Data Center Strategies

Oracle Changes its Position on x86 Hypervisor Support (Unfair Licensing Remains)

I’ve been covering Oracle licensing and support issues in x86 virtualized environments for quite some time, beginning with the January 2008 report “Virtualization Licensing and Support Lethargy: Curing the Disease That Stalls Virtualization Adoption.” You can also view these earlier blog posts for additional background:

• Oracle and the Big Elephant (August 2008)
• A New Year’s Resolution for Oracle (December 2008)
• Oracle Honors its New Year’s Resolution: Non-Oracle x86 Hypervisors are Now Supported (May 2009)

A few weeks ago one of our clients pointed me to a recently published Oracle support article (Metalink 794016.1 published March 27, 2009), which prompted me to write my previous post. That’s when the fun really began. After my last post on May 6th, Oracle published a completely revised version of the Metalink document on May 8th. The document was referenced by the same document ID (794016.1), but had a completely different title and content.

For context, the March 27th version of the Metalink document was titled “Platform Vendor Virtualization Technologies and Oracle E-Business Suite” while the revised May 8th edition of the document was titled “Hardware Vendor Virtualization Technologies on non x86/x86-64 Architectures and Oracle E-Business Suite.” If you recall, the first iteration of the document described how x86 virtualization technologies were supported with the following statement:

The use of platform vendors’ virtualization technologies (both software and hardware based) to host Oracle E-Business Suite 11i and R12 is covered by Oracle’s policy with regards to 3rd-party products – that is, they are ‘not explicitly certified, but supported’.

The support document listed Microsoft, VMware, and Citrix as examples. In the May 8th edition of the support document, the above statement was revised to:

The use of hardware vendors' virtualization technologies to host Oracle E-Business Suite 11i and R12 follows the same policy as Oracle's policy with regards to customizations - that is, they are 'not explicitly certified, but supported'.

Examples of x86 virtualization hypervisors were replaced by the following statement:

This document provides a statement regarding Oracle E-Business Suite (11i, R12) support of Hardware Vendor Virtualization technologies on non x86/x86-64 systems.

The bottom line – the revised support document went from describing support for x86 hypervisors to ignoring them altogether, with the exception of Oracle’s hypervisor Oracle Virtual Machine (OVM). I was told that the revisions were needed to address confusion, but feedback I received from numerous Burton Group clients made it clear that there was no confusion until the May 8th revision was published.

Since early last week, I have had numerous calls with Oracle on the subjects of both licensing and support, and unfortunately the news isn’t all good. Let’s start with the positive. According to Oracle, VMware’s ESX hypervisor “has been supported since November 2007.” As proof, you can view Oracle Metalink document 249212.1 (note that you’ll need an Oracle support account to view the doc). The document states the following:

Oracle has not certified any of its products on VMware virtualized
environments. Oracle Support will assist customers running Oracle products on VMware in the following manner: Oracle will only provide support for issues that either are known to occur on the native OS, or can be demonstrated not to be as a result of running on VMware.
If a problem is a known Oracle issue, Oracle support will recommend the appropriate solution on the native OS. If that solution does not work in the VMware virtualized environment, the customer will be referred to VMware for support. When the customer can demonstrate that the Oracle solution does not work when running on the native OS, Oracle will resume support, including logging a bug with Oracle Development for investigation if required.

The statement goes on to say that Oracle RAC is not supported on VMware environments. If you’re looking for additional background on Oracle support for VMware environments, I suggest reading the following other perspectives:

• “What the Oracle / VMware support statement really means...and why” (Jeff Browning)
• “Oracle on VMware – a manifesto…” (Chad Sakac)
• “EMC attacks Oracle on its VMware support policy” (Alessandro Perilli)

Regarding VMware support, here’s the translation – if you call for support and you have a known bug, you’re good to go. If you’ve found a new (previously unknown) bug, you’re first going to have to reproduce the fault on physical hardware before Oracle will help you. Compared to other vendors that support enterprise applications on VMware or x86 virtualization environments, this is one of the most restrictive policies out there. Most enterprise software vendors only require faults to be reproduced on the bare metal if they are directly related to performance that could be attributed to the virtualization layer.

The recent Virtual Iron acquisition further cements the fact that Oracle is serious about virtualization. Microsoft and Citrix both have clear public support statements about virtualization and the hypervisors they support (I’m mentioning these two vendors because they’re both virtualization vendors and enterprise software vendors). Oracle needs to loosen its support restrictions for VMware and all x86 virtualization environments, and needs to broaden its list of supported (but not certified) hypervisors to include Microsoft, Citrix, Novell, and Red Hat.

Finally, as I previously mentioned, the larger problem here is licensing. Oracle is requiring customers who wish to deploy Oracle products on x86 hypervisors to license Oracle software by physical server CPUs. Suppose you had two Oracle Database VMs (each with two virtual CPUs) running on a two-node ESX cluster that uses two four-socket servers. Since it’s possible that you’d have a VM on each node, you’d need to purchase licensing to cover the 8 total sockets. If you ran Oracle’s hypervisor, you could license by virtual CPU, however this is only allowed if you pin the VM to fixed CPU cores by hard coding CPU bindings. You can read more about that here. This does create a slight advantage for OVM over competing products, but by binding a VM to one or more physical CPU cores, you have to give up advanced virtualization functionality such as live migration. If I’m using application-level high availability features, this configuration may not be a big deal and would in turn favor Oracle; however it is far from ideal.

