Skip navigation

With Apple's OS X Lion announcement yesterday, I was quickly reminded of a situation that plays out in the forums at least a few times a year. With each and every new OS release from any company, the forums quickly fill up with the inevitable "When will VMware support my new OS" discussions. In the table below, 7 Ubuntu releases are shown with their respective release dates and when the OS was supported in both VMware ESX(i) and VMware Workstation. The Ubuntu release dates were obtained from https://wiki.ubuntu.com/Releases and the VMware support information was obtained from the VMware HCL.

 

Version
Release DateWorkstation Support - Guest
Workstation Support - HostESX(i) SupportDays Wait
Ubuntu 10.1010/10/201003/29/2011 (7.1.4)N/A11/15/2010 (4.1)37, 171
Ubuntu 9.1010/29/200901/29/2010 (7.0.1)01/29/2010 (7.0.1)11/18/2009 (4.0 U1)21, 93
Ubuntu 9.0404/23/200908/20/2009 (6.5.3)10/26/2009 (7.0)06/30/2009 (3.5 U4)69, 120, 187
Ubuntu 8.1010/30/200803/31/2009 (6.5.2)10/26/2009 (7.0)03/30/2009 (3.5 U4)152,153,362
Ubuntu 8.0404/24/200809/23/2008 (6.5)09/23/2008 (6.5)08/08/2008 (3.0.3)107, 153
Ubuntu 7.1010/18/200703/14/2008 (6.0.3)09/23/2008 (6.5)04/10/2008 (3.5 U1)149, 176, 194
Ubuntu 7.0404/19/200709/19/2007 (6.0.1)05/09/2007 (6.0)07/31/2007 (3.0.2)21, 104, 154

 

As you can see, the wait for support varies anywhere from 21 to 362 days.  This averages out to approximately 135 days.  There are obviously quite a few variables involved here, especially when you consider VMware's own product launches and their timing.  When you think about everything that has to go into the development, testing, training, documentation, etc involved to support an OS, it is staggering.  The bottom line is that there will almost always certainly be a wait involved for the latest guest operating systems to be supported. As for the duration of the wait, the often used phrase "it depends" definitely applies here.

 

Thanks for reading!

Brian

This is an updated version of my blog entry "SQL Server Data Consistency Levels with VCB Backups" from more than a year ago, and vSphere 4.1 and the VMware vStorage APIs for Data Protection have changed things up a bit.

 

The chart below summarizes the current state of data consistency levels, when using VMware vStorage APIs for Data Protection with vSphere 4.1 for Windows/SQL Server backups.

 

SQL 7SQL 2000SQL 2005SQL 2005 SP2SQL 2008
Windows 2000Crash ConsistentCrash ConsistentCrash ConsistentCrash ConsistentNot Supported
Windows 2003Crash ConsistentApplication ConsistentApplication ConsistentApplication ConsistentApplication Consistent
Windows 2008Not SupportedNot SupportedNot SupportedApplication ConsistentApplication Consistent

 

Not Supported:

Microsoft does not support this combination of Windows and SQL Server. A combination unsupported by Microsoft.

 

Crash Consistent:

This is the state in which a system would be found after a system failure or power outage.

 

Application Consistent:

This is the state in which all databases are in-synch and represent the true status of the application.  In the table above, the light green represents a software snapshot provider (Windows 2003) and the dark green represents a hardware snapshot provider (Windows 2008).

 

As with all things backup, you should always verify the restores periodically to ensure the intended result can/will be achieved.

 

Note: This information is valid as of May 17, 2011 for VMware vStorage APIs for Data Protection and vSphere 4.1 editions.

 

As always, thanks for reading.

Brian

vmroyale Guru
User Moderators

.VMX Hardening Options

Posted by vmroyale Apr 8, 2011

The vSphere 4.1 Hardening Guide was released yesterday, and it contains a wealth of information on securing vSphere environments.  The scope of the guide states that the topics covered are the "initial configuration of the virtualization infrastructure layer," which includes VMware ESX 4 and VMware ESXi 4 hosts, Virtual Machine Configuration (.VMX) files, virtual networking infrastructure, VMware vCenter Server, its database and client components and VMware Update Manager.  General Guest OS and application hardening is not included as part of this guide.

