By The Bell

Intel’s Nehalem virtualization game-changer

A client recently purchased a 4-CPU Dell R900 to expand its VMware environment rather than the 2-CPU Dell R710 because the IT staff felt it would scale better. I hope it's not too late for them to return the server.

The R710 may have only two CPUs, but these are the 11 Generation Intel Xeon 5570 CPUs. The revolutionary Intel 5500 series CPU (Code named Nehalem) was designed to optimize virtualization performance and includes Intel Virtualization Technology for Directed I/O, QuickPath Interconnect, Integrated Memory Controllers, Hyper-Threading, hardware-based memory protection, Extended Page Table, and much more.

Dell's Web site says that the "2-socket PowerEdge R710 achieved the best performance for virtualization, better than all other 2-AND 4-socket servers." This claim is based upon VMmark scores which show a 2-CPU (8-Core) R710 handling slightly more VMs than a 4-CPU (24-Core Xenon 7460) R900. (While the R900 was running ESX 3.5 rather than vSphere for the R710, the results were similar to the AMD based 4-CPU, 16-Core R905 running vSphere).

Dell's server configuration tool shows a cost of $17,203 for a vSphere-ready R710 with 96GB of RAM versus a purchase price of $26,173 for a similarly configured R900 with four 6-Core CPUs. While the R900 scales to 256GB vs. only 144GB for the R710, this adds $19,113 of cost versus $966 to scale the R710 to 128GB. If 256GB is required, it makes more sense to simply purchase a second R710 (or Cisco UCS – see "Alternatives to Dell" below). Another advantage – the R710 uses only 2U of rack space vs. 4U for the R900.

The purchase cost disparity, advanced capabilities and reduced rack space requirements make the R710 the obvious choice as a virtualization host. But using only two CPUs provides further software licensing benefits. Here is a recap comparing three year licensing costs assuming the host is running both VMware vSphere Enterprise Plus and Cisco Nexus 1000V along with both Microsoft Windows Server Data Center Edition and SQL Server Enterprise Edition. The Microsoft products are both licensed according to the physical host CPUs and allow an unlimited number of the virtualized instances of the software to run on the host.

 

Alternatives to Dell

HP and IBM also offer Intel 5500 based servers. If the virtualization environment is sizable, however, we recommend evaluating the Cisco UCS as an optimized virtual infrastructure hosting platform. The superior UCS performance and unified fabric capabilities can further reduce licensing expenses along with rack space, power and cabling costs. Its integrated management capabilities help drive complexity out of operating the environment.

Will UCS Unify IT Staffs?

This article also is available on www.dabcc.com

Traditional data centers tend to be segregated into functional silos - a model that has endured in the physical world largely because of the crisp lines of demarcation. Virtualization is shaking up the status quo and increasinng demand for more efficient IT processes. Cisco's UCS helps automate and streamline the stovepipe model by unifying network, storage and servers with virtualization. It also provides an opportunity for IT organizations to begin chipping away at those silo walls.

 

Dysfunctional Computing

New IT infrastructure tends to be project driven. Large application implementations and other departmental endeavors provide budget for servers and SANs – filling up data centers with a hodgepodge of technology islands in the process. Gartner's number one energy saving recommendation at its 2007 Data Center Conference was to turn off servers that appear idle and see if anyone complains (searchcio.com: Top 10 ways to save energy in the data center).

Function tends to follow form with technology specialists typically managing their own domains in relative isolation. Server administrators are unconcerned with network switch configuration. Storage administrators dole out LUNs as they see fit. The network team controls all traffic. An inherent disregard of the bigger picture excludes efficient leveraging of infrastructure across the organization. Even architectures originally deployed with a holistic perspective tend to diverge as upgrades occur at a functional level.

The requirement to support a multitude of operating and management systems, protocols and equipment drives up costs while serial processing of technology requests leads to lengthy delays. The stovepipe model saving grace is that its shortcomings are often overshadowed by the even greater inefficiencies of a physical data center. It can, for example, take weeks of administrative time to provision new servers – making storage and network resource delays moot.

These same delays become excruciating against the backdrop of a virtualized data center where the server team can provision virtual machines in a matter of minutes. Virtualization also blurs the comfortable lines of functional segregation. Server administrators may need a separate VLAN to enable vMotion or extra SAN spindles to enhance performance. They configure virtual switches, argue about protocols and can take a LUN off-line by over-provisioning VM storage.

The silo model drawbacks are amplified as IT organizations deploy VMware vSphere, the Intel 5000 CPU and the Nexus 1000V virtual switch to obtain 100% data center virtualization. As physical servers become unavailable, IT can no longer circumvent bureaucracy on the fly.

Optimizing Virtual Machine Hosting with UCS

Just categorizing UCS is generating a lot of controversy. Last week HP called it a "giant switch". Howard Marks of Byte and Switch recently called it, "next year's servers…shipping this year". According to ComputerWorld, a Dell executive referred to it as a "one-size-fits-all blade server".

UCS can only be satisfactorily defined within the context of the virtualized data center that it was designed to optimize. It is a specially built hosting platform with innovative advances in both performance and management.

Cisco's patented technology delivers far more memory at lower prices than traditional servers. UCS's unified fabric capabilities resolve the cabling nightmare of hosting many virtual machines on a single server. Its Palo adapter supports up to 128 vNICs and vHBAs – providing select virtual machines with native storage and network characteristics.

UCS revolutionizes virtual infrastructure management by allowing the server administrators to dynamically assign resource pools created by the storage and network specialists to a physical blade. All teams share a common GUI and have visibility into what other groups are doing, yet their role specializations are preserved.

Transitioning to a Data Center Team

Resource pools are enabled through comprehensive service profiles that define UCS blades through software. Initial collaboration of the functional specialists is not only required for their creation, but for ongoing troubleshooting. Zoning and vSAN modifications, for example, should be communicated to the server and network teams. Changing a vHBA should be timed so as to not reset a production server outside of a maintenance window. The server team needs to monitor space utilization and performance of their systems to ensure LUNs are "right sized" by the storage team.

Specialists will likely come under pressure to, at minimum, pick up a strong secondary skill set. The server team needs some understanding of IP routing. The network group should know how to do minor server troubleshooting and basic reinstalls. Storage specialists need familiarity with proper VMFS sizing and best practices. All three groups should understand the implications of how security affects their domains.

