February 4, 2012

Hello Hyper-V : Meet Reality

First off stop reading this and go read Eric Gray’s post on the new Microsoft Virtualization Team post. He does a great job of pointing out the hypocrisy of marketing speak from the MS Virtual Team.

I am writing this blog post to address some specific annoyances in reasoning. While I have made a career in being a Microsoft guy (along with VMware, Cisco, EMC, and Nissan sportscars) I have some serious problems with the marketing pitch around Hyper-V.

So let me attack these head on. First off Chris Steffen states:

VMWare claims to support 4x more OSes that Hyper-V, but what does that really mean? When Microsoft lists an OS as supported, they COMPLETELY support the actual OS installation in the VM and you can call Microsoft support on that OS. Microsoft has support agreements with Red Hat and Novell specifically for this purpose.

So let’s be clear. What does Microsoft support in this case? Do they have support staff on hand that will work with the customer on Red Hat or Novell OS configuration? Would you trust Microsoft to touch your device information files on your Linux host? And to be quite honest, since Red Hat and Novell fully supports their enterprise products within the VMware environment. What is the real difference?

So let’s sum this up.

On vSphere if I have a problem I can:

  1. Call VMware for hypervisor specific issues (experts on this layer)
  2. Call Red Hat or Novell to get full support for OS specific issues (experts on this layer)

On Hyper-V if I have a problem I can:

  1. Call Microsoft for hypervisor specific issues (experts on this layer)
  2. Call Microsoft for OS issues (not-experts) and likely be transferred to step 3
  3. Call Red Hat or Novell to get full support for OS specific issues (experts on this layer)

So the real benefit Chris Steffen points out is an extra possible step. In the end my support coverage is the same at worst. Although I would be very curious about the actual level of knowledge between Linux support/Hyper-V and Linux support/vSphere. But, I can’t prove that point yet. And outside of these two specific operating systems flavors vSphere is light-years ahead. According to the current checklist, vSphere supports 48 flavors of OS compared to Hyper-V’s paltry 13.

Now to the next item:

Also, many of the OSes that VMWare claims to support are only supported by the Linux community – not taking a shot at the Linux community here, but most do not have a formal support organization. This leads me to question why they would be used in an enterprise environment. Also, those Linux distributions can be run under Hyper-V, using the Linux Integration Components Microsoft has available for download and the drivers which are in the 2.6.32 Linux kernel release. In this case, customers wouldn’t be able to call Microsoft for support for the OS, but would work with the Linux community, just as they would with VMware.

So this is pretty simple. The point here is: don’t use open source software. He states that VMware and Microsoft have the same community support so it is just a case of commercial vs. OSS and not a hypervisor argument. I would point out that community support is not only robust for vSphere but also VMware has guides, links, and walkthroughs on their own site (in a very easy to use setup) for how to implement multiple flavors. I wonder how easy it is with the Hyper-V side of things. Since OSS is not the argument here feel free to post OSS success stories in the comments.

 

Now for the fun part:

Reality: The Microsoft solution does not allow for over subscription of critical resources, but you shouldn’t do it anyway.

Oh no! I did not know this. Well I hope he is going to explain why at least.

The core of the VMWare argument is that you can somehow get “something for nothing” – that there is some kind of magic that comes with the over subscription of RAM using VMWare that is the silver bullet regarding memory management.

Wait a second, the argument is “something for nothing”? So efficiency is zero sum result? So I guess there goes thin provisioning, thin-client computing, or any other “thin” (read: effecient) technology. I better go shutdown my Windows Terminal Services farm too because I must be not really gaining anything.

So without the sarcasm, this is utter nonsense. He does not actually attack the technology or approach. He does not talk about direct risk or that fact that all efficiency models require management. Just like you have to manage the amount of users on a Terminal Services server you have to manage use on a vSphere cluster (notice I said cluster, not host. DRS much?). There is always inherent risk in higher utilization rates. That risk is managed by proper operational abilities. With vSphere these are clustering with DRS which allows automated movement of VM’s across hosts based on utilization and vCenter alarms which set low water marks against memory utilization. So with vSphere I have the option to take on operational responsibility for risk in exchange for higher efficiency (see $$$). The reason this is not zero-sum is obvious. I manage out the risk with a mature hypervisor (vSphere) gaining benefit I can never get with Hyper-V. With VDI and newer deployment models using virtualization, this can be a huge cost savings.

To leverage memory management in ESX to the fullest, one would have to fully burden the host beyond the physical memory. If you don’t, you really aren’t using memory overcommit.

Burden. Got to love that word. Puts an emotional spin on it. You can picture it right? The poor ESX host crawling across the data center will all the VM’s on it’s poor weary back.

Efficiency = lower total cost of ownership. The “burden” is your host doing more work for less money. I wonder if trucking companies talk about weight loads as “burdens” upon their poor MAC trucks.

