Home > Blogs > Virtual Reality > Author Archives: Eric Horschman

Author Archives: Eric Horschman

You Can’t Trust Microsoft’s Azure TCO Calculator

[UPDATE: As of January 2, 2018, the double counting of VMware compute category costs has been corrected. The other errors noted in the calculator remain.]

Last month, Microsoft posted an online “Azure TCO Calculator” that supposedly compares the cost of on-premises infrastructure, based on VMware vSphere, with the cost of running the same workloads in the Azure public cloud. Here at VMware, we pay a lot of attention to the Total Cost of Ownership (TCO) of on-premises vSphere versus public clouds. Our customers and industry analysts have told us that a well-managed private cloud often costs less than basic public IaaS clouds. We’ve recently written about a 451 Research study that found 65% of surveyed IT professionals felt their private clouds cost at most only 10% more than public clouds. So, when Microsoft’s blog posts and white papers claimed an 84% TCO advantage for Azure, we knew their figures needed scrutiny.

It didn’t take much digging to find numerous errors and biased assumptions in the Microsoft TCO calculator that wildly skew their results. In fact, the errors were so glaring, we wonder if anyone at Microsoft bothered to test it before going live.

What were they thinking?

The Microsoft calculator provides enough details in its listing of assumptions and in its expanded results breakdown to make it easy to replicate their TCO methodology in a spreadsheet. We did just that, and errors immediately became apparent. Here are just a few:

  • VMware license fees are paid every year of the selected timeframe! Selecting a 3-year TCO timeframe results in vSphere Enterprise Plus license costs being tripled. C’mon Microsoft, we know vSphere isn’t free, but our customers only need to buy it once, not every year.
  • vSphere hosts are grossly overconfigured for CPU and RAM. Using the scenario from the Microsoft white paper of 500 VMs, each with 2 vCPUs and 4GB, results in a server hardware requirement totaling 1,984 cores and 33TB of RAM. That’s 4x the CPU and 16x the memory that should be required! Perhaps they were using Hyper-V host configuration guidelines. Either way, you do NOT want these guys sizing your data center hardware.
  • For Azure Pay As You Go VMs, the calculator is using CPU-throttled B-series VMs that Microsoft describes as appropriate only for small dev/test-type uses.
  • VMs in Azure are powered on only 40% of the time with the calculator defaults. Perhaps a small dev/test team can turn off their VMs when they leave for the day, but the production enterprise workloads that run on vSphere are powered on 24×7, so hosting them on Azure should require the same continuous uptime.
  • The Microsoft calculator assumes vSphere Enterprise Plus, which includes features like High Availability, Dynamic Resource Scheduling, Fault Tolerance, vRealize Log Insight, and more. Azure VM instances don’t have those features – you would need to pay for extra Azure services like Premium storage and Operations Management Suite to come close.
  • On-premises servers, storage and networking hardware are assessed 20% per year maintenance fees in the Microsoft calculator. Are you paying that much? Didn’t think so.
  • The Microsoft calculator doesn’t apply any residual value for the on-premises servers and software at the end of the selected TCO timeframe. I guess Microsoft expects you to send all your servers straight to the landfill on their third birthday.
  • To cap it off, we found that a new math error was recently introduced into the Microsoft calculator. It now double counts all the costs in the “Compute” category for the VMware on-premises configuration – hardware, software, electricity, and virtualization costs are added twice. Really guys, this is getting absurd.

Let’s fix their calculator and see the effects

With all those errors in the Microsoft calculator, it’s not surprising to see that the 500 VM scenario noted above results in an enormous TCO difference between on-premises vSphere and Azure. Fortunately, we can make some simple corrections to their calculator and re-run the numbers. We’ll use vSphere Standard Edition, since it’s more comparable to Azure VM and storage features, and we’ll only charge for it once. We’ll reduce the vSphere server count to properly provision CPU and memory. A more reasonable 6% annual hardware maintenance fee will be assumed. The Azure PAYGO configuration will be upgraded to more realistic F2 VMs needed for steady workloads. Lastly, we’ll keep the Azure VMs running 24×7, as they would be for production workloads, by setting utilization to 100%.

With those simple corrections, the picture changes dramatically. Now, the on-premises vSphere TCO is lower than all but the 3-year reserved Azure TCO – a result much more in agreement with costs reported by customers.

Azure TCO Calculator Inputs

  • 500 compute VMs with Windows, 2 vCPUs and 4GB each, US East 2
  • VM density: 2 vCPUs/physical core
  • 100TB local disk/SAN storage
  • 3-year timeframe

If you really want a VMware-compatible public cloud with low TCO, look at VMware Cloud

So, now that we’ve shown that Azure doesn’t have a TCO advantage over on-premises vSphere, where should VMware customers look for a public cloud provider that can really save them money and increase agility with true hybrid cloud computing? The answer is the VMware Cloud offerings that provide vSphere-consistent infrastructure without the complexity of migrating workloads to an entirely different platform like Azure. You can choose VMware Cloud on AWS that features vSphere, vSAN and NSX with familiar vCenter management, all running on bare-metal AWS infrastructure with optimized access to AWS services. Or, select from the thousands of VMware Cloud Providers that offer VMware Cloud Verified infrastructure-as-a-service. You’re sure to find a provider that fits your needs and TCO budget when you’re ready take the easiest and most direct path to hybrid cloud.

Discover How Much You Can Save by Choosing VMware for Your Private Cloud with Our Updated TCO Comparison Calculator

A brand-new update to the VMware TCO Comparison Calculator is now available for you to compare the Total Cost of Ownership of VMware and Microsoft virtual infrastructure and private cloud platforms. We’ve taken our familiar calculator and updated it with the latest prices, features and independent lab test results to help you evaluate the complete lifecycle costs – combining operational and capital expenses – of a virtual infrastructure and private cloud solution matching your requirements.