The challenge and opportunity for IT management is to institutionalize regular communication, drive appropriate issues to a big-picture decision-making level and utilize increased accountability to promote cross-training. A good first step for many larger organizations is to restructure the storage, network and server groups to directly report to the same person. Creating a data center team will help enable a more efficient operation that is better able to optimize the extensive UCS capabilities.

Chris Reed, Mondy Carpio and Steve Jones of INX were invaluable contributors to this article.

Is VMware More Like Novell or Oracle?

David Cappuccio, in his blog for Gartner, Just a Thought; Will VMware become the next Novell?, draws several disconcerting parallels between VMware's current virtualization industry dominance and a similar position enjoyed by Novell in the early 1990's. Cappuccio wonders if VMware, like Novell, is going to get trounced by Microsoft.

The $60 billion Redmond behemoth is one of the greatest companies of all time and has displaced many industry leaders as viable players. Though it would be foolish to rule Microsoft out of any fight, differences in today's technologies and between the VMware and Novell organizations mean that history is not doomed to repeat.

Virtualization as a Data Center/Cloud Computing Platform

 

Microsoft touts Hyper-V as an operating system feature and arguably views virtualization as a point solution. VMware, on the other hand, is all about virtualization as a platform. This broader perspective has enabled VMware to dominate the market with exceptionally reliable, robust and versatile solutions.

 

While perhaps taking some liberties with terminology, VMware’s declaration of vSphere as a cloud operating system does emphasize the importance of virtualization in enabling a shared resource-on-demand cloud model. As a data center approaches 100% virtualization, embracing cloud computing becomes easier. VMware’s vSphere delivers not only the performance required for 100% virtualization, but also the crucial storage, network, security and management components.

 

Data center virtualization has an exceptional and easily measurable ROI. Achieving the full savings and other benefits, though, requires a homogenized virtualization platform. IT staffs will likely be reluctant to permit pockets of Hyper-V, even if only for test/dev or non-mission critical servers, making it more difficult for Microsoft to establish beachheads against VMware as it did against Novell with NT.

 

VMware’s tight partnership with Cisco adds a further enterprise credibility that Novell lacked. Cisco’s Nexus 1000V switch is practically a given for any Cisco shop wanting to consolidate virtual and physical network management, and Cisco’s UCS optimizes the effectiveness and efficiency of virtual infrastructures. The 1000V only works with vSphere while UCS is tightly integrated with the VMware platform.

 

The Difference between VMware and Novell 

 

In 1994, I ran the 2^nd largest Novell dealer in the San Francisco Bay area. Steve Ballmer invited me and a handful of other Novell partners to lunch in order to hear his pitch on why we should sell NT 3.5. I was particularly impressed by his response to a question about the impending Novell-Word Perfect merger. “We think it’s great!” he boomed. “WordPerfect has these huge budgets. Novell has tiny little budgets. The only thing those companies have in common is that they’re both headquartered in Provo, Utah. It will never work!”

 

Ballmer’s prescience quickly became apparent as Novell sold WordPerfect to Corel two years later. The acquisition was a casualty of Novell CEO Ray Noorda’s resolve to beat Microsoft at any cost – including purchases of Quatro Pro, DR DOS and even ATT UNIX. But by neglecting its core networking competency, Novell opened the door wide for NT to gain a foothold. 

 

Unlike Noorda’s Novell, VMware’s CEO, COO and Executive VP all come from high-level positions at Microsoft. They understand the Microsoft culture, strengths and threat. VMware is a company with a laser beam focus on virtualization that is undistracted by a personal vendetta.

 

Lack of Microsoft Invincibility

 

At the Microsoft MVP Meetings in the early 2000’s, Steve Ballmer commonly dismissed Oracle as a competitive threat, stating that SQL Server had already won. Yet Oracle now sits as one of the three largest software companies in the world with over $22 billion in revenues. Battling Oracle along with other exceptional competitors such as Google and Nintendo will prevent Microsoft from focusing its vast resources only on virtualization.

 

The ongoing improvement of Hyper-V and other hypervisors make a sustained 90% VMware market share unlikely. But VMware’s industry leadership need not diminish. The key will be its ability to continue innovating with virtualization solutions that increasingly obsolete the physical alternatives. As VMware drives the virtualization industry expansion, it will benefit in an absolute sense with continued rapid growth.

 

**** 

This article also available on www.dabcc.com.  Named by VMTN as one of the top 5 weekly blog posts.

http://blogs.vmware.com/vmtn/2009/07/top-5-planet-v12n-blog-posts-week-27.html

Will Cloud Computing See an ASP-like Burst?

The May announcement of vSphere’s debut elevated its status from a mere Virtual Data Center OS to its lofty designation as a Cloud Operating System. One can hardly fault VMware, though, for joining the marketing fray. Even Oracle’s Larry Ellison succumbed as he acknowledged, “We'll make cloud computing announcements. I'm not going to fight this thing. But I don't understand what we would do differently in the light of cloud."

Nearly every manufacturing and consulting firm in IT has seemingly hitched its wagon to the cloud as the hype has grown toward epic proportions. Just try to obtain a unique URL with anything cloudlike: “Versacloud.com” – Ha! “Silverlining.com” – Please. “Muteablecloud.com” – Forget about it. The media and research firms are all abuzz as well. Gartner says that cloud computing will hit $150 billion by 2013.

But then, we’ve seen this before.

The ASP Industry Implosion

A decade ago, ASPs suddenly emerged as the inevitable alternative to more expensive, less predictable and less responsive internal data processing architectures. Gartner Group boldly predicted that ASP revenues would jump from $900 million in 1998 to $23 billion by 2003. The research firm gushed, “Despite market debate, the shift toward the ASP model is inevitable…we believe this model will survive and flourish, simply because the benefits are indisputable.” Gartner was far from alone in its optimistic predictions. IDC projected a $24 billion ASP industry by 2005 while Zona Research estimated it could reach $48 billion by 2003. 

For fiscal year 1999, the top five publicly traded ASPs had combined losses of $485 million versus sales of only $464 million. Despite these horrific numbers, their total market cap reached an astounding $23.6 billion by March of 2000 (Table 1). Clearly, Wall Street shared the research organizations’ exuberance for the ASP model.

