Has VMware won the Hypervisor war? … and who cares anyway… (Hyper-V 3 & SCVMM 2012, RHEV 3, XenServer)

Posted on November 21, 2011

by +

I’ve made it a habit to kick off every larger presentation by asking the audience to indicate adoption of virtualization in their environment. Suffice to say that even just 2 -3 years back barely 10% would indicate any serious evaluation or large scale production use outwith the VMware realm.
And when I talked enthusiastically about Virtuozzo, XenServer, VirtualIron and Hyper-V the eyes of many glazed over until we came back to VMware. And it is difficult to describe but there was (is) a distinct pride, an admiration, almost the feeling of a “cult following” among the VMware users.

Just 2-3 years back barely 10% would indicate use of hypervisors other than VMware – forward to 2011 and over 50% are actively evaluating alternatives …

Forward to end-2011 an the landscape has seriously changed.
Today I barely run a session where less than 50% of the audience lifts their hands and while indication of large scale production use is still limited, the appetite for virtualization alternatives is incredible.
A survey in spring 2010 showed that 78 percent of the 243 respondents said they were using VMware while 38 percent said Microsoft Hyper-V. In the 2011 survey of 250, the gap closed to 59 percent for VMware compared to 53 percent for Hyper-V. Citrix XenServer users doubled year over year from 9 percent to 18 percent of respondents. (statistics on market share vary greatly and are often skewed due to free downloads which don’t result in production use).

2012 will be the first real year of a “hypervisor war” and it will decide the battle in the private IaaS cloud market and its interconnected cloud layers (PaaS, SaaS etc) …

Just two weeks ago I had a client exclaiming that VMware has won the hypervisor war years ago and moved on to fight the cloud war. What war? Let’s face it, there were merely a few “uprisings” without real alternatives…
I’m risking to sound over-dramatic but 2012 is shaping up to be the first real year of a “hypervisor war” and it will directly impact the battle in the private IaaS cloud market and therefore shape the entire ecosystem of the interconnected cloud layers (PaaS, SaaS etc).

What has changed?

  • VMware has evolved from the cool underdog that couldn’t get it wrong even had they tried, to one of the “big players” now subject to the same scrutiny when releasing a new product or changing licensing.
  • Partner loyalty in some areas is exposed, there is increasing portfolio overlap (competition) with their large OEM partners (a critical route to market) and an unsettled technology-partner community that takes a sigh of relieve every time the latest VMware product has not erased it’s ecosystem by branching out into their (management, backup, storage etc) territory. All open to alternatives …
  • Most importantly – the arrival of competitive products will seriously challenge VMware’s position. Most will not challenge VMware’s technology leadership but they will aim to provide (at least) ”good enough“ alternatives or vertical  solution to gain market share. They will target the cost-conscious customer segment (and who doesn’t fall into that category today?) as well as customers desperately looking for a multi-supplier strategy to enable price-negotiation leverage when dealing with VMware.

    So who are the contenders? (and don’t mistake this as an anti-VMWare rant given my genuine respect for the company)

The Contenders

Microsoft
VMware’s biggest threat will undoubtedly be Microsoft with it’s new System Center 2012 release in the first half of 2012 (possibly April). Do not make the mistake and discount this release as “one of those minor “updates”. Especially the new Virtual Machine Manager is clearly an altogether different animal and having followed it since the initial beta in March I can only advise to spend some time evaluating it.

… imagine all the annoying shortcomings SCVMM has today had been magically addressed and add some impressive cloud functionality on top … now you are pretty close to what SCVMM 2012 will look like …

How to best summarize the improvements without going into detail …?
Well, think of all the shortcomings that annoy you with with the current version of SCVMM (whether the awkward dependency on SCOM when implementing DRS-like function, the non-integrated high availability, lack of vApp-like constructs and power management as well as the inflexible virtual network etc etc … ).
Got it? Now imagine most of those shortcomings had been eliminated/improved … then add a fundamentally different fabric management and service template approach, bare metal deployment capability and put an impressive private and hybrid cloud functionality with App Controller (Concero) on top … done …? Now you are pretty close to what SCVMM 2012 will look like (more details in a previous post or here).

It’s important to understand that the SCVMM release will address the hypervisor management layer, not the Hyper-V layer itself. However, the end of 2012 will also see the release of Windows 8 and with it Hyper-V version 3, which will address today’s underlying scalability and functionality limitations. A good summary of what’s to come is here.

HyperV3


Microsoft will have a strong contender by YE 2012 even if you take the – in the meanwhile ongoing – evolution of VMware’s product portfolio into account and many of my recent clients have confirmed that they are actively evaluating it.

Red Hat
One of the biggest surprises at VMworld Europe for me was the interest in our setup of Beta RHEV 3 (Red Hat Enterprise Virtualization). The feedback made it clear that the improvements in this release finally allow Red Hat to provide a credible “good enough” alternative and that not only for clients specifically pursuing an open source based alternative.
My initial impression was “wow, the GUI is very intuitive (yes, I have to admit vCenter-like in some respects … and that is a good thing).

RHEV3

Red Hat has finally ported its RHEV Manager to Java, JBoss and PostgreSQL, removing the dependency on Windows (except for the client which still has some dependency on .Net in this release). RHEV 3.0 also has a nice little “power user” portal that allows users to create and manage their VMs.
There are still a couple of basic things Red Hat needs to (and will) address in the next releases, the lack of live snapshots and live storage migration among them.
Besides the user experience, one of my major concerns with RHEV has always bee the lack of a consistent API and the challenges that poses for the much needed build-up of an ecosystem around RHEV (by enabling ISVs and OEMs to easily create extension) – therefore the addition of a RESTful API for integration with RHEV Manager is extremely welcome.
RHEV already had the advantage of a unified console for server and desktop virtualization and the graphical user experience with the SPICE protocol on our setup in Copenhagen was seriously impressive.

Red Hat is unlikely to challenge VMware head-on but will gain market share by identifying vertical stacks and targeting suitable customer segments that are craving for alternatives or open source based solutions

RHEV 3 (currently public beta) will be released early next year and in conjunction with the previously discussed Cloud announcements Red Hat has an opportunity to make an impact – not broadly challenging VMware head-on but by identifying customer segments and workloads that are craving for alternatives or open source based solutions.

Citrix
Now I really wish I could confidently add XenServer/Center to the list of contenders for 2011 … I have never made a secret of my admiration for Citrix to create (for a long time the only) credible alternative to VMware, sustaining to innovate again and again and to provide a free version in a space totally dominated by VMware and with a Microsoft relationship that makes clear positioning tricky. And it’s not the lack of functionality that makes me hesitate, XenServer/Center 6 is looking strong, but the lack of a well communicated vision for the role and integration of XenServer in Citrix’s Cloud plans.
After announcing project Olympus (OpenStack + XenServer) which reassured the importance of XenServer, the acquisition of cloud.com (managing multiple hypervisors with the CloudStack platform) makes the future role of XenServer unclear. Combined with the fact that Citrix happily collaborates with Microsoft to allow System Center 2012 to manage XenServer and therefore essentially exposes it’s own future virtual infrastructure control point is not reassuring.
Can XenServer seriously survive a Hyper-V 3 onslaught even if Citrix decided to (seriously) take on Microsoft (which is unlikely anyway).