Oracle’s competitors in the database arena allow their products to be licensed by virtual CPU without requiring physical bindings (see Microsoft’s and IBM’s policies), and so should Oracle. Doing so allows VMs to move about the physical infrastructure as required to support IT operations. Binding enterprise software licenses to physical assets is a legacy licensing model, and Oracle is practically alone in their licensing policies.

Oracle’s strategy with regard to licensing is one that I’ve seen before. Oracle is effectively taxing organizations for running Oracle Database in a VM. In most cases, organizations will have to pay increased licensing fees. This policy hurts the customer, and in my opinion is an attempt to stall market adoption while Oracle finishes building out its own x86 virtualization platform.

Oracle, it’s time to classify all x86 hypervisors as “hard partitioning.” Our clients are increasingly deploying enterprise applications on x86 virtualization hypervisors. You’re putting them in a tough position, and many consider the virtual infrastructure the foundation for their cloud architecture. Some clients have told me they are now considering moving forward with DB2 or SQL Server because they are unwilling to pay a penalty to run Oracle on any x86 hypervisor. In the end, our clients shouldn’t have to make that choice. They should have the freedom to run the applications they want on the platform they want. This licensing policy is affecting the bottom line of our clients and could ultimately affect your bottom line too. It shouldn’t have to come to that. Let’s just "right the wrong.” Besides, your “Partitioning” document which describes software licensing for virtual environments was last updated in January 2008. In response to my last blog post, you were able to revise a support statement within two days. How about taking the time to revise a licensing policy that is clearly outdated and places an unnecessary burden on our clients?

Oracle Honors its New Year’s Resolution: Non-Oracle x86 Hypervisors are Now Supported

In case you haven’t seen, Oracle issued a major product support update last month - Platform Vendor Virtualization Technologies and Oracle E-Business Suite - Metalink Note 794016.1 (note that an Oracle support account is needed to view the update). The bottom line – Oracle now offers best effort support for all of its E-Business Suite applications on any x86 hypervisor. Shocked? Here’s a snippet of the support statement:

The use of platform vendors’ virtualization technologies (both software and hardware based) to host Oracle E-Business Suite 11i and R12 is covered by Oracle’s policy with regards to 3rd-party products – that is, they are ‘not explicitly certified, but supported’.

What this means is that while these technologies are not certified, Oracle will not turn away a customer reporting an issue solely due to the use of these technologies. When possible Oracle will triage and attempt to diagnose the issue reported – Oracle support may attempt to replicate the issue in a non-virtualized environment and work with the customer to verify if the problem exhibits in such an environment.

Any specific problem isolated to the virtualization software (i.e. a problem that cannot be reproduced in a standard, non-virtualized environment) will need to be referred to the specific vendor for resolution.

Customers should review all relevant Oracle documentation on the use of such virtualization technologies for known issues and limitations with respect to EBS technology components such as the database, RAC, etc.

Customers intending to use 3rd-party products covered under this policy in production environments should conduct appropriate levels of testing and also have contingency plans to revert to a standard certified configuration (that is, non-virtualized environment)…

So there you have it. Back in December I suggested that Oracle make two New Year’s resolutions:

• Offer best effort support for all major x86 server virtualization hypervisors
• Offer virtual CPU-based licensing for all of its server applications

The year isn’t even half way over, and Oracle can cross the first resolution off its list. Next up has to be software licensing. Oracle considers its own x86 hypervisor, Oracle VM (OVM), a platform capable of supporting hard partitioning (see the Oracle “Partitioning” document for more information). By its definition of “hard partitioning” Oracle allows virtual CPU-based licensing on OVM, but does not allow it on other popular x86 hypervisors such as VMware ESX, Microsoft Hyper-V, or Citrix XenServer. Oracle also allows virtual CPU-based licensing on Amazon EC2, which runs the open source Xen hypervisor (you can read more about that policy here). Updating the support policy was a great first step, and Oracle should be commended for responding to the needs of its customers.

Now how about knocking out New Year’s resolution #2 before the end of June? Oracle, I know you can do it. Your friends in the enterprise software space that offer CPU-based licensing, such as IBM and Microsoft, both allow licensing to virtual CPUs on any major hypervisor. Binding a license to a physical CPU is “so 2007.” Oracle, no doubt you’re in the middle of a major makeover, and acquiring Sun was a good move. I must say, with the Sun portfolio, I love your wardrobe. However, your licensing policy doesn’t reflect your new look or attitude. To stay with the wardrobe analogy, you’re wearing some great clothes, but you need to lose the mullet.

Oracle, let’s complete the makeover. Modernize your licensing policies and your body of actions will show that you are a company that is truly one with the times.

Posted by: Chris Wolf

Desktop Virtualization Webcast

I recently completed a webcast that highlighted Burton Group’s position on the state of the VDI market, along with an assessment of architectural alternatives and technical barriers. This follows the release of our desktop virtualization overview “Desktop Virtualization: Choices and Challenges.” If you’re interested, the webcast is approximately 45 minutes in length and can be viewed below (or click here to view it in a separate browser window). 

VMware Launches the V Series Mainframe

VMware’s clever marketing folks can call their latest release whatever they want (vSphere 4.0 if you’re keeping track), but to me it’s the V Series Mainframe.