 

I updated my security notes, after reading the new guide, and I also decided to include the .VMX options here as a quick reference.  While these .VMX options can help secure the environment, these changes are only a part of a comprehensive strategy.  Also keep in mind that most of these changes will require the VM to be powered off or rebooted, before they can take effect.  With that being said, here are the options:

 

Prevent virtual disk shrinking:
isolation.tools.diskWiper.disable = "true"
isolation.tools.diskShrink.disable = "true"

 

Prevent other users from spying on administrator remote consoles:
RemoteDisplay.maxConnections=1

 

Disable Copy and Paste from VMs:
isolation.tools.copy.disable = "true"
isolation.tools.paste.disable = "true"

 

Ensure that unauthorized devices are not connected (unless needed/required):
floppyX.present = "false"
serialX.present = "false"
parallelX.present = "false"
usb.present = "false"
ideX:Y.present = "false"

 

Prevent unauthorized removal, connection and modification of devices:
isolation.device.connectable.disable=TRUE
isolation.device.edit.disable=TRUE

 

Disable VM-to-VM communication through VMC:
vmci0.unrestricted=FALSE

 

Limit VM log file size and number:
log.rotateSize = "100000"
log.keepOld = "10"
logging=FALSE

 

Limit informational messages from the VM to the VMX file:
tools.setInfo.sizeLimit=1048576

 

Disable certain unexposed features:
isolation.tools.unity.push.update.disable = TRUE
isolation.tools.ghi.launchmenu.change = TRUE
isolation.tools.memSchedFakeSampleStats.disable = TRUE
isolation.tools.getCreds.disable = TRUE
isolation.tools.hgfsServerSet.disable = TRUE

 

Disable remote operations within the guest:
guest.command.enabled=FALSE

 

Do not send host performance information to guests:
tools.guestlib.enableHostInfo=FALSE

 

Control access to VMs through VMsafe CPU/memory APIs:
vmsafe.enable = TRUE
vmsafe.agentAddress=”www.xxx.yyy.zzz”
vmsafe.agentPort=”nnnn

 

Control access to VMs through VMsafe network APIs:
ethernet0.filter1.name = dv-filter1

 

Allow Application Consistent Snapshots in Windows 2008:
disk.EnableUUID=”true”

 

The final option is not actually in the guide, but application consistency can/should be thought of as increased availability.  Many of these options could cause undesired consequences in your environment, so consult the guide and always test before making any of these changes in production environments.

 

As always, thanks for reading!

Brian

The requirement:
A requirement comes down for a set of 4 new virtual machines. Each of the VMs is 185 GB in size and will require a full nightly image backup.

 

The problem:
This additional 740 GB of daily backups is impacting the backup window and using a lot of storage in the process.  With no deduplication, a week's worth of all four VMs requires over 5TB of storage.  Each VM backup was taking 50 minutes to complete, and a total of 3 hours and 20 minutes each night was used just for these 4 VMs.

 

The resolution:
Use NetBackup 7 (which leverages the vStorage APIs for Data Protection) and VMware's Changed Block Tracking (CBT) functionality to reduce both the backup window and storage requirements.  The idea is to take a full image backup of each VM once a week, and then perform a block level incremental backup (BLIB) for the remaining 6 days.

 

The math:


FULL VM IMAGE BACKUPFULL VM w/BLIB
Saturday165 GB, 50 minutes165 GB, 50 minutes
Sunday165 GB, 50 minutes11 GB, 05 minutes
Monday165 GB, 50 minutes21 GB, 10 minutes
Tuesday165 GB, 50 minutes15 GB, 12 minutes
Wednesday165 GB, 50 minutes11 GB, 12 minutes
Thursday165 GB, 50 minutes19 GB, 10 minutes
Friday165 GB, 50 minutes20 GB, 11 minutes
WEEKLY TOTAL x 1 VM1.1 TB, 350 minutes262 GB, 110 minutes

WEEKLY TOTALS x 4 VM4.4 TB, 1400 minutes1 TB, 440 minutes

 

The full backup on Saturday still takes the same amount of time as before, but on Sunday the block level incremental backup takes just 5 minutes to complete.  Monday - Friday offer similar results, and no single BLIB ever takes longer than 12 minutes to complete.  This means that I can get what is essentially a FULL backup of all four VMs in the time it used to take to get just one.  Notice also the amount of data that is backed up.  The entire set of BLIBs for 6 days use less space than a single FULL backup.  Both of these savings are very real and also very easy to implement.