Operational Expenses – latest testing reconfirms the VMware advantage

The VMware TCO Comparison Calculator refresh incorporates the results from new independent testing conducted by Principled Technologies that measured the operational costs of running a private cloud based on the latest product releases from VMware and Microsoft. In the case of VMware, Principled Technologies tested vSphere 6.5 and vRealize Suite 7.0. Their Microsoft testing was on Windows Server 2016 Hyper-V and System Center 2016. By measuring the steps and time needed to complete an array of essential virtual infrastructure and private cloud operations tasks, Principled Technologies provided the raw data the calculator uses to estimate the OpEx costs you can expect to pay. As with similar OpEx studies conducted over the last several years, Principled Technologies found VMware OpEx costs to be substantially lower than Microsoft’s. A few of the contributors to the VMware OpEx advantage are:

  • Patching hosts – Principled Technologies again confirmed that vSphere, with its bare-metal architecture, has much lower patching costs than Hyper-V, which comes bolted to a multi-gigabyte Windows Server OS. [Gartner also just reiterated how much more downtime is required for Hyper-V patching.]
  • Protecting VMs – The new VM encryption feature in vSphere 6.5 is much easier to use than the VM protection features in Hyper-V. A simple storage policy setting in vSphere encrypts all VM data at the hypervisor-level and works with any guest OS. Microsoft’s approach requires separate infrastructures for shielding VMs with newer Windows versions, and completely different methods for Linux and older Windows guests.
  • Operations management and automation – The VMware vRealize Suite makes setting up dashboards for infrastructure monitoring and managing capacity much faster than with Microsoft’s System Center components. Your system administrators can better configure vRealize Operations and vRealize Automation to match their operations management and automation needs. In addition, ongoing monitoring is more efficient.

Capital Expenses – influenced by density, licensing, and management resources

The other half of the TCO equation is capital expenses – what you pay for the servers, networking and storage hardware, as well as the software licenses.

Because most of our customers need to run Windows Server VMs, the TCO Comparison Calculator includes the cost of Windows Server 2016 Datacenter licenses for all hosts in both the VMware and Microsoft configurations. Licensing all hosts this way enables you to run as many Windows VMs as you wish. As a result, you might expect Microsoft to edge out VMware on CapEx because Hyper-V comes bundled with Windows Server. However, that’s not the case. Some reasons why are:

  • Management overhead – As management and automation capabilities get more sophisticated with each release, VMware has better avoided bloated compute and memory resources needed for those services. Management components like vCenter, vRealize Suite, and System Center run in their own VMs. The calculator applies the documented sizing guidelines and best practices of both VMware and Microsoft for the management VMs needed by each vendor. Following those guidelines leads to Microsoft infrastructure that needs almost four times the compute and memory resources for management VMs. And this translates into more hosts, more software licenses, and higher CapEx.
  • VM density – Every time we talk to a VMware customer that comes back to vSphere after trying Hyper-V, they confirm that they can run more VMs on their servers with vSphere. It is primarily due to the powerful memory economizing technologies built into vSphere that operate in any out-of-the-box installation, versus the cumbersome and rarely used dynamic memory feature in Hyper-V. The conservative, but adjustable, VM density advantage factored in by the calculator reduces VMware hardware and software requirements, lowering CapEx.
  • Per processor vs. per core licensing – Microsoft’s switch to core-based pricing for Windows Server 2016 and System Center 2016 has raised their prices on any multi-socket server with more than eight cores per processor. Microsoft licenses for a two-socket server with 18-core processors cost 125% more than they did a year ago – OUCH! VMware users also feel that pain, since most are running Windows VMs, but Microsoft private cloud users pay the new higher prices twice – once for Windows Server and again for System Center – driving their CapEx higher. Fortunately for VMware users, vSphere, vRealize Suite and vCloud Suite have CPU-based licensing, so when Intel and AMD pack more cores into their processors, you enjoy the benefits without paying a “core tax.”

Try the calculator and see how much you can save

The VMware TCO Calculator Methodology white paper explains in detail how the calculator works. Here’s an example of the calculator results comparing the TCO of a 1,000-VM private cloud based on VMware vCloud Suite Enterprise (our top-level private cloud bundle combining vSphere Enterprise Plus with vRealize Suite Enterprise) to one built with Hyper-V 2016 and System Center 2016.

VMware TCO Comparison Calculator example

 

Even with conservative assumptions for VM density, VMware provides healthy TCO savings compared to Microsoft. Use the calculator to see what your TCO savings could be.

Small Virtual Infrastructure or Large Private Cloud, VMware TCO is Lowest – Here’s Proof

We’ve just updated the VMware TCO Comparison Calculator to help customers see the true Total Cost of Ownership differences between VMware and Microsoft. It’s easy to use – just enter the basic parameters for your virtual infrastructure or private cloud environment, such as the number of VMs, type of servers and storage, and the product edition or features you need. The calculator will generate a complete TCO analysis that includes all the necessary elements of capital and operational expenses.

We created the TCO Comparison Calculator after hearing from existing and prospective VMware customers who were being told that alternative solutions based on Hyper-V would be much less expensive, or even “free”. The calculator totals cost elements that our competition leaves out of their oversimplified comparisons, such as: the system administrator labor costs to operate the environment (the largest component of TCO and one that independent testing shows to be much lower for VMware); effects of VM density (where VMware has an advantage according to analysts like Gartner); 24×7 phone support; and the need for third-party software to fill feature gaps.

When all those cost elements are combined, the VMware TCO Comparison Calculator shows that VMware solutions, ranging from a small business virtual infrastructure built with vSphere Essentials to a full-featured large enterprise private cloud based on vCloud Suite Enterprise, have the lowest TCO – often by substantial margins.

When we updated the calculator, we saw that the VMware TCO advantage increased for some important reasons.

  • Our latest vCloud Suite editions now bundle the platform power of vSphere Enterprise Plus together with the management operations and automation efficiencies of vRealize Suite at very attractive pricing.
  • Log analytics powered by vRealize Log Insight is now included with any configuration that includes vCenter Server or vRealize Suite, making VMware solutions more economical than alternatives that must make up for that capability by adding third-party software.
  • Microsoft’s adoption of core-based pricing in their upcoming release of Windows Server 2016 and System Center 2016 makes their solutions more expensive on mainstream servers that have higher core counts. The calculator assumes both VMware and Microsoft hosts are licensed for Windows Server Datacenter edition, so its core-based pricing penalizes customers of both vendors, but the lower VM density of Hyper-V means more Windows Server licenses are needed for a Microsoft platform. Also, System Center is needed to manage Hyper-V and its higher costs with core-based pricing fall entirely on the Microsoft side of the TCO comparison.

Another important enhancement we’ve made to the calculator is local currency support. Users can select USD, AUD, EUR, GBP, or JPY and the calculator will apply VMware and Microsoft list prices from those geographies.

This example from the VMware TCO Comparison Calculator shows that the 3-year TCO for a 500-VM environment built with vSphere with Operations Management Enterprise Plus will be 33% less than a comparable solution based on Microsoft Windows Server Hyper-V and System Center.

vmwareCostGraph

Our customers in the trenches running enterprise virtual infrastructures often tell us they know VMware offers the best and most cost effective solution, but they need help making the case for selecting VMware with purchasing managers or CFOs that have heard from other vendors claiming to be less expensive. If you find yourself in a similar position, use the VMware TCO Comparison Calculator to arm yourself with solid proof that VMware provides the lowest total costs.