VMware is taking mainframe-class availability, performance, and infrastructure management principles and bringing them to commodity hardware. vSphere 4.0’s release, in my opinion, makes it hard to argue against VMware’s intentions of a software mainframe. VMware Fault Tolerance (FT), for example, is one of the new features that provides the availability levels required by many tier 1 applications. This is especially critical for home grown tier 1 apps that do not have built-in resiliency. VMware FT allows a single VM to be kept in lock-step on 2 physical hosts, allowing the VM to be fully resilient to the failure of a physical host. FT is a 1.0 release and does have some limitations. For example, only VMs with a single virtual CPU are supported. Note that this isn’t a VMware-only limitation since competitors such as Marathon Technologies cannot mirror VMs with multiple vCPUs either. It’s a tough problem to solve and one that’s going to take some time.

Back to the announcement. VMware is further cementing their position that they are a cloud OS, and I agree. Many tasks associated with the traditional OS (e.g., resource scheduling, QoS) are moving down to the hypervisor and associated virtual infrastructure, so the roles of the traditional monolithic OS are changing. VMware isn’t alone here. Microsoft is retooling its OS and application delivery methodologies as well.

Many of our clients are interested in internal cloud and are looking for practical steps they can do today to start down the cloud path. Most aren’t able to seriously consider external cloud out of concerns regarding security and regulatory compliance. However, organizations can begin building an internal cloud architecture that is capable of leveraging external cloud resources once the predominant security and compliance riddles are solved. VMware is banking that if vSphere is the foundation for an enterprise’s internal cloud, the enterprise will look at vSphere-based external cloud resources once the time is right.

vSphere 4.0 includes a laundry list of new features, including:

• Distributed Power Management (DPM)
• Availability of the VMsafe API
• Distributed virtual switch and support for the Cisco Nexus 1000V
• Host profiles
• vShield Zones
• Thin provisioned virtual hard disks

I’m not going to discuss all of the features (you can get the details here), but I wanted to talk about a few takeaways from the launch event:

• DPM
• Thin provisioned virtual hard disks
• I/O as a factor in VM placement
• Hot memory add
• Partial host failure detection

When DPM was announced with ESX 3.5, Burton Group advised clients to stay away from using it for two reasons:

• VMware considered the feature “experimental” and wouldn’t officially support it themselves
• VMware’s IHV partners would not officially support it either

I had blogged on DPM back in November, and while the post incited some strong vendor reaction, it served its purpose – move the discussion forward on official IHV support for DPM, an assurance many of our clients wanted before they would consider implementing it in production. VMware now supports DPM, which is a good first step. Burton Group has had serious dialogues with all major IHVs on the topic of DPM for nearly a year, and we believe that official support from the server IHVs isn’t far off, so stay tuned. DPM can result in substantial power and cooling cost reduction by shutting down unneeded servers in a given ESX cluster, and then turning them back on once they’re needed again. Once the IHVs step up and do their part, I expect some Burton Group clients to begin implementing DPM for some of their workloads.

Thin provisioned virtual hard disks (i.e. virtual disk files that grow as data is added to them) isn’t a new concept. VMware Workstation has had this feature since its inception. It’s in other hypervisors such as Virtual Server 2005 and Hyper-V too. It was even in ESX 3.5, but wasn’t officially supported. VMware is high on thin provisioned virtual disks, but keep in mind that there will be a small performance overhead associated with using this feature. VMware has yet to publish a benchmark illustrating the performance tax. For enterprise implementations, thin provisioning is best done in the storage array. For smaller deployments and for deployments involving arrays that don’t support thin provisioned storage, using thin provisioned virtual disk files can result in considerable storage savings.

During Steve Herrod’s keynote, he passionately stated how vSphere 4.0 can run practically any x86 workload, including high end I/O intensive databases. He further went on to tout the values of VMware’s distributed resource scheduler (DRS) and DPM features. Both DRS and DPM relocate workloads to balance performance utilization across an ESX cluster or to shut down unneeded physical servers. That all sounds good, but there’s one problem. The intelligence used by DRS and DPM only take memory and CPU utilization into account when determining VM placement. I/O utilization is ignored. This means that it’s possible that a relocated VM will be I/O bound as soon as it lands on a new physical host. I’ve talked to VMware technology partners who are eager to provide deeper I/O utilization information to vCenter. The problem is, however, that vCenter doesn’t have an API that can be used for this purpose, nor does it have the metadata structure for storing this type of information. Until VMware can factor I/O into VM placement decisions, you should use caution when considering enabling DRS or DPM on I/O intensive workloads. To be fair, VMware’s competitors don’t take I/O into account for VM placement decisions either, but I still see it as something that needed to be pointed out.

Hot resource add (e.g. RAM) is a nice new feature too. One thing to remember though is that hot adding memory to a running VM is only useful if the application running inside the VM can take advantage of the new memory without a restart. If the application must be restarted, the VM is as good as offline anyway. That being said, the way applications are able to leverage hot memory add should be something you’re querying prospective software vendors about, and should be something you include in RFPs.

On the high availability side, I’m still waiting on partial node failure detection. Why is this important? Consider an ESX host that’s online, but due to physical storage controller failure is unable to meet required service levels. So while storage access may remain thanks to multipath support, you may not have enough available I/O to meet service level requirements. Intelligence that allows the cluster to rebalance VMs due to reduced I/O availability, for example, further brings vSphere closer to VMware’s goal of building a software mainframe.