 

The implementation:
Note: The VM must be running Virtual Machine Version 7, in order to use CBT, and the VM must be powered off to enable the CBT changes detailed below.  The first step is to enable CBT on a per-VM basis.  This is done via the "Edit Settings" for the VM.  Next choose the "Options" tab and scroll down to "Advanced" and select the "General" option.  Click the "Configuration Parameters" button and then add the following lines:

 

Name: ctk.Enabled

Value: true


Name: scsi0:0.ctkEnabled

Value: true

 

 

screengrab1.jpg

 

The scsiX:X.ctkEnabled configuration parameter should be added for each disk in the VM, and its value should reflect the Virtual Device Node for each disk.  For example, say there were two disks in the VM with Virtual Device Nodes scsi0:0 and scsi0:1 - there would need to be a configuration parameter added for each disk.  Each disk would be added as a new line.

 

The final configuration parameter (Name: disk.EnableUUID | Value: true) shown in the screenshot is optional.  It is used to enable application-consistent quiescing in Windows 2008 VMs and is discussed in kb 1028881.  Now might be a good time to make this change as well.

 

Once all of the name value pairs have been added to the Configuration Parameters, click OK twice and then power the VM up.  CBT is now enabled, and can be verified with the instructions provided in kb 1020128.

 

Note: If CBT is planned to be used for all virtual machines in the environment, it might make sense to modify any templates to include it by default.


Now that CBT is set up, the NetBackup policy can be set up to leverage BLIB.  Basically, the idea here is to set up the policy's schedule for one weekly full FlashBackup-Windows backup and 6 differential incremental backups.  First, the policy must be set up to use CBT.  This is configured in the "Attributes" tab for the policy and is accomplished by selecting the checkbox for "Perform block level incremental backups."

 

screengrab2.jpg

 

The final step is to configure the scheduling for this policy.  To do this, select the "Schedules" tab and create two new schedules.  The first schedule will be for a full weekly backup to be performed on Saturday.  The second schedule will be for a daily differential incremental backup to be performed Sunday through Friday.

 

screengrab3.jpg

Add the client that CBT was enabled on in the previous steps to this policy, and the policy is complete and ready to be used.

 

After the first full and differential incremental backups are both complete, the results can be verified with the "Backup Archive Restore" NetBackup application.  After selecting the desired VM, the differential backups will show up just like they would for a normal MS-Windows type policy.

screengrab4.jpg

 

The conclusion:

The combination of BLIB with CBT is a very effective solution for lowering backup windows and saving on storage.  For those that are already using vSphere and NetBackup 7, there is no additional licensing or cost associated with implementing either of these technologies.  The information presented here is a very basic approach to implementing BLIB and CBT, and there is much more information about this topic in both the "Designing Backup Solutions for VMware vSphere" Technical Note and the "Symantec NetBackup for VMware Administrator's Guide."

 

As always, thanks for reading!

Brian

http://www.pearsonitcertification.com/ShowCover.aspx?isbn=0789740575&type=f

Last fall, I was fortunate to have the opportunity to be one of technical editors on Eli Khnaser's "Exam Cram: VCP4 VMware Certified Professional VCP-410 Exam."  It was my first time as a technical editor, and it was an interesting and learning experience.  The book was updated from the first edition that covered VI3, and the technical editing was a good use of both my VCP3 and VCP4 certifications.  I recently received a copy of the book, and thought now would be a great time to write about it.  The book is available now!

 

The first thing you notice when opening this book is the Cram Sheet foldout.  This foldout contains 109 facts around the exam that follow the layout of the chapters of the book.  The book follows the exam blueprint chapter by chapter, and along the way there are lots of sample questions, exam alerts and other items that will help prepare you for the VCP-410.  There is also a nice balance between the cram information and the information that Eli includes as tips, notes and cautions.  These items apply more to real world scenarios, and would help any VCP be more rounded in the field.  This information is a nice touch and a real hidden bonus.  There is also a 75 question practice exam, and a very useful reference in the included glossary.  If you still need more practice questions, there are 75 more on the included CD.  If you are looking to take the VCP-410 exam soon, I would definitely recommend picking up a copy of this book!

 

As always, thanks for reading!