There is a lack of a well communicated vision for the role of XenServer in Citrix’s Cloud plans – can XenServer carve out a niche as VDI Hypervisor?

What’s the alternative ..? Citrix is likely to find its vertical “niche” in the cloud business (and e.g. the end-user computing one is not a small niche). XenServer could help fulfilling that very role if Citrix focussed further on the creation of a true VDI hypervisor. Currently we still see the majority of XenDesktop deployed on vSphere but with more vendor specific optimizations like IntelliCache and GPU passthrough Citrix could carve out a niche for XenServer.

Why should we care?

But why should we even care about the boring old hypervisor in the age of Cloud Computing?
Many increasingly (and mistakenly) describe the hypervisor as “replaceable commodity” with little vendor stack dependency. And yes, if you looked at this thin abstraction layer in isolation you would have a case but that would also imply that seamless mult-hypervisor management (to enable “rip and replace” of Hypervisors) is a reality today.

Seamless mult-hypervisor management is not yet reality – today’s solutions are merely marketing tick boxes and strategic directions rather than usable alternatives ..

It isn’t … efforts like Microsoft’s SCVMM’s capability to manage VMware environments and VMware’s XVP to manage Hyper-V are today merely marketing tick boxes and strategic directions rather than usable alternatives. The mammoth effort involved in creating full replacements providing the equivalent functionality to the native vendor management app and the challenge to keep up with the constant flow of new (competitive) vendor releases casts doubts that this will soon become a reality.

So be wary of such messaging – feedback from my clients almost unanimously show that – while all would love to have this “single pane of glass” management for heterogeneous hypervisor environments – they prefer to use multiple native vendor tools (with full functionality) to a single tool (with only a subset of functionality).

Without today’s vSphere footprint there would be no case to use vCloud Director – any hypervisor market share loss will directly impact the VMware’s private cloud market share

Make no mistake, VMware’s strength is the foothold in the virtual infrastracture space, directly influencing the higher management layers – without the vSphere footprint there would be no case to use e.g. vCloud Director (which manages exclusively vSphere environments). Any hypervisor market share loss will directly impact the VMware’s private cloud market share. So yes, we should absolutely care about these dynamics.

It is important to understand that this dependency does not necessarily apply to the higher cloud management layers. We have already seen many examples of 3rd party cloud management platforms successfully drive the underlying e.g. vSphere or KVM-based infrastructures as independent target resource pools.
More on this topic in one of my next rambles … ;)

So it’s undoubtedly going to be an interesting year but with VMware continuing to innovate at incredible speed the main question remaining is how much of this “appetite” for alternatives will also result in “consumption” … ?

Posted in Citrix, Hyper-V, KVM, private cloud, Red Hat, VMware, XenServer | 23 Comments »



VDI – Success or Failure? – or – Why “VDI-bashing” is popular….

Posted on October 16, 2011

by +

I have to admit that it is slightly irritating to fly home from ‘yet another’ client meeting and stumble across ‘yet another’ discussion (OK, let’s call it rant) about VDI on twitter or blogs. Especially when the meeting was actually a positive one that included VDI – so yes – my opinion upfront – there is such a thing as ” successful VDI” (and note that I’m absolutely not saying all projects are)

There are good and even great (critical) discussions and articles with absolutely valid points like this one (discussing VDI’s obvious challenges and shortcomings) – and then there are others, rants driven by frustration, inability to position, feeling mislead and – IMHO the most dangerous ones … are the geniuses and visionaries who already live in 2015 and seem to easily dismiss everything that is today’s (boring) reality.
Don’t get me wrong, we (the industry) need thought leaders – and the best of us all have a visionary in them – but don’t assume genius = flawless/always right. Genius actually = high risk – yes, with potentially high returns. Dangerous, because thought leaders create followers – unfortunately often following blindly, amplifying and distorting originally educated, valid points until they become false generalisations.

Ultimately it’s our own (IT vendors’) fault, we are hyping up, creating bubbles, “re-inventing” our portfolios (translate into “re-badging”) and push our own agendas – so I completely understand where the backlash is coming from. With VDI we have surpassed the hype curve, face adoption and with it a level of disillusion. And suddenly VDI-bashing becomes … well … fashionable.

But enough of my rant, as you know, I am not associated with a VDI vendor, agree with many of the raised technical concerns, I work with server virtualization and cloud environments (so have nothing to “lose” if VDI quietly faded away) but here is my honest view and to make it as short as possible I’ll just paste without changes the comments I posted on Simon Crosby’s blog:  (also check out Tal Klein’s great comments here)

==== paste ====

“To me this boils down to: (and I’ll keep it short as I have the tendency to drivel on)

- Are we trying to solve a problem which does not exists? No. (so I slightly disagree with how you portrait the current state of physical endpoints – time spent on image management is unsustainable for many of my clients)

- Is the answer to the problem always VDI? No.

- Can the answer be VDI? Yes, as a tactical solution for specific use cases where TS does not fit.

- Can VDI be a strategic approach – yes, as a stepping stone (and I am aware of the inherent contradiction here) to enable logical separation of OS and app, mobile & user-centric app delivery.

- Why VDI as the “one fit all”: 1) because the VDI “vendor says so” – 2) because the clients prefers the simplicity of one architecture that can (potentially) cover all use cases (so both dubious but reality)

Bottom line – I can’t fail VDI completely – VDI has a space as tactical building block alongside TS (and others) – combined as strategic hybrid approach to facilitate future app delivery models. (all simplified in the interest of time)

=== paste end ====

This really sums it up for me – unless I’m working in an under-developed / left-behind part of the world – then todays reality for most is not a single “Nirvana-like” solution with a unified GUI, integrated modular sub-components, driving all use-cases of End-user-computing (or ideally all IT services) from the same platform.
If that was the case quite frankly architects would be out of the job. That’s what we do, match use cases with suitable technologies.
Clients have unsustainable problems today – the requirement for tactical solutions in the search of strategic ones is today’s reality, not everyone can “continue and wait”.

And yes, we absolutely should do that with a vision of how to enable emerging approaches. But too often have I come recently into client meetings instructed and prepared to talk about “cloud”, vSphere 5, SCVMM 2012 or RHEV3 and the client kicked off with e.g. “let’s fix today’s problem first – we need to migrate off vSphere 3.5 – so we want a vSphere 4.x (not 5) architecture from you …” (that particular one was one of the largest retailers in the UK just 2 weeks ago).

What’s my point ..? We have a tendency to get ahead of ourselves, forget what today’s problems in the accounts are, we sell one idea then happily move off to the next “fancy” thing, sometimes leaving the client with the mess and giving them a kick in the teeth by turning around saying things like “yes, this VDI thing is rubbish – I could have told you that in the beginning” – “you should really go with TS” or “you should wait for HTML5″ – replacing the same mistaken “one fit all” approach with another label.
The question is not “Does VDI work?” (of course it “does”) but rather is “Where does VDI fit?” and “Does it work for you?”
It’t should be one of our tools in a set of possible solutions to a client-specifc problem, neither the de facto answer to everything nor discounted by default.