VMware TCO Comparison Calculator Now Shows that Stepping Up to Private Cloud Costs Less with vCloud Suite

If you’ve had a chance to use the VMware TCO Comparison Calculator, you know that it factors in all the elements of a proper Total Cost of Ownership analysis to compare the true cost of building a virtual infrastructure on our vSphere and vSphere with Operations Management products to the cost of building a similar infrastructure on Microsoft’s “Cloud OS” – their name for Windows Server Hyper-V and System Center. [VMware has an even more detailed ROI/TCO Calculator to show the financial savings of virtualization and private cloud vs. physical infrastructure.]

image

The results are eye-opening for many users who have seen the comparisons from our competitors that consider only the Windows operating system and virtualization software license costs. Including all the TCO elements shown above makes it very clear that the cost of virtualization software is just a small part of the overall TCO for a virtualized infrastructure.

We’ve just updated the TCO Comparison Calculator with two important new features:

  • You can select VMware Virtual SAN (VSAN) as the storage technology for VMware. VSAN capital costs are significantly less than other storage options like Fibre Channel, iSCSI or NAS.
  • You can see the TCO of upgrading your vSphere infrastructure to a full-featured vCloud Suite private cloud. This option compares the cost of upgrading to vCloud Suite with the cost of migrating to a Microsoft Windows Server Hyper-V and System Center private cloud.

There are three key cost elements that work strongly in VMware’s favor that show up in the calculator results:

  • Better VM density – Being able to run more VMs per CPU has always been a vSphere strength due do its outstanding memory management and DRS load balancing technology. Our customers running both vSphere and Hyper-V report higher densities on vSphere and analysts like Gartner agree, saying, “VMware’s customers have higher VM densities.” Even a conservative VM density advantage translates directly to significantly lower CapEx and OpEx costs.
  • Richer feature set – vSphere with Operations Management and vCloud Suite provide more of the management, data protection and availability, networking and disaster recovery features that customers need. Without those features, Microsoft customers must purchase, integrate and administer multiple third-party products to fill the gaps, driving up costs.
  • Much lower operational costs – Our customers that have tried competitors’ products tell us that running a vSphere and vCloud Suite infrastructure is much easier and more efficient. Third party studies have quantified VMware OpEx cost as much as 80-90% lower than Microsoft and recent studies with the latest product versions show a similar advantage. The OpEx savings from VMware’s greater administrative efficiency are built into the TCO Comparison Calculator.

A quick example from the VMware TCO Comparison Calculator shows just how much of an impact those VMware cost savings have. This example shows the two-year TCO for an infrastructure of 1,000 VMs on vSphere with Operations Management Enterprise Plus (our highest edition) vs. Microsoft Windows Server Hyper-V and System Center.

image

image

You can see that VMware delivers 30% lower TCO from its lower OpEx costs and features that preclude the need for third-party add-ons.

Here’s an example showing that the two-year TCO for upgrading a 1000-VM vSphere Enterprise environment to our full-featured vCloud Suite Enterprise platform comes in 36% less than if that same infrastructure were migrated to Microsoft’s “Cloud OS”.

image

Whether you’re new to virtualization and considering a greenfield server consolidation project or a long-time vSphere user weighing your options for a private cloud upgrade, give the VMware TCO Comparison Calculator a try – you’ll see that you can get the best for less.

Study Shows Higher Costs and Complexity When Managing vSphere Using Microsoft System Center Virtual Machine Manager

The idea of introducing multiple hypervisors into your data center and managing them seamlessly from a single tool might sound appealing, but in reality, products claiming that ability today can’t deliver on that promise.  You introduced virtual infrastructure to simplify operational tasks for your IT staff, so why would you want to handicap them with a management approach that adds costs and complexity?  A study recently completed by the Edison Group and commissioned by VMware shows that is exactly what you will be doing if you introduce Microsoft System Center 2012 Virtual Machine Manager (SCVMM) with the hopes of using it to manage VMware vSphere hosts.

Microsoft touts SCVMM as a heterogeneous management tool with the ability to manage VMware vSphere and Citrix XenServer hosts in addition to those running Hyper-V.  IT managers might find Microsoft’s claims that they can, “easily and efficiently manage… applications and services across multiple hypervisors,” enticing. The suggestion by Microsoft is clear: don’t worry about complicating the jobs of your system administrators by introducing Hyper-V into a VMware environment because SCVMM provides a do-everything single-pane-of-glass control panel. Are their claims true?  Can Microsoft SCVMM really let you operate a multi-hypervisor data center without the cost penalties that come with staffing, training for, and operating across the isolated islands of management that would otherwise exist?

To find the truth behind Microsoft’s promises, we asked Edison Group to test VMware vSphere in their labs using both vCenter and the vSphere Client and Microsoft SCVMM 2012 to complete a set of 11 typical management tasks.  Edison’s analysts used their Comparative Management Cost Study methodology to measure the labor costs and administrative complexity of each task.  The tasks Edison Group studied were those that any vSphere administrator performs on a regular basis, such as provisioning new vSphere hosts, deploying VMs, monitoring system health and performance, configuring virtual networks, etc.

Higher costs and complexity when managing vSphere with SCVMM 2012

The results were clear and conclusive – managing VMware vSphere is much more efficient using vCenter than when attempting to manage it with Microsoft SCVMM 2012.  To complete the 11 typical management tasks Edison Group tested took 36% less time and required 41% fewer steps using vCenter and the vSphere client compared to SCVMM 2012.

Figure 1 Managing vSphere using vCenter takes 36% less administrator time than with SCVMM 2012

Figure 2 vCenter management of vSphere requires 41% fewer steps than SCVMM 2012

Jack of some trades, master of none

It’s not hard to understand why vCenter and the vSphere Client make life so much easier for vSphere administrators. As my colleague Randy Curry wrote, Microsoft SCVMM 2012 just doesn’t do a very good job of enabling vSphere management.  SCVMM’s incomplete or missing support for even basic tasks forces administrators to constantly jump over to the vSphere Client to get any real work done.  Microsoft was apparently more interested in being able to “check the box” for multi-hypervisor management when they built SCVMM 2012 than they were in providing a truly usable vSphere management tool. As Edison Group said in their report (available here or here):

Managing hypervisors using tools that are not specifically optimized to control all aspects of their operations risks impairing reliability, elegance, and ease of management, with potential adverse impact on the bottom line. Creating a truly successful solution requires deep integration and expertise in development.

Adding different hypervisors? Proceed with caution.