Based on the last few paragraphs, it may seem like I'm raining on VMware’s parade, but that’s not my intent. vSphere 4.0 is a major release, and if the massive performance improvements measure up to VMware’s claims, the hardware savings resulting from the associated VM consolidation densities will be enough to cost-justify a vSphere 4.0 upgrade. Of course, the enhanced security (e.g. VMsafe API) and networking features with accelerate adoption as well.

Once again, VMware’s raised the feature bar. Next, I’m looking forward to see how Citrix and Microsoft respond at each of their conferences next month.

Posted by: Chris Wolf

Shhhh! Burton Group Website Easter Egg

Just in time for Easter, Burton Group has put an Easter Egg on the Catalyst Doorway page, offering a big discount for our Catalyst Conference coming this July.

If you've never been to Catalyst, then I highly recommend you go. As always, Catalyst is an intense, 3-day event that covers some of the most important topics in the IT industry. If you've been to Catalyst, then you know what I'm talking about....it's like drinking from an information fire hose. This year, I'm especially excited about our cloud and virtualization coverage. For cloud, we are launching our coverage in earnest (actually the launch is April 20th, another secret) by defining the cloud, discussing usage models, how to build an internal cloud, security, cloud economics, and pitfalls. For virtualization -- client, server and storage virtualization --  we'll delve into top-of-mind concerns such as storage connectivity, virtualization security, network connectivity and virtual switches, application development, new HW technologies, management....basically the whole enchilada.

As always, the DCS team (Richard Jones, Chris Wolf, Nik Simpson, Gene Ruth, and myself) will there to give our perspective and analysis on these topics, but we also have some killer speakers in Paul Maritz (CEO VMware), Mark Templeton (CEO Citrix), Bret Hartman (CTO, RSA), Simon Crosby (CTO, Citrix) there to discuss some of these important issues. 

OH -- and it's in San Diego again this year (my favorite Catalyst venue).

OK, to crack open the Easter Egg (and get the discount), here's what you do:

1. Go to the Catalyst home page (http://catalyst.burtongroup.com/).

Either: click and then drag your mouse off the logo and release the button. OR: roll over the San Diego button but do not click, wait about 20 sec.

2. A message will pop-up stating

"Congratulations! You’ve found an exclusive discount code for Catalyst 2009. Use promo code: Easter Egg and get General Sessions for only $999! Register today – this discount is limited to 50 users and could disappear at any time!"

3. Register.

That's it. Good luck and hope to see you at Catalyst.

[posted by: Drue Reeves]

Hypervisor Competitive Differences - The Aftermath

Following last Wednesday’s announcement and VMworld Europe presentation regarding enterprise production-class hypervisor evaluation criteria, I have received an enormous amount of feedback from both users and vendors. My inbox was flooded with highly positive feedback from more than 150 users (e.g., architects, directors, integrators, and administrators). Vendor feedback, on the other hand, was mixed.

Between the time of the presentation and today, I’ve had the opportunity to speak directly with three of the vendors covered in the presentation, and had an email dialogue with the fourth vendor. Most vendors thought the presentation was fair and balanced, but not all were thrilled with the media coverage. For example, here are two articles in which the announcement was covered:

• VMware outshines Hyper-V, et al. in hypervisor comparison
• Hypervisors Compared: VMware Stands Alone

In the “VMware outshines Hyper-V” article, I was quoted as saying Microsoft would not be happy with the presentation. For background, when I was interviewed by Bridget (the reporter who wrote the article), I had given examples of how each of the four vendors covered in the presentation would not like the results. I mention this because I don’t want it to appear that I was singling out Microsoft. All vendors were treated equally in the presentation, and I didn’t hold punches with any of them. Several folks on the Microsoft virtualization team were present during my presentation, and they were not upset at all. The Microsoft team understood our position and appreciated the fact that I mentioned in the presentation that Hyper-V on Windows Server 2008 R2 will be close to meeting our enterprise production hypervisor feature requirements. Once the product ships, we will conduct a full evaluation against our criteria and go from there.

Citrix XenServer 5.0 and Virtual Iron 4.5 also had respectable showings in the presentation. Sure no vendor wants to hear their shortcomings aired in a public forum, but both vendors, like Microsoft, agreed with our points. That being said, we will stick by our position – only hypervisors that meet our entire list of required features are enterprise production ready. That leaves only VMware VI 3.5 as a product we will recommend for enterprise production roles. As other hypervisors satisfy our feature requirements (one in particular is very close), we will recommend them as well.

Both Microsoft and VMware have come forward and made statements in favor of the evaluation criteria.

“Burton Group is the first technology analyst firm to arm customers with an in-depth list of criteria for evaluating virtualization software. This analysis shows the depth and breadth of Burton Group research expertise, and their ability to deliver substantial value to customers.” - Bogomil Balkansky, VP of Product Marketing, VMware

“As was the case with my colleagues, we thought Burton’s report was a fair representation of the primary commercial hypervisors available to customers. And as we discussed last week, we aim to prove that our collaboration with Citrix, Novell and Red Hat, as well as new features in WS08 R2 and SCVMM r2, will improve Microsoft’s position in Burton’s report.” -Patrick O’Rourke, Group Product Manager - Windows Infrastructure, Microsoft (this quote was taken from this post’s comments)

Of course, user validation is most important, but it’s also important for vendors to accept our criteria and to strive to meet or exceed it. In my opinion, receiving positive backing from the top two vendors in the x86 virtualization space further validates our research and resultant evaluation criteria.