If we sold VDI as the only answer then we were wrong (not architects or consultants) to begin with.

It is easy to criticise – the art is to make it work …

Posted in Citrix, VDI, View 5, VMware, XenDesktop | 10 Comments »
Tagged: , ,



(V) End User Computing-The Vision: Horizon, AppBlast, Octopus and “should I skip the VDI bubble”?

Posted on October 4, 2011

by +

(This is Part 5 of ‘Real-World VDI’ – see Part 1 “Multi-Vendor Stack, Part 2 “High-End Graphics” Part 3 “Cost” and Part 4 “Recent Announcements”)

As I said, the announcements at VMworld caused me to adjust my view of VMware’s end-user capabilities but more importantly on VDI as a whole.

I’ll try to cover this in more detail in future posts but let me outline it on a high level.

Honestly (having attended VMworld every year since the first one in 2004) I thought the moments where I sit in a session and think “wow, that’s amazing” were history … but I was wrong, Steve Herrod’s key note this year was (again) one of those.

VMware’s vision left me with a nagging thought … is VDI really the next logical step for “everyone” or is VDI  just the next ‘IT bubble’ we are trying to sell ?

And it left me with a nagging thought … is “VDI” (HVD) really the next logical step for “everyone” or as someone asked me in the past  “is VDI not just the next ‘IT bubble’ you are trying to sell me?” ;)

During the session VMware created a vision where Horizon takes central stage as an application portal (think “app store”) – presenting users their available (HVD) virtual desktop(s) alongside session based desktops (like XenApp) and even SaaS delivered applications.

ThinApp Factory will have the ability to be pointed to a repository of files (MSI, installed app) and convert the app automatically into a ThinApp-deliverable virtualized package. It integrates with Horizon so that “thin-apped” apps will show up side by side with the above (while ThinApp is likely to retain it’s “agent-less” approach, integration and advanced management can now be done using the Horizon agent)

Add to the picture Octopus, a drop-box like service for data management/sharing, Horizon mobile (new name for MVP, the mobile hypervisor enabling to run personal and business image on the same device) and virtual profiles for persona management. You suddenly have an incarnation of a next-gen user-centric and portable end-user environment.

The wow-factor however was delivered courtesy of AppBlast – a preview of a VMware environment that can potentially deliver ANY application – remotely – to any device with an HTML5 compliant browser. So yes, you can e.g. run MS Visio on your Ipad! Check out this great demo Vladan Seget posted …

A natural question is: Should you really invest in something like VDI if this Nirvana of app delivery is “just around the corner” ….? Can you possibly “skip VDI” and simply wait for HTML5 or similar thinOS/web browser based approaches…?

In a very simplified view, my conclusion is that for many VDI is a stepping stone, a tactical solution that can achieve immediate benefits.

VDI can help to facilitate the (strategic) transformation of your approach to end-user computing.
It helps to modularize OS and applications, encourages user persona/profile management, centralization of resources and forces you to rethink the role of the OS in the future environment.

But more fundamentally it helps you to (strategically) transform your approach to end-user computing. It helps you to modularize OS and applications, encourages user persona/profile management, centralization of resources and forces you to rethink the role of the OS in the future environment.

VDI might or might not be a subset of your future end-user environment. Realistically we will have hybrid environments of physical desktops, session-based remote desktops (aka Terminal Services), HVD and increasingly browser based app delivery for a long time in larger estates. The requirement to match use cases and user categories with the appropriate (financially viable) technology will not go away.

It would be foolish to think that next year all apps will be delivered through HTML5, Silverlight, Flex etc. I also know that many of my clients don’t agree with the statement that the “PC is dead” - and I certainly won’t give my laptop up yet – but (unfortunately) now carry it around (alongside!) my iPhone, iPad and my ability to access my virtual desktop remotely.

But there is no doubt that the role of the (e.g. Windows OS) – which historically abstracted vast amounts of physical HW to provide runtime environments for applications – is changing. The vision to have a ‘thin OS’ (think of it primarily as a runtime for a browser, not more) remotely delivering “any app” to “any device” is extremely powerful.

That is my view. But where do YOU think this is all going …? Are we getting ahead of ourselves or are we pursuing a feasible vision?

Posted in Citrix, private cloud, VDI, View 5, VMware, XenDesktop | 2 Comments »
Tagged: , , , ,



(IV) Rethinking my View of VMware’s End-User Capabilities – Is View 5 a Game Changer?

Posted on October 4, 2011

by +

(This is Part 4 of the ‘Real-World VDI’ series – see Part 1 “Multi-Vendor Stack, Part 2 “High-End Graphics”, Part 3 “Cost” and Part 5 “Future of VDI”)

The announcements at VMworld caused me to adjust my view of VMware’s end-user capabilities and vision.
So what did VMware announce and why will it impact the current pattern and future directions?

There are two streams:

  1. The immediate impact with the latest VMware View 5/vSphere 5 product announcement and
  2. The longer term impact with the announcements around Project Horizon, Project Octopus, ThinApp factory integration and the (potentially amazing) AppBlast

The first stream  is relatively straight forward but has immediate impact (as available now) …

So what has been announced with View 5…?
(among other features)

  • PCoIP enhancements reducing bandwidth (“up to 75%”) and increasing latency tolerance (up to 200 ms)
  • Support for Direct x 9 and OpenGl 2.1 using a virtual 3D graphics adapter
  • View Persona Management which integrates user environment capability (as discussed in ) to manage user personalization when deploying stateless desktops .

The promises on the PCoIP improvements will have to be proven in real-life environments but it is clear that VMware is focussed on addressing one of the major customer concerns aggressively.
The profile management was long overdue (and therefore welcome) and will utilise “just in time” retrieval of profiles in order to minimize the transfer impact. It will not address physical systems or allow profile management across different OSs – so 3rd party apps like AppSense etc. will continue to provide value (but might not be required).

Graphics Enhancements

So what is this 3D function? vSphere 5 allows you to create a virtual 3D adapter in the vm (virtual GPU) which essentially emulates the behaviour of a physical GPU. This has the great advantage that no (sophisticated) physical GPU is needed and that you can run multiple of those 3D capable vms on a single host (achieving potentially higher user density than HDX-3D (which has a 1:1 mapping between user and GPU), at the same time supporting a wider range of graphics workloads than RemoteFX – again, all without the requirement for a physical GPU).

VMware’s virtual GPU has the advantage that multiple users can share a single host for 3D delivery and no physical GPU is required in the server.

However, as the host has no physical GPU to do the work, the CPU ends up executing, which can create significant overhead. How much will depend on the nature of the graphics workload but in offline discussions we have heard that “it can potentially half the number of users per host” … we’ll have to wait for sizing guidelines (or test ourselves).

