functions than C++ version. I will stick with Web client in this post as much as possible.

What’s is vDS?

You can always find this answer from my old post here. Comparing with vSS, vDS provide more virtual gateways (not like vSS, vDS also virtualize Uplink). More control and monitoring on the traffic going through virtual switch and also profile base deploying from vCenter to Hosts so vDS is aware all hosts network rather than working alone like vSS.

However, it does bring lots of other issue if you want to put vDS into production. One of few issues is to rename Uplink.

Why do we need to rename Uplink?

Uplink exists on vDS only. It’s a virtual port group which you connects your physical  nics to. Assuming you have 10 hosts, it’s hard to guarantee all vmnic01 will connect to Uplink01 since vmnic01 may connect to different network in the real world. After a while, you may get confused about what each Uplink for.

Tricks:

Always rename your Uplink before you start to connect anything to vDS.

You need to rename your Uplink ASAP after you create your vDS. Once vDS is hook up something, it simply won’t let you touch Uplink because it may connect to something. Even if you remove the connection to another link, the vDS will still hold same configuration till refresh time. (for me details and solution, please check my old post).

Steps to rename Uplink

Login to Web Client,

After you rename your Uplink, you can start to create vMotion group for vDS.

Create vMotion for vDS

The funny thing for this step is you have to create a vDS port group first before you can do anything lese.

Now, you can create a new Uplink for vMotion

I skip the rest of parts.

Tricks:

I don’t think you can vMotion between vss and vDS. You can only vmotion between same type of vSwitch. Although you can migrate vms from VSS to vDS with few ping drops.

Assign specific vmnic to Uplink

One thing you would like to do is to assign vmnic01 (for example) to a specific Uplink. Please follow these steps.

Add Physical adapters into vDS via web client

change Auto-assign to a specific Uplink

Delete a Uplink (not physical nic connection)

The simple thing I want to do is to remove one of Uplinks. It’s virtual Uplink on vSwitch, it is NOT the physical nic which I connect to Uplink. but this very simple thing almost can’t be done via either vSphere Client or Web client.

To give you a better understanding, a new vDS coming with 4 Uplinks connecting with nothing. What happen if I add more uplinks now and want to remove some Uplinks latter?

The way you add more Uplink is here

Unfortunately, the only way to remove Uplink is either rebuild a new vDS or migrate all your VMs to other switch and remove all physical host nic connection to Uplink and go back to here and to set a LOWER number!

If you set this number to 3, 2 uplinks will disappear but it won’t let you choose which 2 uplinks. Therefore, you better move all VMs and connections between physical host nics to Uplink before you remove Uplink.

This is not just my conclusion, a Vmware Support Engineer was on the phone 1 hour with me and come up with this solution. Maybe there is another way to do it, but we are not able to find out. If you know how to do it, please let me know or leave it in comment.

Conclusion:

There are still lots testing we can do with vDS, but at this stage, I definitely wouldn’t recommend to ditch vSS and use vDS solely. A hybrid environment is what I would recommend.

vSphere 5.1 syslog collector sum up

It’s shame that it took me 2 hours to find out why my Syslog Collector is not working. But I would like to share my experience with everyone including how to debug it.

Syslog collector has two parts.

Part running on vCenter

 

Syslog Collector must be installed first.

It is very important to configure your firewall so your syslog can go through.

the Syslog collector can use 3 different protocol. TCP,UDP,SSL. You can enable all of them.

make sure you have space for this log collector

that’s will install plug-in directly into your vCenter.

Feel free to use your DOMAIN/SERVICE_ACCOUNT to replace local administrator. But you need to make sure that service_account has local admin rights first.

By using different account will make better view in the TASK Manager to see how much memory it consumes.

 

You can replace SSL certificate with local CA certificate if you really want.

Then you can finish installation.

You will see it in your service.

You will see it in your task manager.

Parts you need to configure on ESXi host

 

As ESXi host, you need to configure it little bit more than just PDF file tells you.

You need to configure ESXi Firewall to open the port (which I didn’t. –_-b)

After that, the easiest way to configure is to use vSphere client (not web client).

You can use either tcp://servername:514 or tcp://serverIP:514 or other protocols

Once it’s done, you should have a new folder under your Syslog collector folder immediately without any other actions.

Debug Procedure:

 

Debug from vCenter

You need to check out whether syslog collector service is up

You need to check out whether Syslog appears in Task Manager

Use telnet to check tcp port to see whether port is open / listening

If you want to test UDP port, you can use Microsoft tools PortQryUI to do it. You can find it at this link.

http://www.microsoft.com/en-us/download/details.aspx?id=24009

Debug from ESXi host

Check the firewall and make sure port is open

Use this command on console to check the setting in esxi

Use this command to reload esxi syslog

esxcli system syslog reload

Use this command to test esxi syslog

esxcli system coredump network check

If it is successful, you should see something like this

Verified the configured netdump server is running

You can also use esxi console to configure rather use vCenter

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2003322

The IP is the syslog collector IP address.