 

Ok, one more:

Let’s go back to Basic Computer Architecture 101, and the example of the water pipe. There are limits to how much water you can push through a pipe at any given time, and the more taps that you add to the pipe, the longer it will take to fill up a bucket at each of the pipes. Hyper-V uses the best practice of moving a single VM as quickly as possible, using the entire bandwidth available to complete the transfer. Also, it is important to point out that without a modification of the host setting, VMWare would limit the migration to 4 VMs at a time (presumably for the same bandwidth considerations). The idea of moving 40 VMs all at the same time (as mentioned in the article) is not something that would be recommended, ever, regardless of platform.

Nice of him to explain throughput constraints for a kindergarten class. I would like to show a comparison of VMotion vs. Live Migration speeds (especially on my 10GE FCoE stuff) but instead I will keep it simple.

Why? Why can’t I do this with Hyper-V? Isn’t it because they don’t trust me? Or it is because they can’t make it work without sacrificing stability?
vSphere lets you not only do more but, also lets you do less. In other words, the mechanism is stable enough that throughput is the limitation (the water pipe) and not the stability of the mechanism (Hyper-V Live migration). Microsoft’s limitation on this points out a possible stability flaw and not a risky endeavor. What is also fascinating is the focus on making it “quicker“. Why does it have to be so quick? Are they afraid the VM’s won’t get there on Hyper-V if it takes to long?

 

There is a lot more to point out but, instead I will let someone else have the fun. I am not an anti-Hyper-V guy. I am an anti-F.U.D. guy. I would much rather Microsoft focus on providing a cheap product for small shop markets. In my mind that is what they designed with their product in both cost and feature-set. Even though VMware has some nice offerings also – see here & here

Also, I claim originality rights to the term: “DRS much?”. Feel free to tweet it like crazy :)

Comments and criticisms are welcome and appreciated.
.nick

 

Reblog this post [with Zemanta]

Virtualization & Abstraction : The New Paradigm

This post is inspired by this outstanding post by Chuck Hollis (@chuckhollis) and this one by Chad Sakac (@sakacc).

Chuck mentions my favorite way to summarize what virtualization encompasses: “abstracts logical from physical”. What makes abstraction critical is that it breaks historical dependencies that develop as technologies are built over time. I have said this phrase hundreds of times over the last four years of my career and in my mind it translates into an incredible paradigm shift in data center approach over the next ten years.

A good example of this is the push to service-oriented architecture design principles in the enterprise application space over the last decade. The whole gist was to enable business functionality to achieve independence and agility by breaking hard coded dependencies to platforms and systems. A loosely-integrated system can provide value to multiple business units by removing the overhead of inherited designs. Any ability to move quickly to market with business features brings competitive advantage. In simple terms, removing boundaries opens more opportunities.

Virtualization is the same approach with the end goal of the four food groups (CPU, memory, storage, networking) becoming commodities that can used as needed and where needed.  The status quo has been large CapEx investments in infrastructure where efficiency was limited by the boundaries of the physical needs and the return on labor cost to optimize. Even with a large team invested in tuning and sizing, the organic growth of the business can waste resources quickly as usage patterns change and new infrastructure is purchased.

The virtualization of hardware resources, effectively coupled with the ability to treat hardware as single pools, removes that physical boundary. This allows single investment in infrastructure resources to be carved up for multiple needs. It also allows the refactoring and control of these resources without the large operational cost and risk that physical resources historically bring. This abstraction means that a single server is no more special to an application than another; which means I can move, change, add, remove, or make a host of operational changes without risk. The virtual machine is coupled with the service that needs it and therefore applies fewer boundaries.

With the release of VMware’s vSphere product, critical aspects have been added. This includes the extending of network control to the virtual machine with Cisco’s Nexus 1000v. Now, both management and security of this layer can follow standard process of governance and operational models.

And now with EMC’s new FAST and the deployment models around the V-Max unit, the approach to storage is following the same design principles. If my data workloads are no longer bound to physical boundaries then I can deploy, react, and manage with less risk and more efficiency. I can focus higher expense storage at specific business needs when they need them and maintain cost effective use with lower cost storage on workloads that are diminishing. This translates directly into higher efficiency and lower total cost. Better yet, with approaches like NPIV, virtual machines can be matched to security, quality of service, and metrics. These features extend management, service-levels, and security on the storage layer to the virtual machine as well. Storage abstraction is the last great milestone to the virtual machine becoming the foundation of a data center. EMC is making a huge investment in both technology and people to make this a reality.

Despite the “coolness” factor in abstraction, the one important benefit is simple. Virtualization of servers, storage, networking, and ultimately business functionality brings efficiency. With vSphere I can take a set of CapEx impacting infrastructure and achieve higher utilization, less operational management, and be faster to market with features. To the business, this translates into a competitive advantage that can be measured. Because this hardware can be incrementally grown (EMC V-Max, vSphere), I can horizontally scale with growth demands and bring agility to change as well. And with new models like the VCE vBlocks even the large and complex standup cost in time/effort can be drastically reduced.

In the end only the goal of making people and business more successful is important. The abstraction that virtualization brings opens possibilities where they have not existed before.

As always, comments and criticism  are welcomed.