Brian

I recently had to update the BIOS and iDRAC firmware on a Dell server running ESXi. I did not have the Dell Server Updates DVDs or the time required to download them, and instead used the Dell OMSA CentOS LiveCD and the Update Package for Red Hat Linux files option.  While the steps below are Dell specific, they could be modified very easily to work with just about any hardware vendor.


01. Download the Dell OMSA CentOS LiveCD, and make sure to read the README file.

 

02. Download the required updates from the Dell support site. Be sure to choose the "Update Package for Red Hat Linux" .BIN format for the downloads.  Again, make sure to read any notes or installation instructions included with the downloads.  For this set of Dell updates, the iDRAC firmware had to be updated prior to the BIOS.

 

03. Use ISO Master (or equivalent) to inject the .BIN files to the root of the Dell OMSA CentOS LiveCD ISO image, and then burn the image to disc. The updates could also be copied to a USB drive and then mounted from within the LiveCD.

 

04. After evacuating all VMs from the ESXi host and putting it in maintenance mode, boot the ESXi host to the Dell OMSA CentOS LiveCD.

 

05. Choose the "Boot (run from CD)" option at the menu.

 

06. Wait for the GUI to start, and then issue the following commands from the terminal window:

 

cd /mnt/live
./<NAME-OF-UPDATE.BIN>

 

07. When the update is complete, issue a reboot command from the GUI of the LiveCD or type the following command in the terminal window:

 

reboot

 

08. After the LiveCD completes shut down, remove it from the ESXi host.  Verify that the updated firmware versions are correct, and then test your ESXi host to ensure everything is working as expected.

 

Thanks for reading!

This is a modernized version of the "Single use ESXUPDATE How To for ESX 4" post I wrote over a year ago.  With more and more people making the transition over to ESXi, it seemed like it was time to have a simple set of instructions to be provided for use in patching ESXi.

 

01: Make sure you have the vMA both installed and configured.

 

02: Download the patch bundle directly from VMware Support. This download will be .zip file.  Do not extract it.

 

03: Use SCP (Windows users use WinSCP) to upload the .zip file to the vi-admin user's home directory (/home/vi-admin) on the vMA.

 

04: Obtain local console access, or SSH (putty), to the vMA virtual machine that the bundle file was uploaded to.

 

Note: In all examples below, 10.10.10.10 would be replaced with the actual IP address (or DNS name) of the ESXi host to be patched.

 

05: Use the vSphere client to put the ESXi 4.x host in maintenance mode.  Alternatively, from the vMA use the command:

 

vicfg-hostops --server 10.10.10.10 --operation enter

 

To next verify that the host is in maintenance mode, run the following command from the vMA.

 

vicfg-hostops --server 10.10.10.10 --operation info

 

The following command may also be used to list running virtual machines.  This is for environments without VMotion or for single hosts.

 

vmware-cmd --server 10.10.10.10 -l

 

The following command may also be used to stop running virtual  machines.  This is for environments without VMotion or for single hosts.

 

vmware-cmd --server 10.10.10.10 <vm-path> stop soft

 

06: Verify which bundles are already installed on the ESXi host, using the command:

 

vihostupdate --server 10.10.10.10 --query

 

Note: You will next be prompted for a username and password when running this and any subsequent commands that connect to an ESXi host.

 

07. Find out which bulletins are available in the bundle.

 

vihostupdate --server 10.10.10.10 --list --bundle ~/<bundle-name>

 

Note: In the example above, the bundle-file would be located in the vi-admin user's home directory.  The list here should be reviewed for accuracy and to ensure that the correct bundle file was downloaded and copied to the vMA.

 

08. Find out which bulletins are applicable to the ESXi host.

 

vihostupdate --server 10.10.10.10 --scan --bundle ~/<bundle-name>

 

09: Finally, the bundle is installed using the command:

vihostupdate --server 10.10.10.10 --install --bundle ~/<bundle-name>

 

10. When (or IF) prompted to reboot, use the vSphere client or the following command from the vMA:

 

vicfg-hostops --server 10.10.10.10 --operation reboot


Note: Not all patches will require an ESXi host reboot.  You will be prompted, when required.

