Some questions from the Benchmarking Guide.

[netyale]

Hi folks,

I'm using ESX 5.1 now

I have some questions about the VMmark test!

1.

Is the disk space for VM important??

Because I found for the client template.

The space is running out after I install the cygwin

I'm using Windows 2008, the default 32 GB is not enough

Same thing happened on standby VM

The default space is not enough after Windows Update

2.

When asked to Specify the virtual disk size and provisioning policy, enter either 15GB(for you will be

using Windows Server 2003) or 32GB(for you will be using Windows Server 2008), leave the Disk

Provisioningoptions unchecked

For this disk provisioning, I have to choose one from three methods, I cannot leave it uncheckd

Which one I should check??

Is there any document with installation photos to follow??

Thanks a lot...

Alvin

[Rebecca Yo]

Hi Alvin,

1. The disk size of Standby must stay the same as described in the benchmarking guide. Whether you can create a larger disk size for the client depends on whether you intend to publish these VMmark results. If you intend to publish, you must follow the configuration outlined in the VMmark Benchmarking Guide, including the disk size. If you do not intend to publish them (such as for internal use only), then it is fine to increase the size of the client disk size.  If all Windows updates do not fit in the disk space size, this is okay, just update as much as you can. Both Client and Standby should typically be on a private network anyway.

2. You are right, there are no options for you to check there. This wording may need to be fixed in the Benchmarking guide. Please select "thick provisioned eager zeroed". This is consistent with all the other VMs in VMmark.

No, there are no images accompanying the installation of VMmark. Instead, we are very specific in our installation instructions. As you go through the VMmark installation, please follow each instruction completely and exactly.

Thanks,
Rebecca

[netyale]

Hi Rebecca,

Thanks for your quick reply!

For the cygwin installation

Do we need to install all the packages??

In the guide it shows we need to install all

But the whole package is so large I guess it's more than 10G

Thanks!

Alvin

[Rebecca Yo]

Hi Alvin, I have slightly modified my response above. For consistency, I would recommend choosing "thick provisioned eager zeroed" but it is not a hard and fast rule. Also if you are intending to publish your VMmark results, the size of the Client VM disk needs to remain 32GB.

Yes, you need to install all the packages. The cygwin installation is quite large. It's one of the longest steps in setting up the client.

Thanks,
Rebecca

[netyale]

Hi Rebecca,

I will choose "Thick Provision Eager Zeroed" for all the VMmark test.

The question is

If I set the Client VM disk size to 32GB

It's not possible to fit in the cygwin

I will use a physical server for the prime client

So the disk size doesn't need to fix to 32GB, right?

So you are talking about VM client0 or client1, right?

Do I need to delete anything to make space to install cygwin??

Thanks for your help!!

Alvin

[netyale]

Hi Rebecca,

Sorry to ask more detail questions

I want to make the test as perfect as possible.

According to the Guide p.80

Configure the Windows Server 2008 Firewall to Allow STAFProc.exe to Pass Through

Can I just disable the firewall?

About the STAF software

The current version is 3.4.14

It will launch the startSTAFProc.bat automatically with the system

In the msconfig Startup tab we can also see the setting

Do I need to follow the Configure STAF to Start Automatically on p.80?

Thanks again!

Alvin

[Rebecca Yo]

Hi Alvin, it's ok, I don't mind answering detailed questions.

Yes, we are talking about client0, client1, ...clientN. Technically, to publish VMmark results (meaning you submit the results to the VMmark Review Panel for publication on VMmark.com), you must configure all VMs as specified in the VMmark Benchmarking Guide, so the client disk should be 32GB.

If you are not planning to submit results to the VMmark Review Panel, you can size the client VM as needed.

You should be able to install all packages in Cygwin on the client, if you do not install all of the Windows updates. Please let me know if this is not the case.

Correct, if you are using a physical server as client, then the disk size can be any size you wish.

Yes, you can disable the firewall. Of course you should only do this if you are using a private subnet.