A secret weapon here might be the PCoIP offload adapter from Teradici which will allow to offset that overhead significantly – I’ll cover this in another post in more detail as we have already done testing internally but the card will not GA for a while. Alternatively - if you go to VMworld Europe – why don’t you come round to the IBM booth and I’ll show you that setup live!

Please note that the 3D feature has a dependency on vSphere 5. That means that you can’t get this functionality with earlier vSphere versions (although View 5 will support them). More importantly VMware cleverly introduced a “vendor stack affinity”. That means you will e.g. not get the 3D functionality with XenDesktop on top of vSphere (if this has been your ‘stack of choice’).

Going forward I expect to see more of this “vendor dependency” approach which will drive arguments for a “single-vendor” stack. Another example of this is Citrix’s IntelliCache – which depends on XenServer and VMware’s View Accelerator or “CBRC” which depends on vSphere 5 (the integration of CBRC – think “IntelliCache for vSphere” – has been delayed and is now a “future enhancement” for View).

If we review the graphics capability table now, we see that the picture has changed significantly.

Looking at the above table we can see that VMware has clearly addressed some of its weakest points in an impressive way with the latest release.

Does that mean VMware will displace Citrix and take the crown? Well, only time will tell but there are more aspects to this then “just” the product updates to View/vSphere

  1. Nothing will happen overnight, VMware will need to provide proof points for actual performance, scaling and overhead. The code is new so many will wait and see (regardless of VMware’s track record for code stability)
  2. Citrix end-user heritage and customer affinity is strong
  3. Citrix’s overall end-user portfolio (over and above VDI) is still considered unmatched by many as of today (XenApp, branch repeater, NetCcaler, Citrix Receiver portability etc).

Should we expect VMware’s PCoIP/virtual GPU to deliver better high-end graphics capability than HDX3D?
I see VMware’s key advantage in providing the more flexible (user ratio) solution for “casual – medium” graphical workloads but expect e.g. HDX GPU pass-through to still have the edge when it comes to heavy/high-end graphics (with the trade-off in user density). Also, XenDesktop 5.5 supports now RemoteFX (in conjunction with Hyper-V) so can overcome some of the user density issues with hybrid environments.

I would certainly expect VMware to be keen to deliver (physical) GPU offload and more client-side offload in a future release to complement the current capabilities.

VMware has certainly cleared some major road blocks with the latest release and we should expect increased interest in View-based deployments

Saying all that – VMware has certainly cleared out some major road blocks with this release and I’d be honestly surprised if we don’t see an increased interest in VMware View-based deployments.
I for one certainly consider View 5 a far more attractive proposition than before … and see a good potential for HVD-based deployments.

As mentioned before, beware of increasing stack dependency for advanced functionality when using multi-vendor stacks and evaluate carefully when considering those going forward.

So what about the second “stream” … how is the vison for end-user computing VMware (and others) create for end-user computing impacting the future of VDI …? Have we really reached the “End of the PC era” …? Read on if you are interested …

Posted in Citrix, IBM, private cloud, Uncategorized, VDI, View 5, VMware, XenDesktop, XenServer | 7 Comments »
Tagged: , , , ,



(III) Real-World VDI: Cost – “Help I’m loosing money with my VDI offering!”

Posted on October 3, 2011

by +

(This is Part 3 of ‘Real-World VDI’ series – see Part 1 “Multi-Vendor Stack, Part 2 “High-End Graphics”, Part 4 “Recent Announcements” and Part 5 “Future of VDI”)

Why do we see increasingly see cost concerns in VDI environments? As I mentioned, many of my recent customers already have an existing POC, pilot or (smaller) production environment in place and consider moving to their phase 2. Some of those also use it as external or internal desktop clouds and desktop as a service (DaaS) offerings. And yes, having proven that the concept technically works means that they are exploring to include more complex user categories. But more importantly, moving forward many realise that optimisation is needed to make the cost cases for desktop virtualization viable.

Typically pilots sacrifice optimisation or elaborate business models for the sake of simplicity in order to “make the concept a success” … the result …?

Typically pilots sacrifice optimisation or elaborate business models for the sake of simplicity in order to “make the concept a success”. Aspects like stateless desktops, sophisticated user profiling, application virtualization, storage alternatives, hybrid models with session virtualization, tiered charging etc are often afterthoughts or consciously targeted for “phase 2″ – resulting in challenges for the financial model.

One of the biggest mistakes that can be made upfront is to sell VDI as “cost saver”. One of my first questions in kick-off meetings typically is (it really is) “what is your main driver to adopt VDI?” and if the answer is “cost reduction” then I know we are off to a bad start.

But in the interest of time I will not go into the challenges of quantifying the “soft benefits” of VDI and how much more difficult this is than making the case for server virtualization where the consolidation ratio only typically makes it a “no-brainer”. I’ll rather describe what has been an underlying root-cause for many of the experienced cost challenges with VDI architectures.

Let me simply re-enact an actual conversation I had with a service provider in the Nordics (with a medium production environment – looking to increase capacity for the DaaS like offering but facing cost challenges).

He (client): “Our VDI environment is too expensive, we currently charge this particular client less than it costs us”

Me: “What’s costing you too much?”

He: “mainly our storage” (Enterprise NAS – no names)

Me: “Do you need it”?

He: “Sure! We are using it for our entire virtualization estate!

Me: “You mean server virtualization estate”?

He: Yes … and now also VDI.

Me: So let me ask you again, do you need it?

He: “As I said, yes … well … I think we need it … you tell me! Do we?

Me: “It depends … ” (his smile disappears hearing the “get-out” clause ) ”Do you use persistent or pooled desktops, how are your images delivered etc. …  ?”

He: “Non-persistent, PVS delivered”

Me: “Really? 100% pooled?” (as not very common across larger estates)

He: “yes, this client is” (…ah, sub-set)

Me: “So yes, that might be possible then, we could look into DAS for the PVS write cache, swap etc. and eliminate SAN. “

He: “But my guys told me that we need the NAS with lots of disks to satisfy the user numbers.”

Me: “Yes, we’ll need to look into IOPS requirement and capacity. We could use SSD-based local storage to get the IOPS but also need to check capacity (as they run >100 images on a 4 socket server)

He: “But does that not mean we lose VMotion and VMware HA?”

Me: “Yes.”

He: (a defensive frown appears now) “We can’t do it then!”

Me: “Why?”

He: “That would mean (planned) downtime and increased recovery time”

Me: (deliberately not explaining that you could achieve HA and VMotion-like uptime for VDI through broker and pooled images) – I rather made the point: “Do you provide higher uptime, quicker recovery time, faster deployment and faster reconfiguration times now than previously with physical desktops to these customers?

He: “Of course!”

Me: “Do you charge more for it?”

He: “Ehm, no, our clients said right from the start that they do not want to pay more than for their physical desktops”

Me: “Understand, but you currently lose money because you decided to offer a BETTER service for the SAME cost to them …”

He: (defensive frown disappears) “Hm, I suppose that’s true, so what shall we do?”