I’m not going to revisit the points in the presentation (you can view a 15 minute summary presentation here), but I cannot emphasize enough the importance of the long term commitment associated with server virtualization platform decisions. Exit costs associated with migrating hypervisors can be significant and must be weighed. Migrating from one virtual hard disk format to another may require the time and nearly the same manpower that it took for the physical to virtual conversions that drove an organization’s initial server virtualization deployment. Also, you’re not just getting a hypervisor, you’re getting an ecosystem, and that is why our evaluation criteria focus heavily on factors such as storage, network, security, and management integration. Hypervisors must provide the performance and stability expected by the organization, but also integrate well within the rest of the IT infrastructure.

Many of our clients considering going with multiple hypervisors see a natural divide between server virtualization and desktop virtualization. You maintain separate VM OS images and rely on different management tools in the desktop space anyway, so using a different hypervisor for the desktop environment is a possibility. In most cases, servers and desktops are managed by different divisions in the IT department. The separation is natural in these cases, and the familiarity with associated management tools is just as important. In addition, the refresh cycles, patch cycles, and other updates are typically different between server and desktop. That being said, vendors looking to penetrate established VMware clients need to do a better job talking about how they are better platforms for virtual desktops. One way to do that is by letting their features do the talking; adding memory overcommit support would be a good first step.

At Burton Group, we will be spending considerable time on competitive differentiation in 2009. The VMworld presentation and announcement was really just the start. If you have thoughts on our evaluation criteria, I would love to hear them.

Note: Updated March 5, 2009 to include quotes from VMware and Microsoft.

Posted by: Chris Wolf

VMworld Europe Day One Thoughts

The VMworld Europe day one keynote featured VMware CEO Paul Maritz outlining VMware’s future vision, with particular attention paid to the cloud.  Maritz’s session is already available online, so I’m not going to write a point-by-point summary of his keynote. Instead, I’ll focus on a few key thought-provoking topics:

• Standards-based vs. proprietary cloud architectures
• Security and compliance in the cloud
• VMware/Intel client hypervisor announcement
• Mega merger in the works?

Standards-based vs. Proprietary Cloud Architectures

Maritz spent time on the need for public and private cloud standards that promote application migration between different cloud platforms or service providers. By migration, I’m not referring to any type of live migration. Instead, interoperability should remove lock-in to a particular cloud provider and provide enterprises with greater negotiating leverage when evaluating service providers or renewing existing contracts. VMware is painting a very nice vision here.

Security and Compliance in the Cloud

A number of cloud providers (terremark, IT Structures, EngineYard, and Logica) took the stage to describe their technology, which is built on VMware infrastructure. In each case, the service provider described an easy-to-use management model, but not one provider talked about the issue of regulatory and security compliance. Our clients have continued to indicate the compliance remains the primary barrier for enterprise cloud adoption, so I was surprised that none of the providers addressed this concern. If providers want enterprises to seriously consider their platforms, then they will have to do a better job articulating a solution to the existing compliance and security concerns around public shared infrastructure. Reading between the lines, I think it’s implied that cloud service providers currently do not have a working solution for compliance validation on shared infrastructure. If a provider does have such a solution, I would like to hear it.

Ideally, I would like to see a DoD-like standard security model for cloud service providers. For example, assume that the following options were available:

• Level A: Dedicated physical and virtual infrastructure, including dedicated server and networked storage assets.
• Level B: Dedicated virtual and physical server infrastructure, shared/logically zoned storage infrastructure (clients receive dedicated LUNs, but data traverses a shared physical SAN).
• Level C: Shared virtual and physical infrastructure, isolation provided by dedicated virtual security appliances (e.g., VM firewalls, IDS, IPS).
• Level D: Shared virtual and physical infrastructure, no appliance-based segmentation and isolation (isolation provided via VLANs).

I like the tiered model because it can be easily consumed by an enterprise’s security auditors. Am I being overly simplistic here? Sure. But we need something. And the time to start defining a cloud security model is now. Until that happens, I expect our enterprise clients to continue rolling their eyes each time public cloud infrastructure  is mentioned in a vendor keynote. My sample model is meant to be just a starting point for discussion. We need something, and a tiered model would work.

VMware/Intel Client Hypervisor Announcement

VMware announced collaboration with Intel on a client hypervisor. This announcement was almost identical to a similar announcement from Citrix and Intel last month. To be honest, I was hoping to hear from VMware that they would be doing a little more than Citrix – i.e. – upping the virtual ante. Instead, my impression that was Intel is treating VMware and Citrix as equal citizens. Nothing in the VMware-Intel announcement shows any more innovation by VMware over Citrix. Still we’re far too early down the client hypervisor road to declare a leader, since neither Citrix nor VMware is shipping a product. In fact, the only vendor shipping a client hypervisor today is Neocleus.

Mega Merger in the Works?