Unfortunately, someone forgot to tell the customers. While total fiscal year 2000 sales of the top five publicly traded ASP firms climbed slightly to $547 million (most of which was not due to hosting), total losses soared to nearly $1 billion. Not surprisingly, the subsequent burst of the ASP bubble was far more dramatic than the dot com collapse. By 2005 three had gone bankrupt, and the remaining two had combined sales of $135 million and a total market cap of $111 million – a 99.5% decline in total valuation.  

ASP	12 Month Revenues	12 Month Earnings	March 2000 Market Cap
FutureLink	$  99.9	-$95.2	$  2,147
Breakaway	$  98.6	-$43.5	$  3,680
Interliant	$128.8	-$125.1	$  2,294
NaviSite	$  49.8	-$  58	$  8,790
USInternetworking	$  86.9	-$163.6	$  6,686
Total	$464.0	-$485.4	$23,567

  Table 1: Top 5 publicly traded ASP financial metrics as of March, 2000

Cloudy Forecast

Cloud computing may be all the rage in the media, but market valuations of cloud providers are hardly in the stratosphere. VMware, for instance, recently paid $20 million for 5% of cloud computing provider, Terremark. This $400 million valuation is quite reasonable when considering Terremark’s rapidly increasing revenues and EBITDA which were $250 million and $51 million respectively for the last 12 months.

The overwhelming hype around cloud computing compels CIOs to evaluate it, if for no other reason than being able to demonstrate their diligence. But that doesn’t mean they’re clamoring to roll out cloud services. On the contrary, we’re seeing an unusual skepticism regarding claims of security and SLA capabilities. Organizations want to understand how they can optimally delineate internal versus external computing along with the trade-offs between cost, performance and risk.

Unlike the ASP industry which was doomed from the start by technology limitations, virtualization provides cloud computing with the foundation for technological success. The healthy caution with which organizations are approaching clouds, undoubtedly motivated at least in part by the difficult economic climate, will in the long run help shape a more practical and resilient cloud computing industry.

Why Cisco UCS is Kind of like the Kindle

I've been reading Cisco's new book on UCS titled: Project California: a Data Center Virtualization Server. The book is exceptional, but the title concerns me. The media and potential UCS customers already struggle with how to categorize UCS. By referring to it as a "server", authors Gai, Salli and Anderson may create further confusion. The term doesn't do justice to the new category that UCS creates – a compute platform unifying network, storage and server virtualization.

While obviously very different types of products, UCS and Amazon's electronic book reader nonetheless have quite a bit in common. The paperback sized Kindle 2 combines innovative display, wireless and low power technologies to provide an optimized hosting platform for electronic books. It eliminates the printing, distribution and storage costs inherent in the old-fashioned paper versions. UCS similarly incorporates innovative technologies in memory, network I/O and unified fabric management to optimize the virtual environment. It facilitates a pervasive operation of agile virtual machines that are much less expensive to deploy and manage than physical servers.

Kindle 2 is still rudimentary in some respects such as lack of a touch screen, but it makes so much sense that you know the traditional printed book market is in for long-term trouble. UCS similarly is going to redefine the very concept of "servers" which will increasingly become associated with virtual machines residing on a unified hosting platform. And the desktop market is not far behind. The "desktop" is rapidly being redefined as a virtual machine residing in the data center that is accessed with a PC, laptop, Mac, thin-client or zero-client device.

Hopefully I will be able to purchase the next edition of the UCS book for my Kindle – the printed version is so passé. I have to highlight sections using a marker and look up words with a dictionary. I even had to make my own bookmark from a magazine flyer since I threw away all my paper versions months ago.

Don’t Believe Any Numbers You Don’t Make up Yourself

Microsoft, in emphasizing the value of Windows Server, has stated that open-source Linux is "free like a puppy". It now argues the other side against VMware by constantly equating the "free" licensing cost of Hyper-V with value. The comparison fails to consider how the respective platforms stack up in their ability to assist organizations in achieving their ultimate virtualization objectives.

Where's the Beef?

McDonald's founder Ray Kroc liked to claim, "We take the hamburger more seriously than anyone else". VMware could say the same thing about virtualization. Every page of its Web site educates viewers on different aspects of virtualization, but the term is not even mentioned on Microsoft's home page. Navigating to Microsoft's virtualization section reveals the biggest emphasis to be on price comparisons between Microsoft and VMware. Videos, white papers, blogs, case studies, analyst quotes and even Microsoft's ROI calculator tool underline the cost competition with VMware.

VMware's suite of network, management and automation tools enable a virtualized data center, disaster recovery solution and desktops. Microsoft, alternatively, emphasizes its ability to address both physical and virtual infrastructure. While VMware stresses the superior performance, security and reliability of its hypervisor architecture, Microsoft promotes as advantageous that Hyper-V is incorporated as a feature of the operating system.

Unlike an operating system running an individual application on an isolated server, virtualization integrates formerly disparate silos of compute, storage and network resources. Virtualizing a production environment results in zero tolerance for issues in performance, reliability or security that now affect the entire IT operation. While the many unique capabilities of vSphere such as the distributed virtual network switch, storage live migration and integrated VM level fault tolerance may be argued to be essential for a virtualized data center, nothing is more important than reliability. Duncan Epping of yellow-bricks.com fame gave a succinct response when asked why he used VMware: "Because it's rock solid!"

The Microsoft ROI Calculator

Microsoft's Web site audaciously claims that Hyper-V is one-third the cost of VMware. As Table 1 shows, the cost of Windows licensing is factored in when purchasing VMware although this is only a requirement when running Windows VMs. It also disregards the obvious deployment of Windows Server Datacenter Edition which allows unlimited Windows Server instances on an ESX/vSphere host. Microsoft does concede in its "one-third the price" video that the Hyper-V to VMware cost ratio increases to 50 per cent if using Datacenter Edition.

Table 1: From Microsoft Web site showing Hyper-V 1/3^rd the cost of ESX (ESXi is a typo)

Microsoft also takes the very unusual approach of using its Virtualization ROI Calculator to compare ROI results with ESX. While an ROI metric can be useful in evaluating the economic impact of different platform investments, all of the relevant variables must be factored for each scenario. The Microsoft calculations erroneously treat Hyper-V and ESX exactly the same.