Multi-hypervisor IT shops are a trend that may be growing, but don’t expect a simple single-pane-of-glass management experience if you bring in a different hypervisor.  The testing by Edison Group clearly shows that management costs and complexity will be substantially higher if you attempt to use a partially implemented heterogeneous management tool like Microsoft SCVMM 2012 to manage a vSphere infrastructure. We at VMware realize that operating a 100% vSphere environment is not always possible and we’ve recently introduced our own multi-hypervisor management features with vCenter Multi-Hypervisor Manager and vCloud Automation Center to accommodate those cases. Rather than positioning those solutions as enablers of permanent multi-hypervisor environments, we’re offering them to help our customers manage heterogeneous pools of infrastructure until they can migrate their workloads to a VMware platform where they can benefit from our exclusive software-defined datacenter capabilities.

If you’re weighing possible benefits of introducing a second hypervisor, you may want to take the advice of Gartner’s Chris Wolf and stick to a single hypervisor unless you want maintain and pay for separate islands of management:

Multi-hypervisor… has serious tradeoffs if it’s the end goal for the production server workloads in your data center. Additional hypervisors for one-off siloed initiatives is often practical, but becoming less standardized in your data centers is anything but efficient.

Chris Wolf repeated that message at a session on heterogeneous virtualization we attended at the recent Gartner Data Center Conference. In fact, he stated there that no Gartner clients have succeeded in adopting a single-pane-of-glass multi-hypervisor approach. That’s refreshingly frank advice that should be heeded by anyone lured by Microsoft’s promises of multiple hypervisor nirvana.

Sorry Microsoft; Not Only Does vSphere Cost Less to Buy, It Also Costs Less to Operate.

Microsoft’s wildly exaggerated claims of providing a less expensive virtualization platform than VMware vSphere have been hard to miss if you’ve spent any time on the web lately. We’ve previously pointed out the flaws in their math and our public Cost Per Application Calculator clearly shows how deploying a virtual infrastructure built with vSphere will cost about the same as one built using Microsoft Hyper-V and System Center (or even much less when vSphere’s proven VM density advantage over Hyper-V is factored in.) Now we’re pleased to share recent independent test results that show how vSphere also delivers dramatically lower operational costs compared to Microsoft.

The acquisition capital expense (CapEx) advantage for vSphere shown by our Cost Per Application Calculator is just part of the Total Cost of Ownership (TCO) that diligent customers will want to evaluate when choosing a virtualization and cloud platform. The other key TCO element to consider is Operational Expenses (OpEx) representing the ongoing costs of administering your virtual infrastructure. To help customers assess the OpEx differences between vSphere and Hyper-V, we enlisted Principled Technologies to run both platforms in their labs and measure the system administrator labor time needed for typical recurring tasks.

Five Typical Datacenter Tasks Tested

Principled Technologies selected five common tasks that any administrator of a virtualized datacenter must regularly perform and they measured the administrator labor time taken to complete each one using both the VMware and Microsoft platforms. The tasks tested were:

  1. Shifting virtual machines off a host to accommodate physical maintenance
  2. Adding storage volumes and redistributing virtual disk files across the new storage
  3. Isolating storage-intensive “noisy neighbor” virtual machines
  4. Provisioning new hosts
  5. Performing a non-disruptive disaster recovery failover test

Care was taken to conduct the scenario tasks as realistically as possible using the full capabilities of the latest released versions of the VMware and Microsoft products available at the time of the testing. vCenter Site Recovery Manager was included in the VMware configuration (and the full list SRM license costs were included in the VMware total cost figures.) vSphere 5 delivered a convincing across-the-board win over Microsoft for each task tested: tasks took 78% to 97% less time to complete using vSphere. The time savings provided by the VMware platform arise from the more advanced capabilities built into vSphere and the more efficient and optimized implementation of those features we’ve perfected over our years of focusing purely on delivering the best virtual infrastructure and cloud platform.

PT_TCO_Fig2

It’s important to note that the OpEx dollar savings shown in the chart above derive from only five representative sysadmin tasks. There are many other regular activities performed by administrators of virtualized datacenters, most of which will show similar efficiency advantages for vSphere over less mature and capable alternatives, so customers should expect even greater total OpEx savings from using vSphere.

Scenarios Delivering Big OpEx Wins for vSphere

Contributors to the biggest operational advantages for vSphere were:

  • Storage Distributed Resource Scheduler – When more storage is needed, the vSphere administrator can add volumes and let Storage DRS redistribute virtual disk files to the new volumes automatically with no VM downtime. The Hyper-V administrator must manually redistribute VM storage and make arrangements for VM downtime during the operation.
  • Concurrent vMotion – The vSphere administrator can complete physical host maintenance much sooner because vMotion maintenance mode evacuations of VMs can proceed concurrently and at a faster rate (see comparative live migration testing results here.) Hyper-V hosts can only handle one live migration at a time, so administrators are tied up with much longer maintenance windows.
  • Storage I/O Control – vSphere makes it easy to cap the storage IOPS consumed by each VM to prevent resource hogging by “noisy neighbors.” Hyper-V has no such feature, so administrators can only respond by dedicating storage volumes for misbehaving VMs – a tedious task requiring VM downtime.
  • vCenter Site Recovery Manager – SRM fully automates replication of mission-critical VMs to a remote site and failover in case of disaster. Real disasters may be rare, but full-scale DR tests should be regular events. That’s where the automated and non-disruptive DR failover test features in SRM deliver big operational savings. Setting up DR failovers with Hyper-V requires maintenance-intensive scripting to orchestrate VM replications and restarts and to modify VMs for network isolation.

After making their labor time measurements, Principled Technologies then estimated how many times each task would be repeated over the course of a two-year period in a datacenter operating 1,000 virtual machines. Multiplying the cumulative time taken performing each task by the U.S. national average system administrator compensation rate gave a dollar figure for the OpEx savings. As shown below, the final result was an impressive 91% reduction in operational expenses when using vSphere compared to Microsoft Hyper-V and System Center.

PT_TCO_Fig1

Operational Expenses Dominate Total Costs

The impact of the OpEx savings delivered by vSphere are even more significant when you consider that IT operational expenses are typically much larger than capital expenses. In fact, Gartner survey data shows cross-industry IT OpEx spending is almost three times CapEx spending – even more reason to choose a virtualization platform that will save you money with better operational efficiency long after the initial purchase.

The OpEx savings delivered by vSphere were enough to tip the two-year TCO advantage in favor of VMware in the 1,000-VM datacenter that was the baseline for Principled Technologies’ tests. When you can have the clearly superior features provided by vSphere 5 Enterprise Plus Edition together with vCenter Site Recovery Manager at a lower total cost than Microsoft’s best alternative, it’s easy to make the decision to go with VMware.