As time continues to pass and VMware looks at its hole in the enterprise management space, I see the question of a merger with a major enterprise management vendor as now being about when and not if. Here’s what I’m thinking – a BMC merger or acquisition with either VMware or one of its major partners (i.e. Cisco) makes sense on a number of levels. VMware needs a management and orchestration stack that addresses the physical infrastructure (and so does Cisco for that matter). If VMware wants to be a major player (and not a role player) in the future data center, it will need an enterprise management solution. If we take this one step further, what would the impact of a VMware-BMC-Cisco-EMC merger be? Alessandro Perilli has done a nice job covering Cisco acquisition rumors on virtualization.info (see here and here). Personally, I don’t see the public mutual respect between VMware, BMC, and Cisco as nothing more than a needed group hug. Is the fact that Bob Beauchamp (BMC CEO) has been predominantly quoted in the recent NY times article on Cisco’s data center plans a coincidence? I doubt it. In fact, I can see a Cisco acquisition of BMC as a real possibility. Stay tuned. The drama that is good enough for a virtualization reality show is only going to intensify in coming months.

Posted by: Chris Wolf

Citrix Essentials for XenServer and Hyper-V: First Impressions

Yesterday Citrix announced the release of Citrix Essentials for XenServer and Hyper-V. Yes – Essentials supports both XenServer and Hyper-V. Prior to the announcement, Citrix gave me the opportunity to evaluate Citrix Essentials in my lab, which I naturally accepted. For background, Citrix Essentials is offered as Enterprise or Platinum editions for Hyper-V and XenServer. Table 1 lists each edition’s features.

Version	Features
Hyper-V Enterprise	- Citrix StorageLink technology - Dynamic Provisioning for Virtual Machines
Hyper-V Platinum	- Lab automation and VM lifecycle management - Citrix StorageLink technology - Dynamic provisioning for physical and virtual machines
XenServer Enterprise	- Citrix StorageLink technology - High availability - Dynamic workload balancing (Q2’09)
XenServer Platinum	- Lab automation and VM lifecycle management- - Citrix StorageLink technology - Dynamic provisioning for physical and virtual machines - High availability - Dynamic workload balancing (Q2’09)

Table 1: Citrix Essentials editions and related features

Storage management is provided by a new product – Citrix Virtual Storage Manager (CVSM) and its underlying StorageLink technology (more on that in a minute). Dynamic VM and physical system provisioning is provided by Citrix Provisioning Server. Lab automation and VM lifecycle management is provided as an OEM version of VMLogix Lab Manager and Lifecycle Manager.

Citrix has the lead today with dynamic provisioning since VMware has no tool for automatically provisioning bare metal servers. However, that will be short-lived in my opinion; I think a stateless ESX server that loads on the bare metal (in physical RAM) is an inevitable part of VMware’s future. Still, our world is not yet entirely virtual, and Citrix Provisioning Server can provision any server role (not just hypervisors).

Now I would like to spend a few moments on my initial impressions of CVSM and StorageLink, which I had the opportunity to evaluate for a little more than a week in my lab. CVSM (shown in Figure 1) solves the major Hyper-V storage management hurdles by streamlining LUN provisioning and VM deployment.

Figure 1: CVSM user interface

Consider StorageLink to be the glue the connects physical storage assets and virtual infrastructure management. StorageLink consists of the following components:

• StorageLink Gateway - enables automated discovery and one-click access to native storage services (e.g., DAS, NAS, and FC or iSCSI SAN).
• StorageLink Resource Manager - makes common storage array administrative actions (e.g., partitioning, snapshots, replication, and backup) visible from within the virtualization management environment.
• StorageLink Image Manager - manages a centralized library of VM images that can be instantly converted between XenServer and Hyper-V, and rapidly cloned and delivered to any number of target hosts using array-level features.
• StorageLink Connect - provides a set of open APIs that integrate XenServer and Hyper-V environments with third party backup solutions and enterprise management frameworks.

Tighter storage integration is a big deal, as LUN provisioning can easily drive up virtualization TCO and is often cited by VMware as one of the benefits to VMFS. In fact, about a month ago, one of my friends who manages a virtual infrastructure consisting of nearly 1,000 VMs scoffed at the idea of provisioning a LUN per VM with the thought “Who on earth would want that headache?” The LUN-per-VM model does have its merits, including performance that can be gained through I/O caching at the array (when dozens of VMs share a LUN, it’s hard for the array to make sense of I/O traffic and as a result offer any caching benefit). Note that LUN-per VM (either via a dedicated VMFS volume or as a mapped raw LUN) is also available with VMware ESX infrastructures.

StorageLink currently integrates with arrays via SMI-S and offers custom integration for NetApp, Dell EqualLogic, and HP storage. Note that at this point, SMI-S array-level integration is more of a checkbox than a working feature Citrix could not offer any examples of an array they can be fully managed via SMI-S. But really… who can?