Reblog this post [with Zemanta]

Optimization vs. Scaling: How virtualization affects the scorecard

A typical server "rack", commonly se...
Image via Wikipedia

Many times I have seen situations where an application or process grows incrementally to a point where it is no longer able to meet it’s SLAs (whether official or imaginary). The cause of this can vary but is usually:

  • Overworked/Unbalanced teams -  Too much effort dedicated to new feature-add and not enough to technical debt
  • Poorly planned systems – Designs for immediate need without taking into account needs for things like instantiation or scaling of decoupled components.
  • Poor maintenance/understanding – Lack of knowledge or effort to tune application/process to more effectively use resources. This can exist in both the application and infrastructure groups.

Usually the performance degradation is known early on but accepted because the business users are not making a big enough stink; or at least not big enough to reduce the drive for new features. In addition, lack of monitoring and baselining of application performance is a critical problem. It eliminates the ability to effectively plan for growth and manage team resources.

Eventually the impact reaches a point where someone significant (business executive) resets priorities to fix it (technical debt due date). Many options will be evaluated immediately, from trying to buy time by tuning components to finding misconfigurations. However if no easy answers exist, it usually comes down to two options.

  • Optimize the application/process (Fix the code)
  • Scale the application/process with faster hardware (Throw metal at it)

Both of these options impact the same core factors: time and money. Depending on what time of the year it is, what the next feature would be, and what staffing is available, the choice can go either way.

Optimization has benefits in that better running code has long-term cost effectiveness built-in. But, optimization can have a reduced rate of return when repeatedly used without a complete architectural rewrite. Coupled with this, optimization often consumes productivity from the same teams that were unable to spend cycles maintaining it the first place. Also, the cost of optimizing is can be much greater than the labor involved. It includes what a possible delay on new features does to the firms overall revenue and commitments.

Scaling benefits from not directly affecting the product teams and being more focused on configuration and infrastructure resources. Scaling is also usually easier to estimate and deliver being that both the current application design and hardware resources are usually known. Where scaling can lose ground is in risk and cost. Anytime a change is made, a risk is taken. Moving a application from a set of hardware resources whether server, SAN, network, can end up being more disasterous for the business than the performance issue itself. Failues in configuration, QA testing, implementation, and planning are a dime a dozen. Cost can also be a problem when dealing with a fixed budget. The next incremental step to scale might be a big pill to swallow given the wiggle room available. Along with hardware itself, cost can be found in long-term commitments to power, space, cooling, and staffing to maintain ever increasing data centers.

Another factor is what I call the optimization-bias. If scaling will cause a IT leader to both beg for more money and possibly incur the risk of an outage, she/he may decide to trust in the application team instead. It is better to risk schedule under the covers than business ops and budget above the table.

This is where Virtualization can change the balance of this decision by improving the agility, cost-effectiveness, and reduces the risk of migration for scaling significantly. I see Virtualization as both a layer and a toolkit. It directly changes the balance of choices in the following ways:

  • Reduced configuration, schedule impact, and risk in migration
    • With an application that resides on virtual machine(s) the migration to new servers, SAN, or networking can be performed without a single change to the configuration of the application itself. Technologies like VMware‘s VMotion, Storage VMotion, and partnerships with major storage vendors such as EMC, NetApp, and 3PAR allow the application to truly be treated as an object. This eliminates the need to build platforms in parallel at anything above the hypervisor layer. This can be made even simpler with newer stateless platforms such as Unified Computing System (UCS) from Cisco which can reduce the hypervisor provisioning dramatically. In the end this can remove the need for QA resources and shorten implementation schedules. In some small cases migration can even occur while the application is under load.
  • Better efficiency and life-cycle management
    • Server consolidation has been a foundation of the VMware platform for a long time. When an application is primarily vertically scalable the efficiency of virtualization becomes a part of the life-cycle. An example is a database that must reside on a single server (not easily decoupled). In a physical migration, after the new server has inherited the application, the old server is placed into an equipment stack. Utilizing the old server for another existing application means starting all over again with another migration with the same risks. More often then not, this resource sits idle until it depreciates off the books or is needed in a lab.
    • In a virtual migration the new server would be added to an existing or new VMware cluster. When the application is moved to the new server the old server is still present and available for use. In fact with technologies like VMware DRS, this server would be immediately used for existing application loads. The available resource are equal for both physical and virtual models. However, the virtual model abstracts resources as a pool. This promotes efficiency in the long run and possibly significantly reduces hardware resource management and ultimately data center cost.

The advantages of virtualizing mean scaling as an option gains ground against optimization. Though every situation is different, the added sophistication and agility of virtualization provide any IT leader with possibilities they may not have had.

By far my favorite part of Virtualization is what it can do when coupled with a very well designed application platform. By designing application platforms with the intention of being decoupled and horizontally scalable, Virtualization can be used most effectively. Single components of the system can be quickly migrated to new hardware as needed, loads can be dynamically managed by VMware DRS, platforms can be quickly instantiated for new customers, and resources can be leveraged across physical locations with reduced cost. The major project of my career, of which I am working on right now, is this goal.

As always please comment if you agree/disagree.

Reblog this post [with Zemanta]