The VMA command is little bit different since you normally need to authenticate yourself, hence I won’t list here.

I think that’s everything about Syslog. Please let me know if you have questions.

vSphere 5.1.x vSphere Web Client cannot connect to vSphere Server

Symptom:

 

Well, essentially, this issue exists in both vSphere 5.1 and vSphere 5.1a. No matter what I did to install either vsphere 5.1 or 5.1a, I always not able to use web client to connect to vCenter.

All what you got is this error.

Could not connect to one or more vCenter Server Systems:

https://domain.com.au:443/sdk

 

This is really driving me crazy. The vSphere Web Client should connect to vcenterServer.domain.com.au rather than just domain.com.au.

I have tried many things via vSphere Web Client and none of them actually fixed issue.

Now, Let’s see the root of this issue:

 

Cause:

The root cause is when you install your vCenter Server, the wizard gives you misguide and you input wrong value.

When you install vCenter server, you are required to input a service account in terms of running vCenter Service. Here is what I went wrong.

If you see the account name, it only indicates a simple user name. Clearly, I wanted to use a domain service account to run this service. With this picture, I thought the system is providing a local user on the vCenter server which I don’t want.

so I modified that FQDN by using domain.com.au so I thought I could use domain account rather than local account.

But I was wrong since that FQDN is actually vCenter server and has nothing to do with account name and account password.

 

Solution:

You should reinstall vCenter Server if you do have this issue. You can’t change FQDN of vCenter in look up service easily (at least, I didn’t see any public docs). A reinstall vCenter should fix the issue, but if that doesn’t work, you have to reinstall SSO, inventory service and vCenter server.

 

The interesting thing about that error is once you leave that account, and type password and FQDN, then you click Next and Click Previous to go back and check settings, the simple username becomes domain\username!!

 

 

Other information:

Do not login Web client with admin@system-domain because that SSO admin and it has no rights on vCenter server to see the content.

 

Please let me know if you have more questions.

Disable UCS VIC failover when you use Vmware

What is UCS VIC failover.

Put it into a simple way, each blade can have a VIC card. Each VIC card has 2 10gbit/s ports like the one we are using, CISCO UCS M81KR.

This VIC card will handle all network/SAN traffic from this blade to both IOMs. When there is outage on one path of uplevel, VIC can automatically redirect traffic to another working interface without outage.

For more details, please refer to reference document.

Why we need to disable UCS VIC failover.

According to UCS design document,

All Connectivity May Be Lost During Upgrades if vNIC Failover and NIC Teaming Are Both Enabled All connectivity may be lost during firmware upgrades if you have configured both Enable Failover on one or more vNICs and you have also configured NIC teaming/bonding at the host operating system level. Please design for vailability by using one or the other method, but never both.
To determine whether you have enabled failover for one or more vNICs in a Cisco UCS domain, verify the configuration of the vNICs within each service profile associated with a server. For more information, see the Cisco UCS Manager configuration guide for the release that you are running.

 

UCS VIC failover will have MAC conflict with Host level Nic teaming including Vmware vNic Teaming.

Comparing two solutions of nic teaming failover, Vmware nic Teaming is also providing network load balance and much more controlling over Cisco VIC failover. Hence, we need to disable VIC failover.

How to disable VIC failover

If really depends how you setup your system. In my UCS, I have deployed NIC template and therefore, I will need to modify nic template first.

Notice the nic template type is Updating Template even when service profile template is Initial template, it means the change I will make (untick the Enable Failover) will be push to blade immediately.

The good thing is we have setup our reboot policy ask “User Ask”, so UCS will reboot blade immediately. Instead, it will put request into pending Activity list for approve.

 

Change failover procedure

Now, you will be able to schedule to reboot your blade.

 

 

 

 

 

 

Reference:

http://www.cisco.com/en/US/prod/collateral/ps10265/ps10276/solution_overview_c22-555987_ps10280_Product_Solution_Overview.html

Cisco UCS B series firmware upgrade from 2.0(2q) to 2.0(4a)

 

Why do we upgrade UCS firmware

This is a post which describes upgrade Cisco UCS B series firmware upgrade from 2.0(2q) to 2.0(4a). The reason for this upgrade is simple. A bug.

There is a Cisco Bug in the system which prevent show tech to be generated. Without show tech file, I’m not able to diagnosis any issues. So it has been more and more critical for us.

According to Cisco, 2.0(4a) has fixed this issue. I have attached the pdf in the reference, so you will be able to download and take a look. Basically, the real upgrade is pretty close to this document with minor twist.