 

11: After the ESXi host boots, verify the patch bundle was installed from the vMA with the command:

 

vihostupdate --server 10.10.10.10 --query

 

12: If applicable, take the ESXi host out of maintenance mode using the vSphere client or from the vMA with the command:

 

vicfg-hostops --server 10.10.10.10 --operation exit

 

13: If applicable, restart virtual machines using the vSphere client or from the vMA with the command:

 

vmware-cmd --server 10.10.10.10 <vm-path> start

 

As always, thanks for reading!

I recently ran into an issue upgrading a customer's vSphere environment.  This environment had 4 older servers running ESX 4.0 in a single cluster that were to be replaced with new hardware running ESXi 4.1. The upgrade process was basically to enable EVC on an existing cluster, remove one host at a time and finally add one host at a time back to the cluster.  When it was time to add the first host back into the existing cluster, I received the error "Cannot complete the configuration of the HA agent on the host.  Misconfiguration in the host network setup."  A quick search of the VMware Knowledge Base turned up kb 1019200 - "Configuring VMware High Availability fails with the error: Cannot complete the configuration of the HA agent on the host"

 

The kb article states that "This issue occurs if all the hosts in the cluster do not share the same service console or management network configurations. Some hosts may have service consoles using a different name or may have more service consoles than other hosts."  This makes perfect sense, as the ESX 4.0 hosts are using a Service Console connection type for the management network and the new ESXi 4.1 host is using a VMkernel connection type.  Fortunately, the kb article went on to discuss how to use the "das.bypassNetCompatCheck" option for HA to ignore this mismatched management connection condition.  Basically, turning off HA for the cluster and then turning it  on again with the "das.bypassNetCompatCheck" option allowed the ESXi 4.1 host to join the existing cluster without incident.  Once the fourth and final host was added to the cluster and the cluster contained only ESXi 4.1 hosts, the "das.bypassNetCompatCheck" was removed using the same procedure.

 

Hopefully this information will help others, as I am certain this upgrade scenario is a common one.

 

Thanks for reading,

Brian

I just ran across kb 1026944, titled "Restoring VMware Consolidated Backup images on vSphere 4.1."  This kb article discusses that while VCB support is included in vSphere 4.1, restoring a previously backed up VCB image will require the use of vCenter Converter Standalone 4.0.1 or earlier.  This is because the VCB restore capability is no longer available in vSphere 4.1, vCenter Server 4.1, and vCenter Converter 4.1.  So if you have older VCB backups that may have to be restored into your vSphere 4.1 environments, make sure you have a copy of vCenter Converter Standalone 4.0.1 or earlier available.

 

Thanks for reading!

Brian

vmroyale Guru
User Moderators

VMworld 2010 - Day 4

Posted by vmroyale Sep 2, 2010

Today was the final day of VMworld, and it started off with a very cool general session.  Three speakers highlighted different devices for interacting in the digital realm.  One of the devices was a computer that is worn and included cameras, sensors and more that allowed the user to do things like use a wall as a workspace or copy a piechart from a piece of paper.  Another device allowed brain waves alone to perform actions on a computer screen.  These technologies were very impressive and were definitely things that made you think.

 

After the general session, it was off to SE8421 - "Hypervisor-Based Antivirus and Endpoint Security."  I was lucky to have already seen most of this presentation last week, thanks to the vExpert program, and the content was largely unchanged since then.  Endpoint looks promising, as it offloads antivirus scanning to a security virtual machine and away from the agents and applications currently installed in virtual machines.  For now, the guests will still require minimal agents, but the goal is to remove them eventually.  Agentless antivirus has obvious benefits, but it really makes a lot of sense in the cloud.  A host can provide antivirus services to customer VMs without any changes to the virtual machines. This is definitely first generation, and I have concerns/questions around the availability of this service with 1 security VM per host - what happens if the security VM crashes or needs a reboot for some reason?  Is maintenance mode for the ESX host going to be involved in this?  Or, how can we ensure 100% coverage?  The scans right now are also limited to file - no memory.  Again, this is a new technology and will only get better.  Very cool stuff though!

 

The next session I attended was TA8440 - "10Gb & FCoE Real World Design Considerations."  This looked at the differences between using multiple 1 Gb cards and fewer 10 Gb cards, including cabling/port requirements and pricing.  The amount of bandwidth that vMotion can consume was discussed, as well as multiple approaches to limit or contain this traffic in a 10GB environment.  Some best practices were mentioned here around vmxnet3, jumbo frames and virtual switches.  Test and verify was also recommended, which was refreshing to hear.  Best practices are one thing, but testing is always a great idea!