If you’re running your datacenter on vSphere now, an OpEx win by vSphere probably isn’t surprising. vSphere users benefit from over a decade of optimizations we’ve built into our platform that derive from experience in thousands of production datacenters. What did surprise us was just how large an operational advantage vSphere has over the Microsoft platform. The results are a clear example of why you need to look at more than just a feature checklist and initial price tag when choosing the virtualization platform for your most critical workloads.

Take a look for yourself at the full test report by Principled Technologies here (registration required.)

The vMotion Speed Advantage: It’s Real, and It’s Spectacular

The raw speed of VMware vMotion live migrations for moving single and multiple virtual machines has been a huge timesaver for our customers. Speedy concurrent vMotions let system administrators quickly evacuate the VMs off hosts before a planned server maintenance session so they can get home on time, rather than spending extra hours at work to swap a server power supply or do a firmware upgrade. vMotion also has a minimal impact on VM performance, so mission-critical VMs can be moved during production hours without generating user complaints.

vSphere 4.1 introduced support for up to eight concurrent vMotion migrations per host and 10Gb vMotion networks. vSphere 5 boosted the raw speed of each vMotion over 1Gb and 10Gb networks and added the ability to utilize multiple network links. Now, thanks to independent lab tests conducted by Principled Technologies, we can show just how fast vMotion operates.

Principled Technologies compared vSphere 5 vMotion performance to live migrations conducted with Microsoft Windows Server 2008 Hyper-V R2 SP1. The comparison to Hyper-V was of special interest to us because we’ve seen claims from those in the Hyper-V camp that allowing multiple concurrent vMotions couldn’t possibly be faster than Hyper-V’s one-at-a-time live migrations because Hyper-V can saturate a network link with its live migration memory copy traffic. The Hyper-V advocates discounted the advantages of concurrent vMotions, arguing that they would be bottlenecked by the network.

The results of the Principled technology tests show those arguments are flat wrong. Using identical hosts connected over a single 10Gb network link, a host evacuation simulation where 10 VMs running SQL Server were migrated between hosts showed that vMotion completed the task 5.4 times faster than Hyper-V. vSphere reached its peak capacity of eight of simultaneous vMotions during the test, proving the benefits of concurrent live migrations.

PT_Live_Migration_Fig3

Principled Technologies then looked at live migration of a typical large Tier-1 application. Their test migrated a single 16GB, 4 CPU VM running a heavily loaded SQL Server database. Customers might shy away from such a demanding live migration during production hours. However, the results from Principled Technologies show that vSphere 5 users can migrate Tier-1 apps with confidence, but Hyper-V system administrators might want to plan on some late-night overtime. vMotion migrated the large VM 3.4 times faster than Hyper-V.

 

PT_Live_Migration_Fig4

Maybe more importantly, vMotion’s shorter migration window minimized the impact of live migration on application performance. Compared to a Hyper-V live migration that disrupted the SQL application with long periods of zero throughput, the vMotion VM was able to process 63% more transactions during the migration window.

PT_Live_Migration_Fig5 PT_Live_Migration_Fig6

 

Please take a look at the full test report from Principled Technologies to get the complete test details as well as their findings that vMotion showed perfect reliability, but Hyper-V live migrations resulted in occasional, but reproducible, VM blue screens. With differences like these, it’s no surprise that Microsoft is now talking about augmenting a future release of Hyper-V with concurrent live migration capabilities. As with so many other features that are vSphere exclusives, VMware customers will be enjoying their benefits, while Hyper-V users, waiting for Microsoft to deliver on promised enhancements sometime in the future, will be enviously glancing at them.

Newsflash! Microsoft Charges Less than VMware for Much Inferior Product; Promises to Release Less Inferior Product Sometime in Future

Every year, about the time VMworld starts, Microsoft can be counted to release something and it’s not one of their software products. Instead, they’ve made it a tradition to issue a plea to customers to overlook the VMware products that clearly deliver the best virtualization and cloud solutions and choose Microsoft offerings despite their shortcomings. At VMworld 2008, it was Microsoft’s poker chip fiasco. Last year, Microsoft took out full-page newspaper ads asking customers to wait for their products to mature instead of deploying VMware clouds. This year, right on cue, Microsoft published inaccurate, unsubstantiated comparisons to VMware in the form of a whitepaper purporting to evaluate private cloud economics and a big-budget spoof video that was reviewed as “backfiring on itself.”

As regular Microsoft-watchers might expect, the claims made in their private cloud whitepaper of cost and capability advantages are exaggerated and easy to pick apart. Here are a few of the most blatant inaccuracies:

Microsoft inaccuracy #1: Compare (double) discounted Microsoft prices to VMware list prices and leave out the management server costs

For starters, Microsoft is comparing double-discounted Microsoft prices to full list pricing for VMware’s products. Not only is Microsoft basing their comparison on “Open License” prices that already have volume discounts applied, but they go on to apply “ECI” discounts as well, even though the configurations in their examples are far smaller than the 50-processor minimum required by Microsoft’s own ECI rules.  Microsoft’s next trick is omitting the cost of the multiple System Center management servers needed to deliver the capabilities they claim, even though they include the cost of a VMware vCenter Management Server in their VMware totals. Adding the System Center management server license and support costs and the Windows Server licenses needed to run them and using list prices is enough to more than double the cost of Microsoft’s example private cloud configuration from $18,480 to $41,141, as shown below (we’re still applying their ECI discounts, even though the configuration is too small to qualify.)

clip_image002[4]

License

Quantity

List Price w/ 3 years of SA

Total Cost

ECI Datacenter*

4

$5,473

$21,892

Windows Server 2008 Standard**

7

$1,271

$8,897

System Center Configuration Manager w/ SQL

1

$2,312

$2,312

System Center Operations Manager

1

$1,013< /p>

$1,013

System Center Operations Manager w/ SQL

2

$2,312

$4,624

System Center Service Manager w/ SQL

1

$2,403

$2,403

 

 

Total Microsoft Costs

$41,141

*Microsoft ECI Datacenter bundle purchase rules require minimum purchase of 50 processor licenses making Microsoft’s two-node cost example invalid.

**Seven Windows Server 2008 licenses are required for the System Center management servers.

Microsoft inaccuracy #2: Compare unreleased future products to shipping VMware products

Microsoft always likes to sell their roadmap of future promises rather than the products customers are stuck with today. Their private cloud whitepaper is no exception as it describes cloud features that won’t be available to users until Microsoft ships the next versions of System Center at some unspecified future date. Every component of our Cloud Infrastructure Suite – vSphere 5, vCenter Operations, vCenter Site Recovery Manager 5, vShield 5 and vCloud Director 1.5 – has been recently updated and is shipping today. VMware customers are building and running private clouds today and telling their success stories while, on the other hand, Microsoft doesn’t include any customer stories in their private cloud whitepaper. Once again, Microsoft is asking users to wait for yet one more release to get them the features they need.