Yes, please follow instructions in the VMmark Benchmarking Guide under "Configure STAF to start automatically".

Thanks,

Rebecca

[netyale]

Hi Rebecca,

1.

For the space of the 32GB VM, I will try to install important updates only.

Question: Can I just ignore all the Windows update thing?? Is it gonna change the test result??

2.

For the Configure STAF to Start Automatically thing

After installing the version 3.4.14

It will automatically start it after reboot

The msconfig Startup tab will add an item to launch STAF

Do I need to use Scheduled Tasks to make it start twice??

3.

I tried to choose Download without Installing during the installation of cygwin

Two different mirrors got two different sizes of source file

mirrors.kernel.org

5.53GB

mirror.mcs.anl.gov

2.29GB

Huge difference

Do you have recommended mirror site??

Thank you so much for your help!!  :smileyhappy:

Alvin

[netyale]

Hi Rebecca,

On p.73, The prime client must run Windows Server 2008 (32-bit or 64-bit)

I installed Windows Server 2008 64-bit

And on p.85 Configure the Prime Client for the VMmark2 PowerCLI Service

I tried to run VMmark2-RMQ-Installer-Step2.bat

But it shows the following message

Do you have any suggestions for this problem??

I stuck here for my Prime Client

Thanks

Alvin

[Rebecca Yo]

Hi Alvin,

1. I really can't speak to whether the windows updates would change the test result or not. However, different benchmark submissions have updated their Windows VMs on different dates, and we consider this to be performance equivalent.

Make sure you follow instructions regarding Windows versions and service packs as outlined in the VMmark Benchmarking Guide, for example:

For clients, if you are installing Windows Server 2003, make sure you install 32 bit at least Service Pack 2 on the VM. If you are installing Windows Server 2008, you can install 32 or 64 bit. The prime client requires Windows Server 2008.

For Standby VM, it must be updated to Window Server 2003 Enterprise edition 32 bit, Service pack 1.

2. I understand, but still recommend configuring the Scheduled Tasks for STAF for each client as directed in the VMmark Benchmarking Guide. If you are intending to publish your VMmark results, this is a required step (you must follow all configuration directions as described in the Guide). Moreover, I can't vouch as to whether using the startup tab of STAF is equivalent to the scheduled task or not but I feel it is not.

I can say that when you set a Scheduled Task to start STAF,

  A) The STAF process needs to run under a certain user account which you specify during the scheduled task setup.

  B) On the prime client, STAF needs to start at logon, while on the other clients STAF needs to start on boot (before any user logs on).

I recommend deleting any startup tab STAF task and configuring a Scheduled Task as described in the VMmark Benchmarking Guide.

3. I have not seen such a size difference in mirrors before. I would recommend choosing the larger one, in case the smaller one is incomplete for whatever reason. We do not have recommended mirror sites.

4. Regarding your error message at VMmark2-RMQ-Installer-Step2.bat, this error is covered in the VMmark Benchmarking Guide Troubleshooting section, "Error Installing RMQ".

Thanks,

Rebecca

[netyale]

Hi Rebecca,

About the answer 2

The STAF will show a DOS window when it's running, right?

The default installation of STAF will show the DOS window

Once I follow the document it won't show the Dos window

It's not correct, right??

I should see the Dos window when STAF is running, right?

Any suggestion?

Thanks a lot..

Alvin

[Rebecca Yo]

STAF will only show a DOS window when it's started at logon.

On the Prime Client, the STAF window will show, because on the Prime Client, we set STAF to start in the foreground, at Log on.

On all other clients, the STAF window will not be visible. On the other clients, STAF starts on boot, so it is in the 'background' (no visible window). However, you can confirm that STAF is running on the clients by pinging them from the prime client: "staf clientN ping ping" where clientN resolves to the IP address of the client in question. Or, from the client console, run "staf local ping ping". The correct response is "PONG".