I ran a virtualization scenario on both the Microsoft and VMware calculators for comparison purposes. Both tools have a similar look and feel which is not surprising since they were created by Alinean, and both show a very similar 3-year net cost reduction from the physical scenario. I used the following parameters, but otherwise Microsoft default figures wherever possible:

• Analysis period: 3 years
• 350 total physical servers: 100 2CPU/1 core, 200 2CPU/2 core, 50 4CPU/2 core
• Virtualization hosts: 2 CPU / 4 core
• Users: 5,000 (the Microsoft tool requires an input for # of users in order to work)
• Cost of capital: 0 (for simplification purposes)
• Cost variables to consider: Existing servers only
• Virtualization: 100% (this percentage is fixed for both tools)
• VMs/host: VMware uses a ratio of 8.31:1 (consolidating to 42 hosts), while Microsoft uses a ratio of 5:1, consolidating to 70 hosts (for both Hyper-V and ESX).

The output from the Microsoft ROI tool, shown in Table 2, has a 3-year net cost savings for both the Hyper-V and vSphere scenarios of $2,083,449 – but an initial Hyper-V 3-year investment of $399,801 vs. $1,448,561 for vSphere. Table 3 shows a breakdown of the investment components for each solution.

Table 2: Output from Microsoft ROI Tool with 350 servers, 5,000 users.

Analyzing the Microsoft Calculator Investment Figures

INVESTMENT
		Hyper-V	vSphere
Windows Server Ent. Licensing		$ 88,113	$    88,113
CALs		$   -	$    78,400
ESX Licenses		$   -	$  402,500
vCenter		$   -	$   5,000
Backup Software		$   -	$  287,000
Storage		$ 73,500	$    73,500
Professional Services/Installation		$  238,188	$ 238,188
Training		$   -	$    10,000
Total Initial Investment		$  399,801	$ 1,182,701
Total 3-Year Investment		$  399,801	$ 1,448,561
ROI (over 3 years)		521%	71%

 Table 3: Investment Summary from the Microsoft ROI Tool

Windows Server Enterprise Licensing

The Microsoft tool assumes that all hosts run Windows guest OS's, and then calculates Windows Server Enterprise rather than Datacenter because of an assumed server consolidation ratio of 5:1 (which makes Enterprise less expensive). If the ratio is larger than 5:1, it quickly becomes advantageous to utilize Datacenter Edition.

Although the Microsoft tool assumes identical hosting ratios between Hyper-V and ESX, evidence suggests otherwise. Evaluations by our engineers indicate a ratio of about 2:1 in VMs per host under ESX versus Hyper-V. A quick Google search shows this ratio or higher to be commonplace. A ratio ranging between 1.5:1 and 2:1 was the conclusion of a VMware funded study by the Taneja Group. The 2:1 ratio is supported by comparing the internal consolidation ratios achieved by Microsoft as outlined in its January 2009 Technical Case Study as opposed to those achieved by VMware as presented in its January 2009 Webcast by CIO Tayloe Stansbury. A ratio of 1.67:1 is obtained by using the respective consolidation ratios from the VMware and Microsoft calculator tools.

A larger hosting density for VMware significantly lowers both the investment and three-year operational costs by reducing the number of virtualization hosts and their associated expenses. It also reduces licensing requirements for Microsoft products such as Windows Server Data Center Edition and SQL Server Enterprise.

CALs

$80,000 worth of CALs are applied to the VMware investment cost, but not to Hyper-V. I can find no reason for ascribing CAL costs to the VMware scenario, and assume it is a mistake in Alinean's methodology.

ESX Licenses

The "initial" cost of ESX licensing includes three years of annual subscription and support (SNS) at full list, although VMware offers a significant discount for purchasing SNS up-front. A greater hosting density would result in a decreased cost for both licensing and SNS.

Backup Software

In Microsoft's view, each ESX host requires an additional $4,100 in management/backup software to put it at parity with Hyper-V. Although the aggregate cost of $287,000 equates to 20% of the total VMware investment, I was not able to find an explanation for it.

Storage

The cost of storage appears to be significantly understated for typical virtualization situations. The VMware calculator tool reflects an estimated storage cost of $432,000 for the same scenario. VMware says that vSphere capabilities such as thin provisioning now reduce storage costs by over 50%, a claim appearing to be corroborated by testing results beginning to emerge from the field. For an organization with significant storage requirements, vSphere may have a substantial storage cost advantage over Hyper-V.

Professional Services/Installation

Microsoft likes to tout that Hyper-V is based on familiar Windows interfaces and works with well-known Windows-based technologies. Most midsized and larger organizations already utilize VMware meaning that IT is familiar with its technology and management tools. As an organization climbs closer to 100% virtualization, it is able to leverage its staff expertise to produce still greater gains with minimal further investment in training.

Training

The Microsoft tool applies a cost of $10,000 for training to the VMware scenario but not to the Microsoft scenario (though a note confirms it should be applied to both). This appears to be another Alinean error.

Return on Investment

The cost of the respective investment for both the vSphere and Hyper-V scenarios is subtracted from the savings to result in the net savings ($2,083,449 in Table 3), but then the cost for the Hyper-V investment is subtracted again from both numbers when calculating the ROI. This is clearly an Alinean computational error.

100% Data Center Virtualization

Up until now, most organizations have generally been reluctant to virtualize the majority of their production servers. My soon-to-be-published article, Perfect Storm for a 100% Virtualized Internal Cloud, describes how vSphere 4 provides the performance, fault tolerance, dynamic resource allocation, networking capabilities, security, management and API components necessary to successfully obtain 100% virtualization. The greater the percentage of data center virtualization, the more an organization is able to save and consequently leverage its virtualization investment.

VMware says its internal IT will be 100% virtualized by the end of the month in contrast to Microsoft which says it is shooting for an internal virtualization ratio of 50%. Although Hyper-V R2 will resolve some of the challenges Microsoft's IT department faces today, it will still be hampered by limitations such as an inability to apply network and security policies that are live migration aware. Even if a Hyper-V organization could virtualize 80% of its servers as opposed to 100% for vSphere, it would still suffer a markedly higher operating cost over the analysis period. The number of physical servers in a Hyper-V scenario is increased still further if an organization uses Linux other than SUSE.

Does ROI Matter?