Microsoft inaccuracy #3: Equate ECI to VMware Cloud Infrastructure Suite

The most glaring exaggeration in Microsoft’s private cloud whitepaper is their assumption that the products included in their ECI Datacenter bundle (Windows Server 2008, Hyper-V, System Center, Forefront) provide private cloud capabilities equivalent to the VMware Cloud Infrastructure Suite. In fact, Microsoft’s cloud solution is far less complete and capable as the comparison table below illustrates.

 

clip_image004[4]

Cloud Infrastructure Suite

clip_image006[4]

ECI Datacenter

Cloud Resource

Consumption

VMware vCloud Director and vCenter Chargeback

·  Secure multi-tenancy through Virtual Datacenters

·  Automated, fast provisioning of cloud resources

·  Extensible vCloud API provides programmatic interface for cloud deployments

·  vCloud Connector provides a single pane of glass to monitor cross cloud application mobility

·  Comprehensive monitoring of cloud resource consumption, full cost transparency and accountability for self-service resource requests

SCVMM SSP 2.0, SCVMM 2012 (future release)

·         No multi-tenancy

·         Requires manual resource provisioning and deployment

·         No cloud API’s

< span style="font-family: ; color: ; mso-fareast-font-family: symbol; mso-bidi-font-family: symbol; mso-themecolor: background1">·         3rd party tools required for monitoring consumption

·         SCVMM 2012 only available in Q4 2011 or later

Operations and Management

VMware vCenter Operations Manager

·         Integrated real-time performance, capacity and configuration management

·         Self-learning, patented analytics and dynamic thresholds adapt to the environment

·         Powerful visualization and quick drill down across datacenter, cluster and host levels to quickly isolate problem root causes

·         Tight integration with vSphere, able to take advantage of advanced features

Requires multiple System Center Components

·         No self-learning analytics

·         Uses static, reactive thresholds that do not adapt to the environment generating potentially false alerts

·         Cannot take advantage of advanced features of vSphere such as DRS, Storage DRS, Network and Storage I/O Controls

Disaster Recovery

VMware Site Recovery Manager

·         Automated failover/failback and migration for reliable recovery

·         Simplified management of recovery and migration plans, replaces manual runbooks with centralized recovery plans

·         Cost-efficient replication of applications to a failover site with built-in vSphere Replication and broad support for storage-based replication

Data Protection Manager

·         Only limited site-to-site clustering capabilities

·         Requires manual runbook scripting

·         No host-based replication

 

Security

and Compliance

VMware vShield

·         Virtualization-aware security and compliance for cloud infrastructures

·         Enables offload of file activity to security VM for AV scan on each host, avoiding performance bottlenecks

·         Integrated security capabilities such as network security gateway services and Web load balancing for performance and availability

Forefront

·         Only an anti-virus product, not a full virtualization security suite that can create VM-level firewalls

·         Not virtualization-aware, requires in-guest agents, prone to “AV Storms” during updates and scans

·         No integrated load balancing and firewall features

Virtualization Platform

VMware vSphere

·         Consolidation of shared physical infrastructure into  tiered logical resource pools

·         Advanced features such as DRS, Storage DRS, Network I/O controls to maintain SLAs for cloud services

·         Leading performance and scalability

Hyper-V

·         No ability to create logical resource pools

·         Cannot guarantee SLAs;  lacks advanced features present in vSphere

·         Inferior performance and scalability when under resource pressure

 

 

Microsoft inaccuracy #4: Use unrealistic scaling assumptions for Hyper‑V

Another exaggeration Microsoft makes in their whitepaper is to claim the costs of a private cloud based on Hyper-V somehow remain flat, even as the VM density increases substantially. Even with the advent of “Dynamic Memory” in Hyper-V, Microsoft’s ability to scale up the VMs per host lags way behind vSphere. As shown in independent tests of vSphere 5 and Hyper-V R2 SP1 conducted by Principled Technologies, vSphere outscales Hyper-V enabling higher VM densities. The shortcomings of Dynamic Memory were clearly apparent when the testing showed total SQL Server throughput of a Hyper-V host declining as VMs were added while a vSphere host continued to increase its throughput as VM density
grew. That means that rather than scaling from 252 to 1260 VMs with no additional hardware as the Microsoft paper implies, Hyper-V users will be steadily adding servers and all their accompanying costs as their cloud grows. A simple example that assumes six VMs per processor (realistic for Hyper-V) and Microsoft list prices shows that the flat costs claimed in Microsoft’s paper are absurd and the real costs of a Microsoft private cloud will steadily grow as VMs are added.

clip_image008[4]

Microsoft inaccuracy #5: Make unsubstantiated claims of “best in class” performance

The Microsoft whitepaper declares, “Hyper-V provides best-in-class performance and scalability for Microsoft applications like SharePoint, SQL, and Exchange,” but no head-to-head benchmark test results are provided to support the claim. The truth is VMware vSphere 5 provides superior scalability and performance with mission-critical, demanding applications, including Microsoft’s, and we have the third-party benchmark test results to prove it.

Other distortions that need correcting

Those inaccurate claims found in the whitepaper aren’t the only ones coming from Microsoft lately. We’ve seen them making these other statements below in various places. Our customers know they’re absurd, but we need to set the record straight.

·         VMware sells virtualization, not cloud – False. Customers need to make the journey to cloud computing at their own pace and server virtualization is the most common first step, so of course VMware offers virtualization for small and large enterprises. VMware also has the products and partnerships today to deliver real cloud solutions. Microsoft can only ask customers to wait while it works on promised cloud support in future System Center releases, so we’re not exactly sure what it is they’re selling now.

·         VMware can’t see deep inside apps – Wrong. Microsoft is hoping customers accept their notion that only Windows apps running on Hyper-V can be fully monitored and managed. The truth is apps running on vSphere and VMware clouds are just as manageable, whether it is with VMware tools like vCenter Operations, vCenter Configuration Manager or vCenter Application Discovery Manager, or with Microsoft tools or third-party products. Microsoft’s suggestion that their hypervisor provides deeper insight into Windows apps is nonsense.

·         VMware doesn’t have its own public cloud – Not true. VMware hosts, manages and supports Cloud Foundry,  the world’s first open Platform as a Service (PaaS) offering that doesn’t lock users into a single framework, single set of application services or a single cloud as Microsoft Azure does. In the more mature area of IaaS clouds, VMware’s approach is to cultivate a strong ecosystem of dozens of vCloud hosting provider partners that can match customer needs far better than a proprietary cloud like Microsoft Azure that locks in your workloads with no way out.