If you are on the Prime Client and you do not see the STAF window, you may have accidentally set the Prime Client STAF to start on boot.  In the VMmark Benchmarking Guide, see the note: "For the prime client, select the When I log on radio button instead of the When my computer starts radio button."

[netyale]

Hi Rebecca,

Thanks for your reply.

I'm still working on the test

On p.91

There is a script set_cygserver.sh

But after I execute it

There is no difference between the original and the after

What does it should change?

Thanks

Alvin

[Rebecca Yo]

The script automatically uncomments all values in cygserver.conf, lines like for example kern.srv.cleanup_threads 2.

In the files you showed, the values were already uncommented so there's no change. Your cygserver.conf is correct. :smileyhappy:

[netyale]

Hi Rebecca,

Happy Friday!

I already installed the Exchange Management Tools on my client0

But on p.99

I cannot find Microsoft Exchange System Attendant in my Services

Is it come with Management Tools??

Thanks

Alvin

[Rebecca Yo]

Hi Alvin,

The Exchange System Attendant runs on the Mailserver VM, not the client. So please configure the Exchange System Attendant to Automatically Restart on the Mailserver, rather than the client.  This may need to be clarified in the Benchmarking Guide.

The client does run Microsoft Exchange Load Generator Remote Agent, after you install LoadGen.

[netyale]

Hi Rebecca,

Thanks for your reply.

It really helps me a lot

More question is coming

I tried to setup the LoadGen

But I got this error message

Is it something wrong with my setting??

Thanks

Alvin

[Rebecca Yo]

Hi Alvin, it looks like you resolved this by reinstalling the Exchange Server Management tool on the client as seen here, correct?

[netyale]

I have two problems but I fix it already!!

Thanks...

[netyale]

Hi Rebecca,

New problem...:smileysad:

Is there something wrong in the script??

Thanks for your help!!

Alvin

[Rebecca Yo]

Did you unblock VMmarkPScli.ps1?

1 On the Prime Client, navigate to the VMmark2\tools directory.

2 Right-click the VMmarkPScli.ps1 file and select Properties.

3 At the VMmarkPScli.ps1 Properties window, under the General tab, click the Unblock button.

4 Close the VMmarkPScli.ps1 Properties window.

[netyale]

Hi Rebecca,

It works!! Great.

BUT

I stuck again

Issue 1:

Error: VMmarkRMQmgr Unable to Clear Queues

And I also stuck in the middle of the test

Please give me some suggestion about my incorrect settings

Thanks a lot!!  :smileycry:

Alvin

[Rebecca Yo]

Hi Alvin, glad to hear that cleared up your earlier error.

Your file RMQmgr-out.txt shows the error "The system cannot find the path specified". This is probably due to a missing or misnamed file or directory.

Is this file present on your prime client?

C:\VMmark2\tools\VMmark2-RMQmgr.pl

Also, is this exactly the description of the run that you specified? "1-tile-test" If your run description includes illegal characters (e.g. \ / : * ? | > < ) the illegal characters may not show up in the run results folder, but can still create problems for the run and can produce the above error.

Otherwise make sure that you have directory C:\VMmark2\ which should contain directories results, tmp, tools, xml and files tilescore.pl and VMMARK2.CONFIG.

[netyale]

Hi Rebecca,

Yes. I checked my system

C:\VMmark2\tools\VMmark2-RMQmgr.pl

is on my prime client

But I fix it by reinstall the whole prime client system.

Great.

BUT something new happened!!

I have a question

In the VMMARK2.CONFIG

For example PTD=1

Does it mean I disable the PTD function?

I already marked all the PTD related lines in the CONFIG

But STAX still ask me for the PTD client host

Thanks for your help!!

Alvin

[Rebecca Yo]

Ok, instead of commenting out the line PTD=0, uncomment it. PTD=0 means that PTD is not running.

[netyale]

The config file confused me..

PTD=0/1: Valid values [0/1] Enable/Disable

0: Enable

1: Disable

Thanks..