Array-level integration is significant. In my evaluation, I was able to leverage the NetApp thin provisioning and snapshot features, resulting in considerable storage savings for my Hyper-V VMs. Snapshot integration allows you to quickly provision VMs by taking a snapshot of a source LUN. The result is that newly deployed VMs can leverage the source snapshot image; this means that storage is only required for new writes to the newly provisioned VM. Here are the major steps for configuring CVSM to rapidly provision new VMs:

1. Using the Hyper-V management tool or System Center Virtual Machine Manager (SCVMM), create a new VM without any hard disks. In my case, I created a new VM named “XP-tmplt.”
2. Install the CVMS service and CVSM management tool.
3. Open the CVSM management tool, connect to the CVSM service, and then add connections to the Hyper-V physical hosts you would like to manage.
4. Create a connection to the storage array you would like to manage using CVSM.
5. Create a CVSM storage repository (see Figure 2). The storage repository includes a defined amount of capacity, capabilities, and LUNs for use by CVSM. In my test, I created a RAID 4 storage repository that utilized the array’s thin provisioning and deduplication features.
6. Use CVSM to allocate a storage volume from the newly created storage repository. The storage volume will act as a passthrough (e.g., raw LUN) for the baseline template VM. 
7. Use CVSM to associate the storage volume created in step 6 with the VM created in step 1.
8. Use the Hyper-V or SCVMM console to boot the VM (e.g. XP-tmplt) and install the base OS image, along with any common applications. In my case, I installed Windows XP SP3, created a sysprep answer file (sysprep.inf), and then sysprepped the VM (which automatically shut down the VM).
9. Use CVSM to create a new storage profile that references the volume created in step 6 (see Figure 3). Note that the volume now contains the sysprepped Windows XP image.
10. Use CVSM to create a new VM template (Figure 4). The VM template sets the baseline configuration (e.g. memory, CPU, guest OS) for a VM and associates the VM with a storage profile. The VM template is what is used to quickly provision new VMs.
11. Use CVSM to deploy new VMs from the VM template created in step 9 (see Figure 5) using the array’s snapshot features. You can use CVSM to deploy as many VMs as you like. Each deployed VM will contain a numerical suffix, starting with 00. In my test, I deployed five VMs, and CVSM named them BG-RIC.00 to BG-RIC.04. The five VMs were created and ready to power-on in SCVMM in less than 30 seconds! Figure 6 shows the configured VMs in the SCVMM UI. 

Figure 2: Creating a new storage repository

Figure 3: Creating a new storage profile

Figure 4: Creating a new VM template

 

Figure 5: Deploying new VMs from a VM template

 

Figure 6: The deployed VMs will automatically appear in the SCVMM UI

As you can see, this is a massive improvement for Hyper-V VM provisioning on networked storage. My experience with the software was basically what I’ve come to expect from beta software, and did include one hurdle to overcome – CVSM provisioned storage and created VMs without a problem, but did not create the VM passthrough disks and associate the correct LUNs with each. I had to do that step manually using the Hyper-V manager tool. Note that other beta testers also identified this problem, and it has been fixed. Also, only passthrough (raw) disks are supported today; virtual hard disk files are not supported. I’m hoping that virtual disks will be supported for Windows Server 2008 R2 cluster shared volumes, once they’re available.

Still, after doing the initial work to create the storage repository and VM template, VM deployment was a piece of cake. And what' not to like about spinning up a bunch of new VMs in seconds? Sure, VMware ESX has had this integration with NetApp for awhile now, but that’s not the point. Even if CVSM is a little rough around the edges, it’s a product that is on the right track. We expect serious Hyper-V adoption to start after the Windows Server 2008 R2 release, so Citrix has some time to smooth out CVSM and add some more supported arrays to the mix. This type of technology will be welcome by our enterprise clients who have previously resulted to internal scripting in their efforts to automate LUN provisioning. Naturally, VMware could add a similar feature to vCenter.

Of course, there’s more to provisioning storage than simply allocating a LUN on an array. In fibre channel SANs, you need to worry about zoning and LUN masking too. I asked Citrix’s Pete Benoit about switch and array level integration with regards to LUN provisioning, and this was his reply:

FC zoning is in the V1.0 product.  It requires “credentials” being added for the fabric SMI-S component. We use Brocade’s SMI-S provider (Add those in the GUI and it will discover the fabrics).  For V1.0 if the fabric credentials are not set then the user is expected to have done zoning. As you are aware many enterprises only do FC zoning at well defined change management points. We support both auto-zoning and manual zoning.

The product does the LUN masking in that context. There are also CLI tools to map storage to an HBA/Initiator, if the HBA has not been zoned the action will be performed at that time. If the fabric credentials exist (set by the user) then zoning is done automatically (including multi-pathing) by detecting multiple fabrics that the HBA & Target are visible.

While CVSM features will improve, what I would really like to see is better integration with SCVMM. Ideally, CVSM should be able to be added to SCVMM as a plug-in, but the SCVMM management interface is not extensible - something I’m hoping that Microsoft will change by the next SCVMM release.

VMworld is off to a good start. The Citrix Essentials announcement is the first of what is shaping up to be many exciting announcements this week.

Posted by: Chris Wolf

Production-class Hypervisor Evaluation Criteria

Today at VMworld Europe in my “Hypervisor Competitive Differences: Beyond the Data Sheet” session, I had the pleasure of announcing Burton Group’s production-class hypervisor evaluation criteria. We are frequently asked by our clients for guidance with hypervisor evaluations. As a result, Richard Jones, Drue Reeves, and I spent three months developing criteria to evaluate hypervisor platforms. Along the way, we received considerable input from Burton Group’s Data Center Strategies, Security and Risk Management Strategies, and Network and Telecom Strategies services, as well as input from a number of Burton Group clients.