Microsoft might serve its customers better by putting more energy into shipping real products and less into deceptive whitepapers and propaganda videos, but that’s probably wishful thinking. I expect they have their brain trust locked in a room working on defensive misinformation in advance of VMworld 2012. Until then, we’ll keep building the products and solutions VMware customers need and expect from us.

It’s no surprise that vSphere 5 holds up under pressure, but what about Hyper-V?

Before we head out to VMworld, I want to share with you some fascinating test results just published by Principled Technologies that compare vSphere 5 performance and scalability to Microsoft Hyper-V Server R2 SP1.

When Microsoft released Windows Server 2008 R2 SP1, they added a feature called “Dynamic Memory” that they claimed brought them into parity with vSphere in VM density – the number of VMs doing useful work a host can support. We’d tested previous releases of Hyper-V without Dynamic Memory and found that, without the ability to overcommit memory, Hyper-V would hit a VM density brick wall far before vSphere reached the point of diminishing returns. Would Dynamic Memory yield a breakthrough improvement for Hyper-V? We had our doubts because of Dynamic Memory’s reliance on in-guest ballooning as its only way to reclaim memory from guests to support memory overcommitment. We knew from our history with ESX, ESXi and vSphere that getting good, predictable performance when VM density gets high and host memory is overcommitted requires more than just ballooning. We’ve built an array of technologies into vSphere that have been optimized for over a decade to make it a platform our customers feel comfortable with when pushing resources to the limit.

To get an answer, VMware commissioned Principled Technologies to do a side-by-side comparison of vSphere 5 and Hyper-V R2 SP1 throughput when running a SQL Server workload under high VM densities. They used the well-respected DVD Store Version 2 benchmark to measure total throughput delivered by a host running 24 VMs, and then 30 VMs. With 24 4GB VMs, the 96GB host was just reaching full memory commit, and 30 VMs pushed it to 25% memory overcommit – familiar territory for vSphere users.

The results won’t surprise vSphere customers – here’s how the VM-by-VM score looked:

clip_image002

When Principled Technologies added up the throughput of each VM, vSphere 5 delivered 19% more aggregate throughput (orders per minute as measured by DVD Store) on the host running 30 VMs.

clip_image004

Findings that really pleased our vSphere engineers became evident when Principled Technologies dug into the benchmark results a little deeper. One of the key behaviors we seek with vSphere is fairness across the VMs. Assuming equal resource shares and limits, we want each VM to perform as well as its neighbors. Too much variability would be unfair to your users who might get their workloads stuck on an underperforming VM. vSphere 5 came out ahead in fairness as shown in the figure below with a tighter standard deviation in throughput across the 30 VMs as the smaller height of the vSphere box shows in the chart below.

clip_image006

Another striking validation of vSphere 5’s scalability advantage over Hyper-V R2 SP1 was shown when Principled Technologies compared aggregate DVD Store throughput for the 24 VM and 30 VM cases. For Hyper-V, its throughput dropped by 3% when six VMs were added. Evidently, Hyper-V with Dynamic Memory doesn’t hold up so well when you make your VMs do some real work once the host memory becomes overcommitted. In contrast, vSphere 5 throughput increased by 11% as those six additional VMs were added. vSphere 5 is clearly handling the 25% memory overcommit condition with ease.

clip_image008

So, thanks to Principled Technologies, we have the answer to our question: vSphere 5 holds up better under workload and memory pressure to let our users reliably achieve higher VM densities and that means better scalability and lower costs. You can access the full report from Principled Technologies titled, “Virtualization Performance: VMware vSphere 5 vs. Microsoft Hyper-V R2 SP1” on their Web site here, or we’ve also posted a copy on our site here.

Hypervisor Memory Management Done Right

Memory management in VMware vSphere

Sophisticated and effective memory management has always been a key strength of the ESX (and now, ESXi) hypervisor that powers VMware vSphere.  Back in 2001, when ESX first came out, 2GB was a lot of RAM in an x86 server, so it was essential for a hypervisor to economize on memory to deliver the server consolidation benefits customers were so eager to realize.  Back then, the big attraction of server virtualization was running several lightweight utility and test/dev servers in VMs on a single host and memory was usually the resource that limited consolidation ratios.  Today, x86 machines are pushing 1TB of RAM, but because our customers have now virtualized their most memory-hungry production database, messaging and application servers, memory resources are just as critical as ever.

ESX and ESXi have been able to keep up with these huge increases in memory demands because our hypervisor has always employed a multi-layered approach to memory management that delivers the most efficient memory management, best VM densities and lowest costs to our users.  The methods employed are detailed in an excellent VMware technical paper and I’ve summarize them below.

Let users provide critical VMs with guaranteed memory.  The memory “Shares” and “Reservation” settings let you prioritize memory allocated to each VM and reserve enough host RAM for the active working memory of each guest OS.

Efficiently reclaim memory from VMs when needed.  Simply carving off a VM’s full memory allocation from the host’s physical RAM is tremendously wasteful since VMs rarely use all their virtual RAM.  Instead, ESXi borrows or “reclaims” memory from less active VMs when it is needed by other VMs.  We use four techniques for memory reclamation:

  1. Transparent Page Sharing.  Think of it as de-duplication for your memory.  During periods of idle CPU activity, ESXi scans memory pages loaded by each VM to find matching pages that can be shared. The memory savings can be substantial, especially when the same OS or apps are loaded in multiple guests, as is the case with VDI.  Transparent Page Sharing has a negligible effect on performance (sometimes it evens improves guest performance) and users can tune ESXi parameters speed up scanning if desired.  Also, despite claims by our competitors, Transparent Page Sharing will in fact work with large memory pages in guests by breaking those pages into smaller sizes to enable page sharing when the host is under memory pressure.
  2. Guest Ballooning.  This is where ESXi achieves most of its memory reclamation.  When the ESXi hypervisor needs to provide more memory for VMs that are just powering on or getting busy, it asks the guest operating systems in other VMs to provide memory to a balloon process that runs in the guest as part of the VMware Tools.  ESXi can then loan that “ballooned” memory to the busy VMs.  The beauty of ballooning is that it’s the guest OS, not ESXi, that decides which processes or cache pages to swap out to free up memory for the balloon.  The guest, whether it’s Windows or Linux, is in a much better position than the ESXi hypervisor to decide which memory regions it can give up without impacting performance of key processes running in the VM.
  3. Hypervisor Swapping.  Any hypervisor that permits memory oversubscription must have a method to cope with periods of extreme pressure on memory resources.  Ballooning is the preferred way to reclaim memory from guests, but in the time it takes for guests to perform the in-guest swapping involved, other guests short on memory would experience freezes, so ESXi employs hypervisor swapping as a fast-acting method of last resort.  With this technique, ESXi swaps its memory pages containing mapped regions of VM memory to disk to free host memory.  Reaching the point where Hypervisor swapping is necessary will impact performance, but vSphere supports swapping to increasingly common solid state disks, which testing shows can cut the performance impact of swapping by a factor of five.
  4. Memory Compression.  To reduce the impact of hypervisor swapping, vSphere 4.1 introduced memory compression.  The idea is to delay the need to swap hypervisor pages by compressing the memory pages managed by ESXi – if two pages can be compressed to use only one page of physical RAM, that’s one less page that needs to be swapped.  Because the compression/decompression process is so much faster than disk access, performance is preserved.