Alvin

[Rebecca Yo]

Good point, I can see how that is confusing. We will fix that.

[netyale]

Hi Rebecca,

For the TurboMode I got some errors also.

Error 1

LUNNOTFOUND on 0 Hosts in the cluster and/or 1 VMs are already on the TargetLUN

Error 2

Error with Storage vMotion LUN Check

Error 3

Exchange mailserver: stop mail service failed to start/complete. Returned: RC=2

Error 4

svmotion failed Setup Phase

I think my exchange and storage related setting is not correct but don't know why!!

I setup an iSCSI storage for all the VM workloads

And the Identification in the Configuration - Storage is "Storage"

So I put it in the VMMARK2.CONFIG

Deploy/DeployLUNs="Storage datastore1"

Is it correct??

And for the SVmotion, I put it twice

as the following

SVmotion/TargetLUNs="Storage Storage"

Is it correct??

Thanks for your help!!

Alvin

[Rebecca Yo]

Hi Alvin, this is good. You are progressing further in the VMmark run, and the harness is identifying configuration errors you must fix before you can start the run.

1) Storage vMotion RETURNSTRING = ERROR : LUNNOTFOUND on 0 Hosts in the cluster and/or 1 VMs are already on the TargetLUN

If you look in the file sVMotion-TargetLUN0-Poll.txt, you can see

  Storage LUN found on VMHost 172.16.165.135

  Checking 'standby0' : Error ! : Found on TargetLUN

In VMMARK2.CONFIG you specified

SVmotion/TargetLUNs="Storage Storage"

Standby0 is the 'target' VM for SVMotion. It gets SVMotioned to "Storage" LUN and back during the VMmark run. At the beginning of the run it must be located on a different LUN.

From the Benchmarking Guide:

NOTE In order to provide source and target datastores for the storage relocation operations that are part of the benchmark, VMmark requires a minimum of two datastore partitions.

Additionally, the benchmark requires that all ESX/ESXi hosts used in a test have access to the same shared storage.

It looks like you have only one datastore, "Storage", so please repartition Storage. I recommend making the second datastore rather small (for SVMotion and Deploy operations only) and leaving most of the capacity/performance for the datastore on which the tile VMs are located.

2) This is the same as #1.

3) Process: Tile 0: Exchange mailserver: stop mail service failed to start/complete. Returned: RC = 2, STAFResult = None   

Sometimes on the Mailserver, the Microsoft Exchange Information Store service does not start automatically even when it is supposed to. But it did start manually. I think this is fine, we will come back to it later if not.

4) svmotion failed setup phase: This is the same as #1.

5) Warning: Cluster VM Mapping And Resources returned Warnings : InvalidResources 0

Please power off the VM StandbyN before you begin the run.

6) Also you have only 1 host in your VMmark cluster. If you want to run the vMotion workload, your cluster must include at least two hosts with vMotion capability. Are you planning to run with only 1 host? It's possible but it will not be a complete run.

7) You need to load the DS2 Database before you start the run. See "”Unknown database” Message in ConfigDS2DB Text File" in the Benchmarking Guide.

Thanks,
Rebecca

[Rebecca Yo]

When you create a second partition, suppose you name it "Storage2". All your VMmark tile VMs will be located on "Storage".

You will put in your VMMARK2.CONFIG:

Deploy/DeployLUNs="Storage2 Storage2"

SVmotion/TargetLUNs="Storage2 Storage2"

[netyale]

Hi Rebecca,

Thanks for your reply. Very useful information. :smileylaugh:

Where can I put the new partition for Storage 2??

Should I put on the same iSCSI storage server?

Just reduce the original Storage and create some space for Storage 2??

Any recommended size for Storage 2??

And could you please explain the meaning of the

Deploy/DeployLUNs="Storage2 Storage2"

and

SVmotion/TargetLUNs="Storage2 Storage2"

What's the operation it will do during the test??

Thank you very very much!!