Me: “Do you believe your clients realize that they receive a better service?”

He: (pauses) “Actually… probably not … I mean it’s all working well, no outages, no downtime, but in reality they would only come to me if something failed, not because it’s working well”.

Me: 1) “Tier your offering”

“Create a base (e.g. bronze) offering which is providing at least the same level of functionality as their old physical environment. But do not add functionality if it costs you extra, you will probably find that it will be more flexible anyway. TALK to them about silver and gold offerings (that include those added levels of availability and flexibility). That will a) ensure that they understand what they get and b) tell you whether they are willing to pay for these benefits.
Based on the outcome create those offerings (or consciously not).

Common sense? Probably. But you would be surprised (I was) how many even larger clients fall into the trap. Yes, we all got used to eliminated downtime through live migration, recovery of failed workloads in minutes rather than days through integrated HA mechanisms, deployment and system reconfiguration times of minutes rather than weeks.

Most of us consider this (rightly or not) “standard” in the world of server virtualization.

Do NOT make the mistake and apply your server virtualization base-line automatically to your desktop environment – the challenging nature of VDI business cases does typically not tolerate this approach.

But do NOT make the mistake and apply this base line automatically to your desktop environment – not without understanding the implications – the challenging nature of VDI business cases does typically not tolerate this approach.

You can of course argue that there are various other (possibly more important) issues or maybe you just completely disagree with me here … if so, let us know what patterns and challenges you are seeing out there in the real world …

Interested in recent changes to the VDI landscape …? See Part 4 – Rethinking my view of VMware’s end-user capabilities – The Impact of VMware’s recent announcements…

Posted in Citrix, private cloud, VDI, View 5, VMware, XenDesktop | No Comments »
Tagged: , , , ,



(II) Real-World VDI – Advanced Graphics: DirectX, OpenGL with HDX & PCoIP

Posted on October 3, 2011

by + 



(This is Part 2 of the 'Real-World VDI' series - see Part 1 "Multi-Vendor Stack, Part 3 "Cost”,  Part 4 “Recent Announcements” and Part 5 "Future of VDI")

While most clients initially enthusiastically agree to address first the “low hanging fruits” of VDI, almost without exception I had the “dreaded question” come up consistently in the first meeting: “I have this user group with high-end graphics, video conferencing/3D graphics/engineering application (replace this with your graphics intense application) requirements … Can we deliver this with our approach…?

Make sure you understand what the clients view of “high-end graphics” really translates into … For some it was a 720p streamed video for others it’s an OpenGL 3 based CAD app …

Without ignoring the fact that many of these will require proof of concept type of activity in order to verify suitability, the first step for you should be to understand what the clients view of “high-end graphics” really translates into…

For some it was a 720p streamed video for others it’s an OpenGL 3 based CAD app. Understanding which version of the rendering API is used (e.g. OpenGL, DirectX) is a good first step to determine feasibility for HVD in general, alongside with the expected bandwidth requirements based on resolution, number of monitors etc.

Reality is that vendor/technology choice is crucial here – you will be able to achieve different levels of graphics capabilities and (important!) user density for a given use case. And always ask yourself throughout the process whether VDI really is the right answer for this use case.

So, are the capabilities really that different? It’s actually not that easy to get a simple view of how vendors compare here, so I created this little high-level table:

Please note that this table only addresses “advanced graphics capabilities” (like DirectX and OpenGL) which typically require GPU-like functionality in either the host or client device.
In order to e.g. display an avi video you won’t (e.g. for Citrix) require HDX3D (with host-side GPU) and will therefore not incur some of the restrictions and requirements listed below. In the same way, View 4.x with PCoIP is perfectly capable to display “casual graphics”.



Comment: Important, see “my changed view” for latest updates as it changes this picture significantly

References:
- XenServer 5.6 Multi-GPU Passthrough for XenDesktop HDX 3D Pro Graphics
- System Requirements for HDX 3D Pro

It becomes clear why XenDesktop/HDX has been favored by many when it comes to graphics delivery. Again, the last 3 large clients told me pretty much the same story after their initial evaluations or POCs. Efficient bandwidth utilization, good WAN performance and the option to display high-end graphics for e.g. workstation replacements were some of the key decision factors to select HDX. When using HDX3D (typically a small subset of the overall user base), the 1:1 user to host ratio is however very limiting. The GPU-Passthrough function (which is experimental prior to XenServer 6) is a great feature, but still only increases the ratio to one user (vm) per GPU.

Due to VMware Views 4.x lack of host (or client) GPU rendering offload capability none of these advanced capabilities can realistically be achieved. Standard graphics workloads can be delivered with similar user experience than HDX (but sometimes higher bandwidth utilisation).

RemoteFX overcomes the restrictive user to host ratio of HDX3D but can’t deliver full “advanced graphics” functionality and lacks WAN support, still a good fit for those use cases where only casual Aero-like workload is needed.

You might argue that high-end graphics capabilities shouldn’t be the main decision factor for a VDI platform – reality is that most customers want to have at least the option to cover those use cases in the future …

You might argue that high-end graphics are only a small subset of the overall user base requirements and shouldn’t be the main decision factor for a VDI platform – and I partially agree, reality however is that most customers I worked with want to have at least the option to cover those use cases (even if in the future) – so it has almost always been a consideration.

What one can’t argue with is the importance of the cost case for VDI … so let’s look at a common mistake made with that next HERE …

Posted in Citrix, VDI, VMware | 1 Comment »
Tagged: , , , ,



(Part I) Real-Life VDI – Patterns in the field: Multi-Vendor Stacks with Citrix, VMware, Microsoft and others …

Posted on October 2, 2011

by +

(This is Part 1 of a ‘Real-Life VDI’ mini-series – see also Part 2 “High-End Graphics”, Part 3 “Cost”, Part 4 “Recent Announcements” and Part 5 “Future of VDI”)

In this mini-series I’ll share my real-life experience based on some of my recent VDI client engagements. So it’s a reflection of what customers actually implement (and struggle with) all mixed with my comments, evaluation and opinion. In particular I’ll talk about the following aspects:

-   Software Stack – or – “Is there a common software stack for VDI in the industry today?”
(Is it Citrix, VMware, Microsoft or another vendor’s stack?)

-   Advanced Graphics – or – “I need advanced graphics, what shall I do, is it always out of scope?”
(Looking for an “at a glance” view of advanced graphics capability by vendor …? Then read on …)

-   Cost – or – “I can’t get to a viable price point for my VDI environment, what am I doing wrong?”

- Future/Vision – or - “Is VDI for me? How will recent announcements impact the future of VDI?”

While I’ve been involved with desktop virtualization for several years, this article was triggered by the emergence of “patterns” in my VDI projects over the last year (based on field experience, not some nebulous industry projections).
These patterns are currently common across customer from various vertical backgrounds – although a large segment of my recent engagements has been with European system integrators and customers in the financial sector, typically of larger size with targeted environments of 2k seats up to 100k seats. That also means that some of these implementations were deployed as private desktop clouds and Desktop as a Service (DaaS) environments

Most of those already had an existing POC, pilot or smaller production environment in place and consider moving to “their phase 2″. Phase 2 typically involves a larger production user base (> 500 seats), often aiming to include more demanding user categories (power users and higher graphics requirements) and distributed environments (i.e. branch offices).