 

Next up was TA8133 - "Best Practices to Increase Availability and Throughput for VMware."  There were a lot of best practices, some new and some old and even some confliciting advice from other sessions.  This was a good session, and I have at least a couple of other new things to do as a result.

 

My final session of the day/week was TA8218 - "VMware Storage Vision."  One of the basic ideas around this session was that storage as a utility needs to happen for the cloud to work, and Amazon EC2 was given as an example.  There were also some ideas around physical storage that scales linearly and managed trivially.  VMs managed through policy with Qos and complianc reporting was also discussed.  The ideas here tied in very strongly with many of the cloud themes from the week.

 

And with that, VMworld 2010 is over.

vmroyale Guru
User Moderators

VMworld 2010 - Day 3

Posted by vmroyale Sep 1, 2010

The first session of the day was BC8274 - "VMware Fault Tolerance - Best Practices, Usage Scenarios and Performance."  The session began with the question of how many people in the audience are currently using FT and then another of how many aren't using it specifically because of the lack of SMP support.  We were then told that FT SMP support is being actively worked on and that it will be available in a future release.  This session then went on for far too long with general availability marketing and background information on FT and how it works.  I feel like a session whose title started with "Best Practices" is clearly geared towards people that are already familiar with the product.  With that being said, the technical information was good and explained well. Some interesting stats that came out of this session were that 20% of VMware customers use FT in Production, and 90% of VMware customers use HA in Production. That means there are 38,000 customers using FT in their production environments, based on the customer numbers stated yesterday in the general session.  It was also stated that databases (mostly smaller SQL Servers) are the most common workload used with FT currently.

 

My second session on the day was TA7121 - "Next Generation VM Storage Solutions with vStorage API for Array Integration (VAAI)."  This was an excellent session and it offered up a lot of information around the current state of the VAAI and two sneak peaks at future technologies.  I learned that a new VMFS driver (3.46) ships with 4.1 that contains new algorithms for hardware accelerated functions. Now I need to find out if this version of VMFS is required for the VAAI features to work - one more thing to do.  The futures section of the presentation focused on vStorage API for Thin Provisioning and vStorage API for Storage Awareness.  Both of these technologies focus on giving more responsibility to the storage arrays and taking the work away from VMware hosts and ultimately away from VMware administrators.  In the Q&A portion of this session, a question was asked about when the (2 TB - 512 bytes) limitation in VMFS was going away.  The presenter could not answer this question, but responded with, "you will be happy very soon."  This was actually one of the questions I had hoped to have answered this week, and since VMware doesn't officially talk futures this answer is close enough for me.

 

I spent some time after lunch walking around the Solutions Exchange, now that it isn't quite so crowded down there.  There are some really cool demos and booth candy in there. After a stop by the Social Media and Blogger Lounge, it was off to TA8065 - "Storage Best Practices, Performance Tuning and Troubleshooting."  This was another great

session that was packed and contained a lot of valuable information.  Topics covered included sizing LUNs for performance, limits on VMs per datastores, queue depths, latencies, balanced paths, partition alignment, esxtop for troubleshooting and more.  This was one of the better sessions I have attended, for pure information, and I have a few more things to do for it.

 

My final session of the day was supposed to be TA7743 - "ESX iSCSI News, Configuration, and Best Practices." Unfortunately, I had to miss it since the previous session ran over.  The next stop is the St. Regis Hotel for the vExpert meeting and then off to the VMworld party.

 

Day three is almost done! Looking forward to day four already!

vmroyale Guru
User Moderators

VMworld 2010 - Day 2

Posted by vmroyale Aug 31, 2010

VMworld 2010 Day 2 started off with the first general session.  I saw people in line at 7:30 this morning - for a 9:00 start!  There were lots of announcements made today around new VMware products, acquistisons, strategy and vision. Even after having a preview of some of this information last week, it is still a lot of information to take in and comprehend. 

 

Some interesting items this morning:

  • VMware has identified three phases of virtualization progression:

  • - Phase 1 is cost savings.

  • - Phase 2 is QoS/Business Prop - Most VMware customers (190,000) are in phase 2 right now.

  • - Phase 3 is IT as a Service, which will be driven by hybrid cloud computing. 