So what’s all the fuss about? In summary, Burton Group has defined requisite hypervisor criteria in areas such as high availability, networking, storage, and management. Burton Group considers hypervisors that are unable to meet the requisite criteria as not enterprise production-ready. We have also defined preferred and optional features as well. The complete evaluation criteria and technical justification for each, along with checklists that can be used to write RFPs, is in our 70 page Server Virtualization Hypervisors technology and standards report, which will be available the first week of April. A series of product profiles will follow the release of the Server Virtualization Hypervisors document, including analysis of Citrix XenServer, Microsoft Windows Server 2008 Hyper-V, Virtual Iron, and VMware vSphere 4.0.

I have said on numerous occasions that server virtualization is a core infrastructure technology, and as such requires a long term commitment. Our evaluation criteria will help you identify the features that really matter to your organization, and also provide the depth to differentiate between mere marketing checkboxes and enterprise-ready hypervisor features. If you would like to see a sneak preview of the evaluation criteria, take a look at my 15 minute summary presentation below. In the presentation I describe each of the required hypervisor features that were identified in the report. 

Posted by: Chris Wolf

Citrix and Intel Take Major Step on the Path Toward a Client Hypervisor

Today Citrix and Intel announced a partnership that will bring embedded client hypervisors to market by the end of the year. This is part of the Citrix "Project Independence" client solution.

Those that attended our Catalyst Europe conference last October saw Citrix's Ian Pratt conduct a live demonstration of a Xen-based client hypervisor. Following the demonstration, I asked Ian about Citrix's plans to bring such a solution to market. Ian proceeded to make his best poker face and replied "We currently have no plans for a client hypervisor." Members of the audience (which included VMware CTO Steve Herrod, Neocleus CTO Etay Bogner, and Virtual Computer CEO Dan McCall) weren't convinced, and neither was I.

Here's a basic view of the client hypervisor architecture.

The idea is pretty simple (in theory) - run a hypervisor on the client device that allows you to partition it into multiple virtual desktop systems. I say "in theory" because getting the hypervisor to work with a myriad of client-side wireless adapters, video adapters, and sound cards, for example, is no easy feat. Also, "getting the hypervisor to work" with devices like video adapters alone isn't enough. A user with the latest video adapter may want to leverage its full power to run games in his personal VM, for example. It's important to note that the Citrix-Intel announcement involves running a bare metal Xen hypervisor, which is far different than a hosted (Type II) hypervisor such as VMware ACE, VMware Workstation, or Virtual PC. With the hosted hypervisor model, the hypervisor runs on a host operating system (e.g., Windows XP), and unless the endpoint OS is fully trusted, questions of risk will always arise. With the bare metal hypervisor, the only code between a corporate (IT-owned) VM and the hardware is the hypervisor. TPMs could also be leveraged to validate the authenticity of the hypervisor.

Users can maintain a personal virtual desktop VM and would be free to install personal software on it as well. Furthermore, IT could implement policies that only allow the corporate VMs to connect to the organization's internal LAN. You could get creative with personal VMs as well, such as giving them wireless access to Internet resources via an isolated network subnet (similar to how visitors may be granted Internet access today).

The client hypervisor also opens doors to employee-owned PCs, where employees can receive vouchers and select from a list of allowable laptops, for example. In such cases, hardware and personal VM support could be turned over directly to the laptop OEM. This frees the IT staff to only have to worry about the corporate VM.

Let's take this a step further and include a synchronization component within the desktop hypervisor. This could allow a user's personal VM data to synchronize with a replica VM in the cloud. That would allow a user to access his personal VM from any browser-based device, including a PC, iPhone, Blackberry, or smart phone. Replication could be configured to allow file recovery and desktop VM rollback (in the event of corruption or configuration error) as well. IT could follow a similar model by allowing the user's corporate VM to synchronize with a duplicate copy hosted on a server in the organization's "internal cloud." Unlike traditional VDI models, with the client hypervisor, the virtual desktop would execute locally on the endpoint system, hence reducing the number of servers required to maintain the virtual desktop environment.

If you haven't already noticed, I'm very high on client hypervisors. Still, we do need to keep the hype in check. For starters, I see the client hypervisor as something that all desktop and laptop vendors will eventually ship on a flash chip on most desktop and laptop systems. I'm optimistic that laptop and desktop vendors will offer the ability to change hypervisors with a simple flash update. VMware first mentioned their work on a desktop hypervisor at VMworld North America last September, and I think it's a safe bet that they will be bringing a desktop hypervisor (as a downloadable package and as a vendor OEM) to market as well. To not make such a move would all but concede the desktop space to their main rivals.

To me, the embedded client hypervisor is an inevitable part of our future, with several options available by early 2010. When Citrix acquired XenSource in 2007, Citrix CEO Mark Templeton confidently stated "The virtual desktop market is ours to lose." The Intel partnership, without a doubt, gets Citrix closer to Mark Templeton's assertion. Still, the game's far from over. Burton Group clients with substantial VDI deployments have been evenly divided between Citrix and VMware. However, the majority of our clients with VDI plans remain undecided, and they are looking for vendors to give them concrete reasons to commit to a particular solution. As with their server virtualization investments, our clients view client virtualization as a long term strategic commitment and want to choose the platform they can confidently move forward with over the next several years. Many of our clients have interest in leveraging employee-owned PCs, and today's Citrix announcement plays right into those plans.

Now I'd like to hear what vendors with competing solutions have to say. So VMware, Neocleus, and Virtual Computer, what do you think? For our clients and interested readers, I would also love to hear your thoughts. Is the client hypervisor part of your 2-5 year strategic plan for desktops?