This article will be mainly restricted to VDI (HVD “hosted virtual desktops” based architectures). However, I hope we all agree we’d be pretty poor architects/consultants if HVD was our only answer to every desktop infrastructure scenario, but please allow me to leave it at that …)

… I hope we all agree that we’d be pretty poor architects/consultants if HVD was our only answer to every desktop infrastructure scenario …

I started to write this article just before making my way over to VMworld Vegas and some of the announcements and discussion (including a particular enlightening one with Massimo Re Ferre from it20.info ) changed some of the dynamics in this space so I’ve added “Rethinking my opinion of VMware’s End User Computing capability and vision“.

But let me first share my thoughts as originally intended – that will also make the significance of the recent announcements clearer.

Let’s start with the software pattern … is there really something like a “typical” stack for VDI components …?


Pattern 1: The Status Quo of VDI Software stack: VMware or Citrix?

OK, I admit, this can be seen as a dubious title. Is there really a status quo (and can there realistically be one), why restricting it to VDI and what about the other players in the game …?
And that’s exactly the point, it’s currently not a clear-cut single vendor stack we see typically being implemented.

So the question I typically get is “Am I the only one with this odd software stack, there are just too many combinations of products from different vendors – what are others (in my industry) using…?”

“Am I the only one with this odd software stack …?”

The majority of customers asking me this question actually then go on to describe a very similar (multi-vendor) stack …

The majority of customers asking me this question actually then went on to describe a very similar (multi-vendor) stack -  including the last 3 large customers in a row over the last couple of months … and that stack was:

L1 (Hypervisor and hypervisor management): VMware vSphere

L2 (Broker/Session Management): Citrix XenDesktop

L3 (“Application Virtualization”): Microsoft App-V

L4 (Profile Management): roaming profiles/folder redirection and/or pilot to implement AppSense

As you can imagine, I am simplifying the layers in the interest of time and again, I am not making any claims that this is an objective representation of all customer segments, sizes and regions by any means. (I will cover other solution like Virtual Bridges’ Verde – which I’ve recently worked with – in a future post).

Affinity typically decreases with the higher layers so for instance App-Virtualization and Profile Management are less likely to be nailed down. Also, before I attract the wrath of VMware View fans, I am not making a general stack recommendation here, The purpose of this article is to convey what customers decided to implement after careful consideration.
(also please check out  ”rethinking my view of VMware’s end-user capability“)


Hypervisor of choice for VDI: ESX

A pretty obvious choice you might say… arguably the most mature, feature-rich hypervisor in the market.

In reality this is often an evolution of what the client did for server virtualization (rather than based on specific evaluation of suitability for VDI) … existing skill set, license agreements, confidence and relationships often make this the natural choice… .

Many clients confirmed that they are actively evaluating hypervisor alternatives for VDI attempting to reduce cost

Most however confirmed that they are actively evaluating alternative hypervisors attempting to reduce cost (with e.g. XenServer but actually increasingly Hyper-V), with one large customer actually stating that the directive given by their management is Hyper-V “whether they liked it or not” due to existing licensing agreements with MS.
The SP1 addition of Dynamic Memory helped and RemoteFX has the potential to address some of the graphics requirements, however missing NAS support (except CIFS) for Hyper-V and the upcoming release of SCVMM 2012/Windows 8 (which many see as first credible competitor to vSphere) seems to be holding actual adoption of Hyper-V as platform for VDI back …
Comment: MS just announced that full NAS support will be part of Hyper-V3 (Windows Server 8).
XenServer is considered when a single (Citrix) vendor stack is preferred (support, integration, skill set) and features like IntelliCache and GPU pass-through are hypervisor specific considerations customers take seriously (like RemoteFX for Hyper-V).

Broker of Choice: XenDesktop

There is a multitude of “connection broker” technologies in the market (Quest vWorkspace, Leostream, Virtual Bridges Verde etc.) with varying capabilities and scope (session management, deployment etc.). There were endless discussions and many points raised but when asking the clients “what’s the main reason for picking XenDesktop” the common points were:

-   multi-hypervisor support – “If we decide to switch hypervisors to XenDesktop or hyper-v we can keep XenDesktop in place.”

-   HDX protocol advantages - mainly bandwidth efficiency and high-end graphics (3D) capabilities over PCoIP and RDP (comment: see “Section for high-end graphics” and “changed view” for updates on VMware capabilities)

-   Citrix Portfolio – this is a little more difficult to quantify but it became clear from the discussions that many customers appreciate the breadth and heritage of Citrix’s end-user capability.

The “Flexcast” concept Citrix promotes and the technologies it provides to enable this “one does NOT fit all” approach resonates well with clients.

The fact that they can not only get VDI and streamed applications but also hosted shared desktopss (aka presentation server/terminal server/XenApp) from the same vendor alongside various network appliances for delivery optimization seems to play an important role alongside the fact that Citrix receiver promises connectivity from “any” device.
The overall value proposition of these capabilities made XenDesktop the product of choice.

OK, so what about Application Virtualization and User Profile Management …?

Application Virtualization: Microsoft App-V

Let’s not get into a definition of app virtualization (as there are different models) but here I am referring to the ability to sequence an application, deliver it to the end-point (e.g. stream it to a virtual desktop) and have it run in an encapsulated runtime environment “bubble” with the usual advantages of portability and isolation.

Many vendors provide solutions in that space (VMware with ThinApp, Citrix – XenApp, Symantec SWS and others).

From my experience app virtualization in “phase 1″ is often not fully explored. One of the biggest drivers for app virtualization is the goal to deliver stateless desktops but reality prevents in many cases the use of 100% stateless instances in larger environments, whether for technical reasons (e.g. typically not all apps can be virtualised) or organisational (user insisting on locally installed apps). However, some administrators I worked with even started to virtualize and stream apps to physical endpoints before looking to virtualize the desktop itself..

Technologically ThinApp has the advantage that it runs in “user-mode” which in simple terms means that the app is less likely to affect other processes in case of failures, it is also using an agent-less approach – generally favoured to avoid yet another update and certification point. But at the same time it is this agentless model which makes it less integrated into the central management environment.

Administrators were pretty consistently telling me that the level of integration and central management of application virtualization makes them pick one solution over another

Administrators were pretty consistently telling me that it is this integration and central management which makes them pick one solution over another (on a very practical level) and currently favour App-V over ThinApp even though some improvements have been made with basic integration in recent View releases.
While XenApp is clearly used in many accounts for Server Based Computing (aka terminal services) I have seen only relatively limited use for app streaming in multi-vendor “best of bread” environments.

Because of the emerging nature of app virtualization however the affinity to a product/vendor in accounts is here less strong than e.g. with the hypervisor or broker – often still more a vision than final choice in “phase 1”, while in “phase 2” it is typically recognized as a required enabler for (cost) efficient VDI environments using stateless images.