Alvin

[Rebecca Yo]

If you have another datastore besides the iSCSI storage server, you can use that. Or yes, you can put it on the same iSCSI storage, in which case you would reduce the original datastore size and use the remaining space to make a second datastore. The second datastore could be 100GB or less I would say.

Keep in mind that VMmark has quite intensive storage IO requirements, see this post. Those requirements refer to the datastore hosting the tile VMs.

Deploy/DeployLUNs="Storage2 Storage2"

and

SVmotion/TargetLUNs="Storage2 Storage2"

means that Deploy operation will clone a new VM onto datastore "Storage2" and that SVmotion operation will SVmotion Standby0 onto datastore "Storage2" and then back to its original datastore.

[netyale]

Hi Rebecca,

So I can use the datastore on ESX for ClientN for test purpose, right??

For

Deploy/DeployLUNs="Storage2 Storage2"

Does it mean the test will deploy to Storage2 twice??

Thanks!

Alvin

[Rebecca Yo]

If you are referring to the ESXi host's local datastore, yes you can use it to host some other non-VMmark VM "clientN" but that VM should not be powered on or active during a VMmark run. All active client VMs need to be located on a different host than the host with the VMmark tile VMs.

Also if you want to be able to run all workloads, including vMotion, you will need at least two ESXi hosts dedicated to the VMmark tile VMs. Do you want to run all workloads?

The reason for listing Storage2 twice is because the number of concurrent infrastructure operations in VMmark increases as the number of tiles / hosts increases. However, you will only get 2 or more concurrent operations if you have 4 or more hosts and 4 or more tiles.

[netyale]

Hi Rebecca,

I refer to the Storage2. Can I put it on the ClientN ESXi host??

I want to run all workloads. But I need to make sure the setting for my Tile0 is perfectly correct first.

Can I test run the Tile0 only first??

Do I need to modify something to run the single tile test??

Thanks

Alvin

[Rebecca Yo]

Yes, you can make datastore Storage2 available to your client ESXi host. Ideally you would use a separate datastore for clients vs VMmark tiles, but it is not required.

To run VMmark on a single host, change the entries in VMMARK2.CONFIG to

INFRASTRUCTURELIST="Deploy SVmotion"

(this turns off the vMotion workload) and

DoClusterMapping=0

You can troubleshoot the other workloads in the meantime until they are all running.

However, you will not be able to get a Score_1_Tile_Test.out file (run results summary) until all workloads are enabled and there are at least 2 hosts in the VMmark cluster.

Thanks,

Rebecca

[netyale]

Hi Rebecca,

You need to load the DS2 Database before you start the run. See "”Unknown database” Message in ConfigDS2DB Text File" in the Benchmarking Guide.

Do I need to do this even if I import the OVF from the VMmark website??

So what I should do is in the section

Prepare the DVD Store 2 Database

Is it correct?/

Thanks for your help!!

Alvin

[Rebecca Yo]

Yes, even if you import the DS2 VM, you must prepare the DVD Store 2 database. The database does not come preloaded in the ovf, please follow all instructions carefully in the Benchmarking guide. The guide DOES direct you to prepare the DVD Store database even if you used the OVF.

[netyale]

Hi Rebecca,

So I need to start to prepare the DS2 database from installing the MySQL??

I tried to run the 1 tile test yesterday

It seems work without any issues

So Happy! :smileylaugh:

But I didn't prepare the DS2 database yesterday

Why the Harness didn't show this issue??

And I can start to build my second tile, right?

Please help me to check my result files

Thanks

Alvin

[Rebecca Yo]

Hi Alvin,

DS2 is not working correctly because the database is not initialized. The harness can't tell whether the DVD Store database has been initialized or not.

You don't need to install mySQL. Follow the directions under
"”Unknown database” Message in ConfigDS2DB Text File"

in the Benchmarking guide.

Olio is also not working correctly yet. Is this the correct IP address for OlioDB0?

192.168.1.10

If not, you will need to update the hosts file on OlioWeb0.