Virtualization is a rare IT technology that provides a "hard" ROI without quantifying intangibles such as lowered risk and increased agility, or even IT staff benefits. Consolidating servers by 90% or more slashes associated costs such as power, cooling, cabling, power whips, generator and UPS slices, rack space, distributed network switch ports, core network switch ports, SAN ports, HBA ports and maintenance contracts. Over a period of a few years, these cost savings can easily amount to hundreds of thousands of dollars for organizations with around 25 servers or more, ranging up to millions or tens of millions for larger organizations.

Running varying virtualization scenarios using my own ROI tools all show vSphere as the superior economic choice. But as the title of this blog says, you shouldn't believe any numbers that you don't make up yourself. I encourage IT leaders considering virtualization to not just rely on manufacturer calculator tools, but to take the time to carefully evaluate the economics of different virtualization solutions in their unique environments.

ROI can be very helpful in deciding whether or not to virtualize, but it is typically not a good metric for comparing different products. The huge economic and other benefits resulting from a successfully virtualized data center inevitably make the cost comparison between individual components irrelevant in comparison to their ability to facilitate the desired outcome.

------------------------------------

Follow-up:  Tom Pisello of Alinean wrote a thorough response to this blog.  Alinean seems to agree with most of my points, and is even "posting an update to proactively address" the issue of CALs (this section of their tool was defined by Microsoft).  Alinean apparently did include notes regarding the backup software (I was unable to find them when writing the blog, but will look again when I get a chance), and found no calculation error with the ROI metric (I went through this carefully before reporting it - but since the error was repeated for both Microsoft and VMware scenarios, it wasn't material enough to warrant me running through all the scenarios again to replicate the issue).

Overall, I thought Tom's response was great - nondefensive and clearly representative of a sincere intention to provide organizations with as accurate results as possible.

Open Letter to Steven Chan of Oracle

Steven Chan is a senior Director in the Oracle Applications Technology Integration group.

Steven,

Your April 10, 2009 blog discusses Oracle's support policies on virtualization products such as VMware and Citrix, referencing Metalink Note 794016.1, which said that Oracle E-business applications are "not explicitly certified, but supported" on "both software and hardware based" virtualization technologies. The link to the note in your blog, though, is no longer active. 

In his May 14th blog, EMC's Jeff Browning reveals why.  He says that as of May 8, Oracle changed its support policy to no longer cover software based virtualization techologies.  Given Oracle's pending acquisition of Virtual Iron, Jeff concludes, "It appears that Oracle is carefully and intentionally excluding VMware from its support statement".  Jeff poses some questions for Oracle including:

• Given that there is abundant evidence from extensive testing (including at EMC) that VMware virtualization of Oracle databases is both viable and compelling, what is your real issue here?

• ...when will you extend official best-efforts support to VMware which you have already done to other equivalent virtualization products?

Many of our clients run Oracle on VMware today, and we anticipate that number to rapidly escalate now that Oracle 11G and even large Oracle OLTB applications run faster as virtual machines on vSphere 4 than they do on physical servers.  The ability of vSphere's Fault Tolerance to enable real-time fail-over of Oracle applicatioins will be another compelling driver to virtualize Oracle on top of VMware.

You make a point of emphasizing your "Blogging Code of Ethics" which specifies that, "This blog's "first priority is the needs of the Oracle community."  Given that your April 10 Blog is no longer accurate, and on behalf of the 18,000 VMware partners, I'd appreciate you clarifying Oracle's support stance so that we in turn can report back to our enormous base of Oracle customers who deserve to know your intentions.

Respectfully,

Steve Kaplan

Microsoft blog on Hyper-V versus VMware pricing

http://blogs.technet.com/virtualization/archive/2009/05/06/microsoft-hyper-v-server-2008-r2-release-candidate-free-live-migration-ha-anyone.aspx

(I wrote the following response to Microsoft's blog, but it turns out MS doesn't post reader comments).

I have spent 20 years developing, writing and speaking about ROI as the appropriate tool for making intelligent IT platform decisions.  TCO is inadequate because it fails to consider the context of the overall savings realized.  Such is the case here.  When organizations successfully virtualize their data centers, they are typically able to save hundreds of thousands, millions or tens of millions (depending upon the size of the environment).  Moreover, because the bulk of virtualization cost is absorbed up-front in terms of hardware (servers/storage), consulting and training – the marginal ROI that results as an organization virtualizes closer to 100% of its data center is even higher.

The enormous savings at stake make it imperative that an organization considering virtualization should evaluate its overall projected savings/cost avoidance versus the overall costs to achieve those savings.  Data center virtualization typically enables a truly amazing reduction in costs relative to the investment, resulting in payback periods commonly less than 12 months.  And these savings/cost avoidances are not the reduced risk or increased agility type of ROI commonly touted.  While virtualization certainly includes these benefits, we’re talking about only very hard savings of hardware, data center space and power that the CFO will validate and sign off on.  When calculating ROIs for our clients, we don’t even bother to quantify the large reduction in IT staff time required http://www.dabcc.com/article.aspx?id=9425, because this figure is a little more fuzzy and can therefore lead to skepticism.  Besides, there is no need – the other savings are compelling enough on their own.

The fist step to a successful virtualization project is therefore to calculate an ROI for which we only need a general idea of the investment in order to derive an order of magnitude for the savings/investment ratio.  Assuming the numbers overwhelmingly justify an enterprise virtualization deployment, the optimal platform (HS/SW) for successfully enabling the virtualization project can then be evaluated.  While the comparative cost of any specific solution can be a factor, much more important is its ability to facilitate the overall goal of 100% virtualiztion with all the performance, reliability, scalability and manageability required. 

 

A lot has been written about the superior capabilities of ESX to any other virtualization platform in these areas including features such as Storage vMotion, DRS, Distributed Power Management, etc.  Even Redmond Magazine, “The Independent Voice of the Microsoft IT Community” gave its 2008 Editors Choice award for the most reliable IT technology to VMware ESX (the IBM mainframe came in #2).  Now vSphere takes these capabilities up a level with new features such as Fault Tolerance, vmSafe, storage thin provisioning and around 150 more.  And the enhanced performance capabilities of vSphere mean that even the largest workloads can now run as fast or faster as virtual machines.