 

Download firmware

There is no drama here. Just log in Cisco.com with your cisco account, and follow instruction on document so you will be able to download the bundle file.

In my case, I only have UCS B series, so I only downloaded two files.

 

Preparation

There ain’t much to do with preparation. My personal suggestion is:

make sure you have enough space on bootflash

Then, you can upload those two files into system easily from local Server.

Backup your current configuration.

You need to make sure you have filename written in the field otherwise it may not able to backup configuration.

 

Create Host Firmware Package

This package will delivery quite few firmware updates and will only be deployed to service Profile. In another word, your server must associate with service profile in terms of getting those firmware.

Now, with different environment, firmware package can contains different components.

Adapter:

In our system, UCS blade has one DCE which is M81KR. However, I didn’t include adapter firmware in the package according to PDF doc. But Cisco tech support said I should include it in the firmware.

BIOS:

BIOS is a must.

Storage Controller:

Because we use RAID-1 local disk for OS. so we need to upgrade that as well.

Board Controller:

Comparing with package version, there is no new version. so we don’t need to upgrade this one.

 

Disable Call Home Service

 

 

Update Firmware for Adapters, CIMCs,IOMs

Update firmware is just to load new version to backup Version slot. The new version will kick in as start up version once you restart components.

For just Update firmware, you can select ALL, it will not cause any harm.

 

 

Activating firmware on adapters and CIMCs

You need to do these steps in order. You can’t select adapters and select CIMCs settings and hope to click ok to apply both components at once. It will cause issue. If you somehow did select both, click Cancel.

DO NOT select ALL in the filter to activate everything in once!!

Activate firmware for Adapter.

Notice Active status is Pending Next boot

 

Activate CIMCs

CIMCs is separate component from data. so It will restart itself but no disruption for production data.

CIMCs will become 2.0(4a)

Activating UCS Manager Software

This will cause console,KVM to restart. No data disruption as well.

Activating IOM

IOM is important module and will cause data disruption. so this module will reboot when you reboot FI. If you have 2  FI as redundant, you can reboot one FI at a time. When you reboot FI-A, IOM-A will reboot as well. Therefore, we will only load new version to Startup version and wait for reboot.

 

Activate Fabric Interconnector Firmware

With fabric Interconnector, we need to identify which one is subordinator. We will update subordinate first, then switch role to new FI with primary and update another FI. You need to make sure your redundant system is working otherwise, you will experience downtime on blades.

In my personal experience, you can actually give FI (subordinator) a reboot before you update firmware so it will clean up lots of stuck issue and processes.

 

 

If FI come up with status like that, it means it’s all good for update another FI.

check all connections including network and VIFs

essentially, if you see connections on both FI-A and FI-B, then it means it is right. Just be aware that some command line has changed once you upgrade your version of UCS Manager.

You will do the same step for the other FI but remember to switch other FI to become subordinator first.

Update blade BIOS, SLI logic controller, and others

This is the last step. Before you do anything, you need to make sure you have management policy setup correctly like this.

 

then you need to make sure your host firmware packages is attached with template or service profile.

Once you made change, it should pop up to reboot or not.

 

Choose No to reboot at your own time.

 

Thank you for reading. Hope it helps

 

Reference:

http://www.cisco.com/en/US/docs/unified_computing/ucs/sw/upgrading/from2.0/to2.0MR/b_UpgradingCiscoUCSFrom2.0To2.0MR.pdf

Cisco UCS Design problem with Vmware

Recentely, we have finally got upgrade to new environment which is Cisco UCS 2.0. We are all excited with new toy but we ran into some design issue s which I would like to record here so you can avoid it in the future.

 

FC Uplink needs to be at right ports.

I think this is basic common knowledge but  clearly, we don’t know. With Fabric Interconnector, we need to configure FC port to connect to Uplink FC switch. At first, we put FC ports in the middle of switch and put Ethernet Uplink at the end of switch (like port 31/32). Then, we realize it’s not doable once we get into configuration.

click that to get into FC port configuration

Click Yes

We put FC link in the middle, which is wrong. Ethernet port at the end. As you can see, there is slide bar to slide to configure. Once you slide, you will see this.

All ports on right side of bar will be FC ports. So you can either put FC to expansion model or you have to change your ports.

 

UCS Memory is bigger than your hard disk

Well, this actually sounds ridiculous. But it’s one of reasons why we bought UCS. Our Blade has 196GB memory and we will put Vmware on them. We also bought 100GB SSD to increase swap file speed. Unfortunately, at that time we purchased, we didn’t realize that to put swap file of vms on local disk, we need at least same size 196GB as memory so vm swap file can use local disk rather than precious of SAN storage. Even with new vSphere 5 feature (Swap host cache in SSD), that function won’t help much only we have memory contention. So if we balance it out, we should buy some big size of SAS to cover that.

vMotion is No!