  • In 2009, for the first time, virtual machines were deployed more than physical machines.

  • Virtual Machine growth is 28% a year.

  • VMware wants to decrease OPEX (estimated $6-8 spent on OPEX for every $1 on CAPEX).

  • The secure hybrid cloud uses private cloud and service provider cloud in combination.

  • A virtual datacenter is a collection of apps that share a set of common policies.

  • SaaS is growing everywhere; the pc form factor is threatened; new devices fuel new applications.

  • View desktops are now priced at less than $500, which is cheaper than real PCs

  • Split the datacetner into virtual datacenters to achieve different SLAs. 

  • Abstracting business production from user consumption is a goal.

  • IT as a Service - the IT service catalog maybe becomes something like an app store.

 

My first session on the day was EA7860 "Java Apps on vSphere: Technical Best Practices and Why Now Is Better Than Ever" and it was a good session.  There was good information on how Java works that laid the groundwork nicely for the best practices.  Best practices always make more sense to me, when you can fully understand the "why" and not just the "how."  And the list of things to do grows a bit more, as I now have a few JVM items to investigate when I return home.  An interesting comment by the speaker was made about VMware using Java as the development platform.  That statement certainly makes sense with all these new web interfaces I am seeing this week..

 

After lunch and hanging out in the blogger's lounge for a while, it was time for TA8233 - " Prioritizing Storage Resource Allocation in ESX Based Virtual Environments Using Storage I/O Control"  this session was good, but I was personally a bit disappointed.  I had printed out copies of both the "[Storage I/O Control Technical Overview and Considerations for Deployment|http://www.vmware.com/files/pdf/techpaper/VMW-vSphere41-SIOC.pdf]" Technical White Paper  and the "[Managing Performance Variance of Applications Using Storage I/O Control|http://www.vmware.com/files/pdf/techpaper/vsp_41_perf_SIOC.pdf]" Performance Study documents to read over on the plane ride to SFO.  This session was mostly a repeat of these documents, but a couple of really good best practices were covered that made the session worth it.

 

My final session on the day was EA8004 - "SAP solutions on vSphere." This was another excellent session with a lot of good information to use later.  One of the more interesting statements I heard was that VMware actually staffs support employees at SAP to handle incidents with SAP customers using VMware.  SAP also has leveraged the SDKs to pull performance information from VMware hosts into its performance monitoring tools.  To me, these actions show a dedication from both VMware and SAP and is further validation that SAP on VMware can be a viable option.

 

Stopped by the Solutions Exchange to talk to CA about vSphere integration with their ARCserve product.  CA isn't shy about giving away things, and their giveaways are always very impressive.  With that being said, they are giving away a car this week!  Had an invite to go to their party, and am regretting not making it. I'm sure it was over the top!

 

Day two is done! Looking forward to day three!

vmroyale Guru
User Moderators

VMworld 2010 - Day 1

Posted by vmroyale Aug 30, 2010

VMworld started yesterday for me.  Registration was open, so I took the opportunity to get registered and pick up the welcome kit.  The line for the welcome kit was long, and the event staff had actually stopped the line at one point.  Fortunately, it was a quick wait and candy was delivered in line!  From there I spent a little time learning the Moscone Center and figuring out where everything was.  I ran into some fellow vExperts, after leaving the Moscone, and took a chance to unwind and talk with them.

 

The start of Day 1 began with a stop by the Social Media and Blogger Lounge, where I saw John Troyer and also got to meet Alex Maier.  From there it was off to the first session of the day - EA7850 "Design, Deploy, and Optimize Microsoft SQL 2008."  The session covered licensing, vSphere capabilities, VM sizing and P2V conversions, best practices for performance, high availability strategies for SQL using different options (MSCS, HA, SQL mirroring) and backups (including a plug for SRM).  This was a good session that covered a lot of ground and it was packed for the duration.  I also left with a few ideas and things to test when I get back.

 

The next session I attended was MA8030 - "Saving Time with vCenter Orchestrator" and it was a good session, but I'm afraid I just don't understand how Orchestrator is supposed to be used.  It appears to be for 3rd parties or systems integrators.  Another item added to the "to do" list from VMworld.  I did notice that the interface for Orchestrator was being used on Safari, and this has we wondering about the web-based GUI as a standard for VMware products. 