 

One of the major inhibitors to organizations virtualizing their data centers today has been the lack of networking support within any virtualization hypervisor.  This becomes even more important in facilitating a seamless cloud computing strategy.  vSphere includes the vNetwork Distributed Switch enabling QOS and the application of security and network policies that follow virtual machines as they migrate across hosts.  By adding the Cisco Nexus 1000v (which only works with vSphere), network administrators can use the same Cisco command line tools in the same way they manage physical switches.

 

The competition to VMware vSphere isn’t Hyper-V or any other hypervisor, it’s inertia.  By not virtualizing, organizations continue to purchase servers, network and SAN ports, rack space, power and cooling, UPS and generator slices, cable runs, etc.  It makes far more sense for an organization to virtualize today with the best platform available and achieve what is likely to be an amazingly short payback on the investment while also reducing IT staffing requirements.

 

Since I've covered the ROI, I do want to bring up a TCO item you highlighted. You say that you leave the cost of the guest OS out of your calculations because, "That cost is the same whether you're running Hyper-V Server 2008 R2 or VMware so I didn't bother to include it".  This is silly - if we're running a Linux VM on vSphere, we obviously, as opposed to Hyper-V, do not require a copy of Windows Server.

 

Finally, even when evaluating Hyper-V and vSphere on a TCO comparative basis, we still need to consider the context of the overall cost requirements for each.  We estimate that Hyper-V will require twice the number of servers as vSphere to handle a similar number of virtual machines (Microsoft and VMware’s recent documents reflect this ratio today for the organizations’ internal virtualization projects).  This means a requirement not only for twice as many hosts, but also more rack space, network ports, SAN ports, maintenance contracts, generator and UPS slices, etc.  Additionally, Gartner says that it now costs more to power and cool a server than it costs to purchase the machine itself.  So when factoring in all of the additional costs that Hyper-V requires up front, it doesn’t even compare favorably on an isolated TCO basis.

 

Will the Threat of Swine Flu Boost VDI Sales?

Despite the extraordinary hype around VDI and the rapidly burgeoning VDI industry, it’s difficult to find sizeable enterprise deployments.  Part of the issue is that of the two real contenders in the market, VMware’s View 3 has only been shipping for 4 ½ months, while Citrix’s Xen Desktop represents a complete paradigm shift away from the 98.5% of its sales that are Microsoft Terminal Server based.  While it hopefully will pass without much more pain, swine flu unfortunately does have the potential to accelerate deployment of VDI.

Citrix got its big break in 1994 when Novell started referring its resellers to Citrix's OS/2-based remote access product, WinView.  As the Citrix technology matured, though, it was not positioned nor implemented as a widespread hosted desktop solution.  The limitations of Terminal Server made such an environment more confusing for users and harder to effectively administer.  The simplicity of VDI is appealing to users because their virtual desktops have the same look-and-feel as their normal PCs.  It appeals to IT administrators because it eliminates the management challenges of an enterprise Terminal Server-based deployment.   

As with all big IT changes, though, VDI is taking time to work its way into corporate consciousness as a viable alternative to traditional desktop computing.  Just the threat of pandemic disasters such as swine flu may give IT managers an incentive to start setting up virtual desktops for users as a preemptive strike.  VDI would let users work from home or other remote locations as easily as from headquarters.  As was the case with Citrix WinView many years ago, this could spark many more instances of VDI.  The advantage of the new technology, though, is that now IT managers are familiar with the concept and savings of a server virtual machine.  It won't take them long to realize that they can achieve much greater benefits by making their remote access desktop virtual machines the organizational computing standard from headquarters as well.

Check out my article on VDI coming imminently on searchvmware.com.

 

VMware has its Head in the Clouds, but its Feet are Planted Firmly on the Ground

Note:  This blog has been transformed to an article on DABCC.com. 

http://www.dabcc.com/article.aspx?id=10627.  DABCC is a leading virtualization Web site with huge traffic.  Doug is also a very sharp observer who is literally world renown.

 

Microsoft Vs. VMware: Internal Virtualization Usage

Virtualization.info http://www.virtualization.info/2009/04/how-microsoft-and-vmware-use.html and vInternals http://vinternals.com/2009/04/microsoft-myths-and-realities/ provide excellent analyses on recent internal documents/presentations from Microsoft and VMware respectively on utilizing their virtualization products internally.  Microsoft's internal use of Hyper-V from a January 2009 Technical Case Study http://technet.microsoft.com/en-us/library/cc974012.aspx and VMware's internal use of ESX from a VMworld Europe 2009 presentation by CIO Taylor Stansbury http://sessions.vmworld.com/mgrCourse/launchCourse.cfm?mL_method=player.

To summarize, VMware uses server virtualization internally for 97% of its servers, and the remaining servers will be virtualized by the end of this quarter.  Microsoft, on the other hand, expects to have 50% of their server instances running on virtual machines by the end of Q2.  VMware is able to handle twice the number of VMs per 8 core/32 GB host as Microsoft (Microsoft actually uses a combination of 8 core and 12 core hosts, so the discrepancy is even greater).  VMware's deployment has no performance or stability issues and is both flexible and easy to manage.  Microsoft, on the other hand, has this to say about their deployment:

"Because of the required brief outage every time a virtual machine is moved from one host to another, Microsoft IT found that coordinating the server update processes with virtual machine owners was difficult.  Because one physical host could contain several virtual machines, Microsoft IT had to communicate with each of the virtual machine owners and coordinate host server maintenance with virtual machine maintenance."

Microsoft discusses its limitation of 23 LUNs because each LUN must be available to the Parent Partition" (Host OS). Windows does not deal with drives in terms of numbers, it deals in terms of letters (of which there are 26).  "A" and "B" are traditionally reserved for disk drives and, as such, cannot be assigned to a LUN.  "C" is the local hard drive and cannot be assigned to a LUN.  Therefore, Hyper-V can only address 23 LUNs.  Microsoft says it had to acquire new storage frames that supported more LUNs because they ran out with its 1 LUN: 1 VM strategy.  