After you get these issues resolved, then you can start cloning out your second tile.

Thanks,

Rebecca

[netyale]

Hi Rebecca,

Long time no see!!

I'm back!!

I check everything and setup 2 tiles for my testing

And here is the result without any error

But I still cannot get the score in the Score_2_Tile_Test.out

Could you please help me to check it??

Thanks for your help!!

Alvin

[Rebecca Yo]

Hi Alvin, welcome back.

To find whether a workload is working correctly or not, look in Score_2_Tile_Test.out in the table "Tile_N_Ratios:" where N is the tile number.

TILE_1_Ratios:  mailserver        olio   dvdstoreA   dvdstoreB   dvdstoreC     standby    Geo.Mean

p0                    0.00        0.99        0.46        0.43        0.42        1.00        0.00

p1                    0.00        0.97        0.49        0.45        0.42        1.00        0.00

p2                    0.00        0.94        0.47        0.49        0.51        1.00        0.00

A score of 1.00 would be a ballpark 'normal' score. Ratio is a workload's normalized throughput. P0 is the first third of the run, p1 the middle third, p2 the last third of the run. As you can see, Olio is around 1, which is good. Standby is also working correctly. DVDStoreA, B C are functioning but at below expected levels (~0.45) Mailserver works briefly at the beginning, but then becomes overloaded with queued requests and shuts down.

In my opinion, these results are because you are facing a shortage of some resource on your VMmark testbed, probably slow storage. For example, your SVMotion ratio is .56, meaning that your SVMotion only completes about half as many operations as you would see on a 'typical' VMmark run.  A frequent cause of these low scores is a storage bottleneck, see here for a description of VMmark storage requirements. I would say a bare minimum is a 7-spindle RAID0 LUN for 1 tile of VMmark VMs, plus one additional LUN as an infrastructure operations target.

How many spindles make up your Storage LUN? And what RAID level are you using for that LUN?

Try running only 1 tile and see if your scores improve.

Thanks,

Rebecca

[netyale]

Hi Rebecca,

I just change the storage for my iSCSI

I'm using Fusion-io iScale and 10G NIC as the Tile storage and vmotion storage

It's really looks better than yesterday

(I was using only 2 hd and RAID 0 storage)

But I still see some issues in the result files

What else I can do to improve the result??

And what is the result for the following?

Unreviewed_VMmark2_Applications_Score   :     2.83

Unreviewed_VMmark2_Infrastructure_Score :     0.00

Unreviewed_VMmark2_Score                :     0.00

Is it means the result of my system is 2.83 tiles?

By the way,

The setting for the VMmark seems correct, right??

Thank you for your help!!!!! :smileylaugh:

Alvin

[netyale]

Hi Rebecca,

I also tried to reinstall the client system

But it stuck at the following issue

Checking client0 Time Sync failed to start/complete.

Do you have any suggestion??

I'm sure all the systems time are synced.

THnaks

Alvin

[Rebecca Yo]

The reason you are getting a VMmark score of 0 is that your Deploy is failing. If one workload fails, the entire score is 0.

To troubleshoot Deploy, try a 'manual Deploy' operation as described here.

In regards to the time sync error, it is because perl is not operating correctly. You can see in the log

'perl' is not recognized as an internal or external command,  operable program or batch file.

Check your perl installation by running "perl -v" in a prime client cygwin window, which needs to print perl version. Did you set the environment variable for cygwin? Do you have the correct bitness cygwin install for your prime client operating system?

Thanks,

Rebecca

[netyale]

Hi Rebecca,

I checked my prime client, client 0 and client 1

perl -v shows the version correctly

Is the subversion of perl matters??

Do I need to install the same version of cgywin??

THanks

Alvin

[Rebecca Yo]

If you are running 64 bit windows, you can use either 32 bit or 64 bit cygwin. If you are running 32 bit windows, you need to have 32 bit installed.

Do you see the correct 'perl -v' output if you run it in a command window instead of cygwin?