Ruben Spruijt from PQR has created a whitepaper Application Virtualization Smackdown covering this topic in great detail - definitely worth a read if you are exploring this area.

Profile Management: Windows Roaming Profiles, AppSense

Arguably the least defined and therefore potentially most complex aspect of desktop virtualization (not just for HVD) is the profile management.

Least defined because products have evolved from initially addressing the “user profile” issues to include an ever increasing scope of “user environment management” (UEM).

User Environment Management is undoubtedly the component virtualization architects who have grown from server virtualization into desktop virtualization typically struggle with most.

It’s also undoubtedly the component virtualization architects who have grown from server virtualization into desktop virtualization (like me) typically struggle with most. When I first worked with AppSense years ago I was dazzled by the capability but also repelled by the complexity of user management in general – something desktop admins have no issues with.

In simplified terms UEM goal is to achieve a consistent user experience across images and devices, reduce login times and address consistency issues with classic Windows user profiles. However, products like AppSense or RES provide capabilities far beyond that by including aspects of resource, security, license management and reporting capabilities.

Out of the last 4 larger engagements, 3 initially used Windows roaming user profiles and were at varying stages of piloting AppSense (one also considered RES). One client stated that they were not going to investigate/use another approach anytime soon (stayed with roaming profiles and folder redirection).

In my view there is no easy answer or recommendation without deeper analysis and a generalisation is dangerous, so let me be clear – while there is no doubt that you will need to address the issue of user environment “mobility”, it is a very individual technology decision which will require you to invest in piloting the approaches and determine what scope you want it to address.

OK, that was the software stack … so what about the Achilles’ heel of HVD … High-End Graphics …? What is really achievable with today’s technology?  See  Part 2 “High-End Graphics” …

Posted in Citrix, private cloud, VDI, VMware | 2 Comments »
Tagged: , , , , , , , , ,



vSphere 5 Announcement – The Aftermath – Or – Will Microsoft, Citrix and Red Hat be quietly smiling?

Posted on July 14, 2011

by +

#### Update 29/07/11 – VMware is making changes the initial vRAM licensing allocations for vSphere 5 ####

Summary of changes are:

  • VMware vSphere 5 Essentials will give a 24GB vRAM entitlement
  • VMware vSphere 5 Essentials Plus will give a 32GB vRAM entitlement
  • Max vRAM in Essentials / Essentials Plus will be maxed at 192GB vRAM
  • VMware vSphere 5 Standard vRAM entitlement has changed to 32GB (assumption)
  • VMware vSphere 5 Enterprise vRAM entitlement will be doubled to 64GB
  • VMware vSphere 5 Enterprise Plus vRAM entitlement will be doubled to 96GB

And - interesting for users targeting to utilize the recently announced (up to) 1TB vm capability –  The amount of vRAM that counts against the overall vRAM license pool will be capped to 96GB per VM. That means, even if you assign e.g. 1TB of RAM to a VM, it will only take away 96GB of RAM from your license entitlement!

Now, this is certainly looking much better!

Cynics might say that ‘the damage has been done’, the ‘sleeping dogs have been woken’. However, this change would certainly calm the nerves of  many folks who projected their memory growth over the next years and realised the long-term impact of the initial vRAM licensing structure.
#### Update End ####

If I had to sum it up … Great Products, Questionable Communication Strategy and Dangerous Timing.

Before going any further – I actually think it’s a shame that the licensing change and the subsequent online “riot” overshadowed the excellent new features announced yesterday and yes, I am painfully aware that by writing this article I am fostering that very pattern …
However (while having great and honest respect for VMware) I’ve also always been very open about my attitude towards customer choice and healthy competition in this space … so here it goes …

Dangerous timing

Let me put it simple … had VMware done this only 2 years ago I would have said “ok, that’s the luxury a market leader has, that’s the way business works”… .

There is no doubt that VMware still is the market leader BUT many customers have now “good-enough” alternatives…

It is not that customers didn’t want to look at alternatives 3 or 4 years ago, no, they actually did and said “it’s not good enough – it does not what we need it to do” …
… But 2011? This year it’s all coming to a head!

  • Microsoft closed the (initially massive) gap with the announcement of SCVMM 2012 – directly trying to match vCenter features and squaring up to vCloud Director. (yes, it’s only a beta at the moment)
  • Citrix just announced the beta of XenServer 6 and also the acquisition of cloud.com the very same day VMware announced vSphere 5 and vCloud Director 1.5, demonstrating how serious they are to compete in this space.
  • Red Hat has announced their CloudForms cloud platform and pushes KVM through alliances like OVA as open source alternative to VMWare. Big names like Intel, IBM and HP are in this too. A beta of RHEV 3 is expected soon and promising to be a very competitive release.
  • Gartner has just added Microsoft and Citrix to the leader quadrant of it’s Magic Quadrant for Server Virtualization 2012 – as sign of the time!

There is no doubt that VMware still IS the market leader BUT many customers will have “good-enough” alternatives by the end of this year.

The biggest mistake VMware can make is to erode it’s strongest pillar – todays small, medium and large Enterprise virtualization customer and sacrifice this for an (arguably strategic but unvalidated) cloud market …

The biggest mistake VMware can make is to erode it’s strongest pillar – today’s cost-concious small, medium and large Enterprise virtualization customer and sacrifice this for an (arguably strategic but unvalidated) cloud market.
Yes VMware has the most comprehensive and advanced feature set but customer will look for (and I use the term again) GOOD-ENOUGH alternatives.  Yes Storage DRS,  auto-deploy and 1TB vms are sophisticated technologies but are they a requirement for the majority of today’s virtualization customer, something they really can’t do without?
More importantly, will everyone look today at vCloud Director, SRM and vShield and accept the value they provide as today’s strategic “must-have” – that’ s a key question here!

“Hypervisor and vm management” – The importance of this control point has often been greatly underestimated and is in reality exactly what VMware’s stronghold is …

Exposing this pillar and opening up the door for Microsoft, Citrix and Red Hat to even just provide the “lower architectural layers” (e.g. hypervisor and base vm management) would be extremely dangerous!
The importance of this control point has often been greatly underestimated and is in reality exactly what VMware’s stronghold is. Today it’s often belittled as “commodity” in the shadow of “cloud” discussions – but without it VMware would not dominate the Enterprise virtualization market. And without its dominance in the Enterprise virtualization VMware have little leverage to enter the wider cloud market as it’s strategy is primarily based on expanding from virtual infrastructure to cloud through a service provider/hybrid approach.

So IMHO the timing for this change could not have been worse – will Microsoft& Co be scared by the latest technology leap or see this announcement as an opportunity … time will tell …

So what happened?

The Good

VMware announced their next generation “Cloud Infrastructure Suite” with vSphere 5, SRM 5, vCloud Director 1.5 and vShield 5 (click links for the respective – good – “what’s new” flyers).