 

After lunch, it was a long snaking line for EA7849 "Design, Deploy, and Optimize Microsoft Exchange Server 2010 on vSphere."  This was another packed session with people standing along the walls. There was actually a lot of duplicate content from the Microsoft SQL 2008 session earlier, but in all fairness best practices are often

duplicated. 

 

Monday's final session was TA8245 "ESXi Internals: Better Understanding for Better Management and Troubleshooting."  This was labeled as an advanced technical session, and it delivered in that regard. The key point from this session was that if you still use disks to install the ESXi host on and couple this with syslog use, you are in

pretty good shape.  Diskless ESXi hosts definitely have their advantages, but it appears that their use with ESXi will require some additional setup/configuration in comparison to disked hosts.  One final item on the "to do" list is to look into the debug VIB available for ESXi.

 

The day wrapped up with a stroll around the Solutions Exchange.  It was very crowded and almost impossible to really talk to the vendors.  It was good to see who the vendors are, but those visits will have to be later in the week.  Picked up some more cool buttons for the badge, and typed this up.

 

Day one is done!  Looking forward to day two!

 

Thanks for reading,

Brian

I recently attended the new VMware vSphere: Manage for Performance course and thought I would summarize how it went.  From what I understand, this course was put together by Scott Drummonds.  I've been a fan of Scott in the forums and on his personal blog for a long time, so I had a fairly high set of expectations going into this course.

 

I prepared for the course by reviewing the course outline and then finding relevant material on VMware's website.  A few of the items I read (or mostly, re-read) in preparation for the course were:

 

 

 

 

 

 

 

 

The course was well organized and generally covered the "core four" components of cpu, memory, storage and networking.  The student manual is very good and contains a good amount of concise information that can be very useful in troubleshooting performance issues.  Quite a few of the metrics and thresholds looked very similar to what Duncan Epping put up at Yellow Bricks in his excellent esxtop entry. I will say that a certain amount of the material contained in the student manual was also repeat reading for me, either due to my preparation work and/or following of Scott Drummonds' blog.  Many times that I had a question or needed more information/clarification, a quick search of Scott's site would provide the detail needed. As with any new course, the labs had a few minor issues.  There were no show stoppers here though, and for the most part the labs were very effective at demonstrating the concepts they attempted to enforce. 

 

Where the course most excelled was in the way it taught the concepts and/or mechanics of how things actually work and then used a lab exercise to demonstrate these principles. Knowing that a certain CPU Ready value is bad is one thing, but actually understanding why this situation occurred to begin with is infinitely better. I would recommend this course for anyone interested in learning more about how VMware vSphere actually works and how to use this information to your advantage.  I left with a couple of pages of notes and a handful of burning questions that need answers and some scenarios that needed testing.

 

As always, thanks for reading!

Brian

I was recently investigating a situation where certain NetBackup 6.5.4 backups that utilized VCB 1.5 U1 were taking an unusually long time to complete.  For example, a 350 GB backup of a virtual machine was taking almost 8 hours!  This was a LAN-free backup of a virtual machine sitting on a Clariion SAN.  Something was definitely wrong.

 

It didn't take too long to find Document ID 323230, where Symantec reports a bug described as slow VCB backups and explained that "This occurs when the NTFS File system is handling large files."  There was also a fix introduced in  NetBackup 6.5.5 that allows the ability to export virtual machine disk (vmdk) files as 2GB chucks instead of a single monolithic file.  This sounded like a promising fix.

 

Now facing a change request for a NetBackup upgrade, I went to look at the release notes for the latest version of VCB - 1.5 U2.  Under the resolved issues section of the release notes, it states that "Performing backup of virtual machines containing .vmdk files greater than 50GB with the vcbmounter -M1 (monolithic export) command takes longer duration compared to vcbmounter -M0 command. The issue is fixed in this release."  NetBackup was definitely using the monolithic export, so I decided to simplify things a bit and just upgrade VCB to the latest version first.  I perceived the VCB update as a quicker win with less risk. 

 

Upgrading VCB to the latest version was simple.  I stopped the NetBackup services, installed VCB 1.5 U2, verified the VCB install was a success (as detailed in the release notes), and finally restarted NetBackup services.  Running the same backup job for the 350 GB virtual machine now completes in less than 3 hours.  The next step will be to upgrade to NetBackup 7, where these times should improve even more!

 

As always, thanks for reading!