Also please check the forum for the same error: https://communities.vmware.com/search.jspa?view=content&resultTypes=&dateRange=all&q=%22Time+Sync+failed+to+start%2Fcomplete%22&rankBy=relevance&containerType=14&container=2427&containerName=VMware+VMmark%E2%84%A2&userID=&numResults=15

[netyale]

Hi Rebecca,

I fixed most of the problems during the weekend!!

And finally I got the score in the end of the result output file!

But I still got the warnings about the mailserver & Olio issues

I check some articles from the community and still not sure the root cause

I increase the CPU core number and didn't help a lot

And I also manually run the LoadGen, It passed and shows it's success

I guess Max TQL : 56 is acceptable, right??

Please give me some direction to fix it

Thanks for your help!!

Alvin

[Rebecca Yo]

Hi Alvin,

This result does look much better. Olio, DVDstore, Standby, SVmotion and Deploy all seem to be working fine.

vMotion is working, but with a high degree of variability. Half the time, vMotion takes 35 seconds, the other half of the time, it takes 300.

Secondly, your mailserver is working correctly but with very high latencies. The application latency can be seen under the heading "TILE_N_QoS(ms) :" An asterisk by the value means that the latency is above acceptable limits. This is why your run failed: "Run_Is_NOT_Compliant". The latencies for Olio and DVD Store, however, are quite good. Can you think of any difference in network or storage configuration between Mailserver vs. Olio and DVD store that may be responsible for the difference? Secondly, is there anything in your environment that may be causing spikes in network or CPU usage, such as shared traffic? The variability in vMotion is very odd.

Thanks,

Rebecca

[netyale]

Hi Rebecca,

First, if I want to change the vCenter IP address, should I reinstall the vCenter??

I cannot connect to the vCenter anymore after change the IP.

The service "VMware VirtualCenter Server" won't start after changing the ip.

Where did you check my vMotion result? (Half the time, vMotion takes 35 seconds, the other half of the time, it takes 300.)

mailserver & Olio are on the same storage & network switch.

So I will try to find out the cause of latency

Thanks

Alvin

[Rebecca Yo]

Hi Alvin,

I do know that changing the vCenter IP address can cause some various issues, especially with the registered hosts. I don't think you should have to reinstall it but you can always do so as a last resort. Try searching the other forums or google for solutions.

All workload results are in the *.wrf files (Workload Results Files) in the results folder. That's where I can see how long each one of your vMotions is taking to complete. (vmotion-t0.wrf) Why don't you check out all of your .wrf files. It will help you catch errors and get familiar with what is happening during a run.

Thanks,

Rebecca

[netyale]

Hi Rebecca,

Long time no see...

I changed my storage setting

It looks OK now

Is it means the setting is all correct and the result is meaningful??

Unreviewed_VMmark2_Applications_Score: 2.92

Unreviewed_VMmark2_Infrastructure_Score: 1.09

Unreviewed_VMmark2_Score: 2.55

How to explain it??

Is it means my system can run 2.55 tiles?

Thank you soooooo much!!!

Alvin

[Rebecca Yo]

This run looks great, the settings are correct. Yes, your combination of systems / storage can support 2 tiles with score of 2.55. If you add additional tiles, it may be able to support even more.

The score is 2.55. This is not the same as the number of tiles it supports. The number of tiles you can run is a 'coarse grain' measure of system capacity. The score is the 'fine grain' measure of performance. Generally, the score is most important.

From VMmark.com:

The unit of work for a benchmark of virtualized consolidation environments can be naturally defined as a collection of virtual machines executing a set of diverse workloads. The VMmark benchmark refers to this unit of work as a tile. In VMmark 2.x, the total number of tiles that multiple systems in the data center can accommodate, while experiencing administrative operations in the background, provides a coarse-grain measure of that data center's consolidation capacity. The number of tiles and the performance of each individual workload within the tiles determine the overall benchmark score.

Congratulations on a great run :smileyhappy:

Rebecca