Well, maybe it’s just me that I’m get used to always vMotion everywhere. Once I installed new blade and join them to our vCenter. I tried to offload my vms to new host. Then, I got this error.

Of course, what you need to do is to turn it off and migrate. But then, that’ s outage or you have to EVC.

All those errors can be avoid easily but it’s matter of experience, I guess. Hope it helps.

 

How about that? Virtual Cloud Day has changed date!

Well, if you are like me, you probably still receive email from Vmware and said Virtual Cloud day is open at 24th May.

 
But once you actually jump on Vmware site, it’s changed to July.

 

It makes me thinking why it changed? Is that due to new Hyper-v 3 and SCVMM 2012 SP1 totally changed the Virtualization and Vmware are unprepared? ho ho

 

Hyper-v 3 Part 1 Storage vMotion

Well, with my last post, I was not able to join Hyper-v 3 to SCVMM 2012 RC. Therefore, most of labs can’t be done. After I cried loud, our MS TAM help me to find a new copy of SCVMM 2012 CTP. This version is designed for Windows 8 and all functions are in Go status!!

With this exciting news, I managed myself (thanks a lot to my team mates!) to get 3 HP G5 servers to build a test environment today. Just like all test labs, I spent lots of time to work hard to find solutions and I hope my post will save your time and let’s uncover Hyper-v 3 true power together!!

Again, if I made any mistakes, please feel free to point out.

Introduce my Test Lab:

My test lab involves 3 physical hosts in the same subnet. There is no FC SANs available nor NAS exists. Each Hosts has it’s own local disks. I installed Windows 8 Beta on all 3 servers since this is one of conditions to do labs. 2 Servers I have enabled with Hyper-v 3 and I installed SCVMM 2012 CTP on the last server.

Note:

There is no issue of installing Windows 8. But with SCVMM, you need to install AIX and .NetFramework 3.5. You can’t tick the .NetFramework inside of W8, that would fail with not able to download. What you can do is following:

And restart server afterwards

Hyper-v 3 Storage vMotion is quite good

The one thing Hyper-v R2 can’t do is storage vMotion. Without this feature, I’m not able to migrate a live VM from one share LUN to another shared LUN without powering off VM. In the Hyper-v 3, MS not only managed to do a share storage vMotion in a cluster with SAN, you can storage vMotion between two standalone Hosts with a windows 8 share folder (SMB 2.2) !! (No more expensive VSA?).

Create a share folder

Live Migration is just normal vMotion with disk files sitting on the share storage. In this case, it’s a windows 8 share folder.

Since I have only local disks, I want to have some shared storages as the part of habit I got from Vmware. With Hyper-v 3, all what you need is a Windows 8 server and create a traditional share.

That’s it. No drama. No need for any specific share rights. This share purpose is to enable you to see this share when you add it into VMM. Once it is added, VMM will modify the accessing rights.

There are 2 steps to add this Share resource. The first steps is to add it to VMM library. so all VMs will sitting on that. You also need to add share resource to individual hosts so VMM will add hosts ID to security list.

Add to VMM Library

Add to individual host

Once it is added, your host is able to migrate VMs to share library.

This is how VMM setup access rights on the folder.

Storage vMotion (between host and share folder)

Basically, you can migrate VM files from local disk of your host to a share folder and from share folder to share folder or move it back to local disk.

According to my calculation, there is no ping drop during moving. The ping value sometimes is increase from 2ms to 65ms.

Storage vMotion from host to host

It’s interesting that MS say Hyper-v 3 is able to storage vMotion between two standalone host but they recommend to do it with a share folder accessable by both hosts. So does that mean you can Storage vMotion + vMotion from Stand alone host to stand alone host?

From my test result, It actually tried but failed at 81%. What it happened was Storage vMotion kicked in and copied 9GB data from one host to another host. During the copy procedure, the ping value jump up to 600ms. But the file is copied completely. Then, ping drop back to 2ms and tried to translate the last bit in the memory and that’s where it failed.

Overall, Hyper-v 3 can do Storage vMotion and with at least 2 sessions at same time. The ping value jump up can be contributed with my 100Mbit/s network limitation. Hyper-v 3 can use a normal windows 8 share folder as storage space which saves tons of money. so no more NAS and VSA? The new Windows 8 vhdx supports 64TB so there is no worries on how big a single file can be.

Bugs I have found:

1.With Windows 8 beta, RDP connection seems dropped if I started to copy large files via network. The whole server actually seems to be frozen status until copy is finished. My guess is that W8 didn’t prioritize RDP over copy session.

2.When I storage vMotion a VM which use differencing disk, the original disk (or parent disk) are not moved with storage vMotion.