Microsoft says it uses 802.1q trunking but not 802.3, and doesn't support NIC teaming.  This means that there is no network redundancy for the VMs in a Hyper-V environment.  This is a huge failing on the network side — and much more important than VLAN tagging they do support.  If a VM network port goes down, all of those VMs are offline until an administrator intervenes (Microsoft plans to fix this in the R2 release of Server 2008).  VMware, on the other hand, not only fully supports 802.3 and NIC teaming today, but with the vNetwork Distributed Switch (VDS) of the upcoming vSphere, will span many ESX hosts allowing quick scale up of networking capacity.  vNetwork also includes Private VLAN support, network VMotion, off-loaded network I/O and the ability for organizations such as Cisco to build VMotion-aware networking applications on top of it.  Cisco's Nexus 1000V virtual switch is expected to be announced on 4/21 in conjunction with vSphere.

When a VMware virtual machine wants to communicate with the network, it must be processed by the Guest OS and then by the Hypervisor.  Once the Hypervisor gets hold of it, the traffic gets pushed out to the network at large.  On Hyper-V, the traffic must be processed first by the Guest OS, then by the Hypervisor, and then by the Parent Partition (the Host OS).  The added layer (host OS) is the one that has traditionally been a bottleneck (which is one reason why ESX performs so much better than Hyper-V).

Here is a summary of the internal virtualization environments at VMware and Microsoft 

 

  Internal Virtualization Environment    VMware Microsoft

Current % of servers that run as virtual instances   97% < 50%

% of Servers that will be virtual instances by end of Q2    100%   50%

# of Production VMs per 8 core/32 GB host     10.4     5.7

High Availability enabled?    Yes No

Migration of live server instances without downtime?    Yes     No

Natively supports NIC teaming (network redundancy)      Yes No

Can add vSwitches without requiring host rebooting  Yes    No

Efficient communication of VMs with the network    Yes    No

LUN limitation    255    23

# of VMs managed per administrator    145 ?

Thanks to Jason Coleman and Steve Jones of INX who explained the networking and storage issues for me (though any mistakes in transcribing their remarks are solely the author's).

 

 

Presentation on Hyper-V vs VMware Sponsored by Microsoft and Dell

Hyper-V vs. Market Leaders http://www.bitpipe.com/detail/RES/1237562441_234.html

IT expert Nelson Ruest has impecable credentials as a well-published author.  Then again, I've co-authored many Osborne/McGraw-Hill technical books, and you wouldn't want me anywhere near your data center.   But even I understand enough to see that this is a pretty weak attempt to show Microsoft Hyper-V as a legitimate contender to VMware.  Take for instance, the only criticism of VMware from the presentation:

“Hyper-V is based on Windows Server 2008 and because of that you can run these machines on pretty well any hardware configuration, any hardware configuration that is designed to support Windows.  So, that means that you have hundreds and hundreds, if not thousands, of configurations that you can run Hyper-V on  whereas because ESX Server, VMware is more…a more limited edition…a more limited product because of that.  You can only run VMware on maybe dozens or fewer number of server configurations than you can run Windows.”
 
Is this really relevant to anything?  How many ESX host options does any organization possibly require?   Furthermore, new VM hosting platforms such as Cisco's UCS will start to have a major impact in data center virtualization, and UCS is optimized for VMware.
 
Check out these links which do a great job of dispelling Microsoft FUD:
 
http://itknowledgeexchange.techtarget.com/server-virtualization/vmware-fires-back-against-microsofts-mythbusters/
http://vinternals.com/2009/04/microsoft-myths-and-realities/
http://itknowledgeexchange.techtarget.com/server-virtualization/microsofts-embarassing-hyper-v-vs-vmware-video/
http://vteardown.com/2009/04/06/how-stable-is-hyper-v/
http://www.vcritical.com/2009/04/would-you-buy-a-hypervisor-from-these-guys/
 

 

 

 

Burton's Report: The Dark Side of Virtualization (4/6/2009)

http://www.burtongroup.com/Research/PublicDocument.aspx?cid=1464

Unfortunately, unless you're a Burton subscriber, you can't read their document.  Nonetheless, I'll go ahead and address the issues it raises.  Burton has a lot of clout in the industry, and although I disagree with some of the specific content of this paper, I recommend subscribing to their service for those who value generally well researched reports.

At a high level, I agree that virtualization has a dark side.  The dark side, though, has little to do with the purported issues presented in the Burton report.  Rather, it’s an implemenation done without the financial analysis, assessment, planning, proof-of-concept testing, piloting, beta deployment, project management and the carefully executed roll-out that should be a prerequesite to any successful enterprise deployment.  This tends to frequently limit the expansion of VMware deployments that were implemented originally as point solutions, and then began expanding beyond the original intention.  The organization consequently misses out on tremendous cost savings and other advantages while having to manage both a physical and virtual environment.

 

I think that The Burton Group knows this – they clearly reach during the document for any “dark side” bits that they can drum up in an attempt to write an interesting paper.  Their conclusioin, though, states that “Virtualization is an overwhelming success story in most IT shops and delivers on promised ROI.”  I agree with everything the conclusion says, including “awareness and taking action to address these new challenges”.  The interesting thing, though, is that they just tip-toed around the only true challenge of successful virtualization (other than lack of treating it as an enterprise deployment as mentioned above), which is the network. 

 

While the VMware virtual switch has been a remarkable way to achieve much faster inter-virtual machine communication on an ESX host, it limits the enforcement of security best practices which in turn impacts the environments that can be virtualized.  The VMware virtual switch is rudimentary; it is a basic layer 2 switch that doesn’t route, does not support QOS and provides no visibility to individual virtual machines.  Virtual machines running on an ESX host communicate with each other without going through the physical network –administrators cannot apply traditional security practices such as access control lists and private VLANs.  Administrators also have no way of easily isolating traffic to an individual virtual machine, and even if they do, a VMotion or DRS instance leaves them in the dark once again.  Managing virtual switches is painful because each one must be administered on an individual basis.  Since the server administrator is doing the configuration rather than the network team, misconfigurations and inconsistent policies often result.

 

Fortunately, the Virtual Distributed Switch (VDS) of vSphere addresses these issues.  And Cisco shops will be able to purchase the Cisco Nexus 1000V which makes managing the vSwitch just part of the overall Cisco network management scheme.

Let’s look at some of Burton’s “dark side” points:

 

Organizational Management

 

Burton claims that the difficulty of managing storage and network along with servers can cause a challenge.  Again, if a virtual infrastructure is put in without planning, then this could be a true statement.  Incorporating planning, however, enables a simpler environment than physical – even with the existing networking deficiencies.  Here’s an article I recently wrote about this topic that should shed some light  http://www.dabcc.com/channel.aspx?id=232.  Again, with proper planning and training, IT staffs inevitably really appreciate all of the advantages of a virtualized data center.  If you ever go to VMware’s VMworld, it’s like a 12,000 person love fest.