What about “Dynamic Memory”?

The days of ESX being the only hypervisor that could support memory oversubscription are over now that Citrix and Microsoft have added a feature they both call “Dynamic Memory” to their products. But how effective is Dynamic Memory compared to the multi-layered memory management technologies in vSphere?  In both XenServer and Hyper-V, Dynamic Memory relies on just a single method – guest ballooning – to reclaim memory from guests when the hypervisor needs to find more memory for other VMs once physical RAM is fully allocated.  As we found when engineering ESX, ballooning is important, but it isn’t fast enough to cope with rapid spikes in memory needs or times when all your VMs are using their full memory allotment.  That’s why we designed ESX and ESXi with all four levels of memory reclamation technology, including the last resort safety net of hypervisor swapping, which is essential to avoid guest freezes when guests are all working hard and memory is under pressure.

Ballooning alone isn’t sufficient.  The Taneja Group took a look at XenServer 5.6 and its performance under conditions of memory oversubscription in their “Hypervisor Shootout” study of VM density. The results were striking.  XenServer 5.6 with Dynamic Memory was able to keep up with vSphere 4.1 VM density, but only when the guests were running a light load.  When the workload in the guests was increased, XenServer suffered, “a performance penalty that was surprisingly high and unfortunately consistent.”

Taneja_fig_10

Will the new “Dynamic Memory” feature Microsoft has added to Hyper-V R2 SP1 fare any better when memory is overcommitted and VMs are busy?  According to Microsoft, Dynamic Memory improves their VM density by 40% over the previous version that maxed out before memory could be overcommitted.  It’s an improvement, but it won’t be enough to catch up to vSphere 4.1, which showed in Taneja Group’s testing, “an overall VM density advantage of two-to-one.”  As seen in XenServer 5.6, reliance only on ballooning hits a wall when your VMs actually need all the virtual RAM you’ve allocated to them.  You can see signs of that effect in the painfully slow 2 second max response times in the VDI test Microsoft ran to justify their 40% improvement claim.  Apparently, entire Hyper-V R2
SP1 hosts can also find themselves frozen by memory shortages when Dynamic Memory is enabled.  Just as Microsoft has reluctantly come around to adopt the VMware way of doing things such as live migration and ballooning, perhaps we’ll see them grudgingly admit that hypervisor swapping and memory compression are essential to handle those times when memory pressure hits a peak.

Other Dynamic Memory Curiosities.  To permit many VMs to be packed onto a Hyper-V host, Microsoft has come up with a scheme where VMs are powered on with a low “Startup RAM” setting (512MB is recommended) and memory is then hot added as the guest needs it.  You can probably think of lots of reasons why this is a bad idea.  How about:

  • Hot-adding RAM to VMs will play havoc with system monitoring tools that will now see the memory allocations of VMs changing randomly – administrators can expect a barrage of “Low Memory” alarms.
  • Hyper-V Dynamic Memory will now only work with the latest versions of Windows that support hot add RAM.  Running older Windows 2003 service packs or Windows XP?  You’re out of luck.  Linux support isn’t even mentioned.
  • Since Microsoft hasn’t yet invented “hot remove” of RAM, Dynamic Memory can only ratchet up the RAM allocation of a VM.  Getting back to the “Startup RAM” setting will require a reboot of the VM.
  • Applications requiring more than the “Startup RAM” amount that perform a pre-check will fail immediately.  Microsoft’s own software installers run into this problem.  Microsoft’s workaround is to trick Hyper-V into hot adding memory to the VM by running MS Paint before starting the installation – I’m not kidding.
  • Microsoft has now added a whole new range of complications for software vendors to test and certify now that their apps must cope with suddenly changing amounts of memory.  Don’t expect your software vendor to support their product in a Hyper-V VM with Dynamic Memory enabled until they’ve done lots of recoding and testing.

With VMware vSphere, VMs look and act just like physical machines .  Any guest OSs and any apps or monitoring tools in the VMs see a consistent, fixed amount of installed RAM.  We thought it would be easier for everyone if we put the memory management intelligence into our hypervisor rather than relying on the guest OS to perform some unnatural acts to enable good VM density.

There’s no need to disable Windows security to get great VM density with vSphere

Lastly, I’d like to clear up some confusion on Microsoft’s part.  They’re claiming that VMware is telling customers to disable a security feature in Windows guests called Address Space Layout Randomization (ASLR) to improve VM density.  ASLR is intended to make things harder on malware writers by scrambling the locations of Windows processes in memory.  [Whether it’s meeting that goal is up for debate as Microsoft’s own security team has acknowledged that the bad guys have figured out how to bypass ASLR.]  ASLR makes it somewhat more difficult for the ESXi Transparent Page Sharing feature to de-duplicate memory, so you’d expect that turning it off would boost VM density in vSphere.  The Project Virtual Reality Check team confirmed in their September 2010 report that disabling ASLR produced a modest 16% improvement, but the authors clearly warned, “that Project VRC does not blindly recommend disabling ASLR. This is an important security feature.”  I thought the Project VRC findings were interesting and I mentioned them in a session I gave at VMworld 2010 Europe called, “Getting the best VM density from your virtualization platform.”  I also cautioned that disabling ASRL would reduce security and was best reserved for isolated testbeds where absolute maximum VM density was desired.  Here’s exactly how my slide put it:

ASLR

Note the “lessens security” warning.  Somehow the folks at Microsoft read that as a recommendation by VMware that our customers put themselves at risk to use our product.  Maybe they’re also the sort of people that see hot coffee as an unacceptable danger to society.  However, I think most would agree that the indignant posturing is just a way to distract users from the less than dynamic qualities of “Dynamic Memory.”