 

 Systems Management and Monitoring

While the performance and management tools are certainly evolving, again I’d argue that they are already superior to anything you can do in a physical data center given proper assessment/planning is required.  A good example of this took place with one of my clients a few years ago – when the technology was much less advanced than it was today.  An IT manager for a bay area bank told me that he liked everything that I told him about virtualization, but he worried that his security guy would never allow it.  I asked him, “How many servers do you have again?”  He replied, “Between 250 – 300”.  “But you don’t know exactly?” I asked.  “Not exactly,”  he said.  “If I asked your security guy how many servers you have, would he know?”  He replied, “IF I don’t know, then he certainly won’t know.”  I said, “So how is a virtual environment where you can view every server at any time, know the individual and aggregate resource consumption and know who was on what server at what time than in your existing environment where you don’t even know how many servers you have?”

 

Security

 

Burton makes a big deal about VMware having 22 vulnerabilities.  Every OS has vulnerabilities.  Even Cisco IOS has vulnerabilities.  The big question is whether there is any Exploit code available that goes along with the vulnerabilities.  As far as I know, the answer is no.  I’ve yet to hear of a single case of a VMware bypervisor-based security breach.  VMware even provided the NSA with their source code several years ago, and the NSA couldn’t hack between virtual machines.  The VMware hypervisor is extremely thin (probably over 1,000 times fewer lines of code of Windows Server), and therefore very difficult to exploit.  Burton Group again hedges here by saying that “Virtualization increases risk by adding complexity, but can be used to reduce risk by providing another means of separation.”  As my banking example above shows, when you virtualize, it is intuitive that fewer control points and much more granular monitoring and tracking capabilities in turn reduces security risk.

 

Burton says that, “customers will have to spend a lot on upgrades from the vendors for so-called ‘virtual appliances’.  First of all, virtual appliances are not by any means a mandatory part of a VMware deployment.  They are terrific, low-cost VMs that are generally superior to the physical versions – but are certainly not mandatory.  They also bring up the network security vulnerability, but as I mention above, that is addressed fully in vSphere and can be handled today through other means with proper planning.

 

Compliance and Audit

Clearly regulatory compliance is simplified in a VMware environment.  We were seeing this four years ago with the very heavily regulated pharmaceutical industry.  In that industry, products can be life or death.  When you get a prescription filled, it better be exactly the same as the product you received two years ago.  Regulation trumps all, and the FDA requires rigorous “validation” of all servers.  But since VMware virtual machines are provisioned from software and are thus 100% identical, pharmaceutical industry companies have long found it easier to receive and maintain FDA validation of virtual machines than their physical counterparts.

 

Vendor Licensing and Support

 

Burton first complains about the MS policy prohibiting movement of server OS licenses between hosts more than once every 90 days, and then goes ahead and gives the obvious resolution which is to simply purchase Windows Server Data Center Edition for each ESX host.  What they don’t discuss is just how economically advantageous this can be. 

 

Suppose, for instance, that an organization has 100 copies of Windows Server Enterprise.  We’re going toconsolidate those onto three Dell R910s with four 6-core CPUs each and 256 GB RAM.  Assuming that Windows Server Enterprise and Windows Server Data Center Edition both go for around $2,000 street price (they both list for around $3,700), then in the physical realm this organization has to spend $200,000 on Windows Server license (Windows Server Enterprise is licensed by physical server).  In a virtual environment, however, they simply license Windows Server Data Center Edition for the underlying CPUs of the ESX hosts, and can then run unlimited instances of Windows Server.  So the cost becomes $2,000 X 4 X 3 = $24,000.  The organization therefore saves $186,000 every time they renew their Microsoft Server licensing.  Other large licensing advantages can be obtained with SQL Server.  And with vSphere, there will be no reason to purchase Microsoft Clustering software any longer – thereby still saving more money. 

 

Regarding Oracle licensing in a VMware environment, well – I don’t understand Oracle’s licensing model (but then, I’m not sure anyone else does either).

 

Storage, Networking and Architectural Considerations

 

Again, proper planning will address these issues.  And VSphere, with vStorage and the Virtual Distributed Switch, takes storage and network management to a level beyond anything possible in a physical environment.  SRM already enables storage management for disaster recovery beyond anything possible in a physical DR scenario.

 

Burton says they advocate going with low server consolidation densities on blades so that a server failure doesn’t impact many servers.  This is baffling to me as, 1)  Servers rarely fail and 2) If a server does fail, the VM is off-line for only an average of seconds while it reboots – since it’s software, it reboots very fast.  Of course, with the Fault Tolerance feature of vSphere, there will be no down-time at all so Burton’s argument is totally moot.  We also disagree with their advocacy of blades for most situations as we wrote in an article last year, and our opinion is shared by most virtualization industry experts with whom I have spoken http://www.dabcc.com/article.aspx?id=9114.

 

Burton says that the ROI may not be there for smaller organizations.  We’ve found that organizations with as few as 7 servers can still achieve a nice ROI, even when purchasing a new SAN as part of the deal.  But, the key of course is to conduct the ROI analysis prior to virtualization in order to make sure.

 

Burton says that, “Virtualization sprawl has been a potential worry that has only been kept in check by the added management overhead that an IT shop has incurred to manage multiple virtual platforms.”  This is kind of silly.  Most physical data centers still have server sprawl despite the large expense and effort it takes to purchase, configure, rack, stacked, cable, connect, power, cool, protect, back-up and maintain physical servers.  Virtual machine sprawl can easily be contained either through policy or through automated workflow utilizing VMware’s Lifecycle Manager.

 

Recommendations

Burton says if you have an ROI of 18 months or less, you should virtualize (actually the terminology should be payback period as ROI is not a unit of time measurement, but we’ll let that slide).  I don’t think you can set a hard number in any case.  It depends on what else the organization can do with the money, and what other not quantified benefits result – examples being advantages such as no more server failure, no need for hardware-based maintenance windows, greatly enhanced disaster recovery, and on and on.  The rest of their recommendations pretty much imply that organizations should plan their deployments.  I agree.