The feature sets of vSphere 5 are impressive and VMware has in many respects extended it’s lead as visionary and executioner in the virtualization space …

The feature sets are impressive to say the least and VMware has in many respects extended it’s lead as visionary and executioner in the virtualization and (to some extend) cloud space. There is already a wealth of information on the new features available – so I have only attached a few “recommended links” at the bottom of this post.

The Bad (?)


New vSphere 5 Licensing

With that they also announced a change in licensing for vSphere 5 – adding a memory-based charging model to the existing “per CPU” license. In simplified terms,  memory usage will be “capped” to 24, 32 and 48GB per (old CPU socket) license, so exceeding these memory to CPU ratios means you have to buy additional licenses with vSphere 5.

How Bad?

In reality the implications of this will depend on your specific environment, mainly your current “memory to CPU ratio”, your consolidation ratio and your existing license levels.
To get a quick indication, calculate your current “memory” to CPU” ratio and see if this exceeds these capping levels listed above. (Bear in mind that it is the total configured amount for your vms, so if you are currently heavily over-subscribing memory then you might have an unpleasant surprise).

But it transpires that for a good chunk of actual customers reality will be less rosy. Read some of the comments at the bottom of Gabe’s post (good post – dubious title) to see some interesting real life examples.

If you are using Enterprise Plus only to run your 12 core AMD CPUs then you are in luck!

If you are attempting to run those “monster vms” announced with vSphere 5 – think twice!

Two extreme examples:

  • If you are using Enterprise Plus only to run your 12 core AMD CPUs then you are in luck (as long as your memory ratio is below the capping levels). Switch to any suitable lower edition and save the price difference.
  • On the other hand, if you are attempting to run one of those “monster vms” announced with vSphere 5 – think twice! A single 1TB virtual machine can cost you $76,890 in vSphere licensing fees (22 Enterprise Plus licenses).

Now you can question my sanity based on how many systems out there are actually using 1TB and how much that physical system would be – but then I’d have to question the announcement to support 1TB vms in general and more importantly the ill-considered timing of announcing these two together – as one basically negates the other!

Yes, change “always hurts” and an initial adverse reaction to change is not necessarily an indication that the change is wrong/bad. In reality most will sit somewhere in between those two use cases, more real-life analysis is needed to fully understand the impact and it is important not to fall into the trap of competitive exploration.

Communication

Yes, there was only some disclosure to analysts and selected customers (I had actually a client asking me months ago on my opinion about vRAM charging – long before I was disclosed).

We all know why VMware changed licensing – the fact that customers scaled their environments by buying CPUs from Intel/AMD – rather than buying VMware licenses …

But I believe much more could have been done to prepare the community and also that a lot of the uproar was caused by the somewhat patronising way VMware positioned the change as “simplification of licensing”. Now, the VMware user community is arguably comprised of the most switched-on individuals out there … we all understand why VMware is ultimately doing this. The fact that customers scaled their environments by buying new CPUs with more cores – rather than buying VMware licenses - must have driven VMware mad.

In initial discussion with a client I was asked how VMware can seriously position this as “simplification, fairness and flexibility” (words from the licensing guide).
“Simplification and ability to project costs? Now I don’t even know what my licensing will be based on, it could be memory or CPU …? Flexibility – configured vm memory – what happened to my flexibility of oversubscribing memory? Pooled licensing – well, to be honest we had that since the introduction of a license server.
…visionary licensing – well possibly but to be honest I’m not even using charge-back internally and “cloud” is literally a vison for me – not todays reality, today’s reality is keeping cost down … . ”

So yes, it is fairer to VMware (if there is such a thing in business) a “loop-hole” created by hardware evolution has been closed – if customers with existing subscriptions however feel the same remains to be seen …

Any hardware vendor with scale-up systems and memory extending capabilities will feel the pain – and will be open to offer alternatives.

Many OEM Vendors and Service Providers won’t be happy

Another important part has not been widely considered – the OEM hardware partners shifting a massive share of VMware’s licenses  …
Now – being with IBM – I can’t and won’t comment officially here at all but it will be clear to everyone that this licensing announcement has an impact on the hardware platform. Vendors have advocated the value of scale-up approaches in accounts for a long time. Any vendor with scale-up systems and memory extending capabilities will feel the pain. Any customer who has invested in such a strategy will go back to the drawing board and review – now I’m not saying that scale-up and memory extensions are “dead in the water” at all – I’m actually convinced that for many customers it will remain to be an efficient approach – but some of the value proposition has clearly taken a knock (that is of course … if you deploy with vSphere – … back to the point …).
It will also come as no surprise that those systems are typically the ones with the largest profit margins – so vendors will care!

An even more important segment will be the service provider community – which VMware will need on their side to make vCloud a success – I am curious to see what charging models VMware will come up with to keep them happy!

All that brings us to the most important point of all of this – where will this lead the VMware customer base – or … where will they be lead to …?

Recommended Reads:

The VirtualizationMatrix will be updated as soon as the vSphere 5 code is available for download

Posted in Hyper-V, private cloud, Red Hat, VMware, XenServer | 6 Comments »
Tagged: , , , , , , ,



Citrix announces XenServer 6 Beta – The ‘beginning of the end’ for XenCenter?

Posted on July 7, 2011

by +

Citrix announced the public beta for their next version of XenServer 6.0.

There are various changes and new features which you can review HERE but I have listed some of the relevant ones below. Some of the key changes are around Citrix’s cloud capabilities (as expected).

What strikes me is the overwhelming ‘generosity’ of promoting another vendor’s management platform as part of their own announcement. So all of us aware of the Microsoft’s SCVMM 2012 beta of course knew that it can manage XenServer … but why is Citrix so ‘keen’ to put it in as a ‘feature’ in their own announcement? Is this not the management control point one would want to keep … ?

Now don’t get me wrong, as ‘consumer’ I absolutely welcome open choice and I’m under no illusion that SCVMM 2012 will not be a full replacement for all native XenCenter functions. But that’s the point, using SCVMM will promote the “microsoft way” of doing virtualization management. SCVMM 2012 promises to be a major improvement over the previous version and shapes up to be a credible alternative to vCenter and vCloud Director management and – compared to XenCenter – will potentially offer  superior functionality in many respects.

So yes, we completely understand why Citrix and Microsoft promote respective management capabilities of VMware vSphere (hypervisor) – so down the solution stack  -  … entice folks to shift eventually onto their management platform.
But I quite frankly fail to see why I would do the opposite and promote someone else’s management stack … unless I am prepared to indeed adopt it as the product of choice. Let’s be honest, if you go down that route (and you can argue that this is already EXTREME speculation – but a valid question) then where does that leave the relevance of XenServer as hypervisor in such a stack…?

Given the recent upheaval of  Crosby and Pratt leaving Citrix to start Bromium this is not necessarily the message Citrix wants to convey to customers right now (as indirectly as it is). The announcement to discontinue Citrix Essentials for Hyper-V could have been interpreted as effort to focus on its own XenServer management capabilities – or was it just the recognition that Microsoft is winning this battle and a strategic alignment longer term (avoiding duplication of efforts) …?

The XenServer product line has seen impressive improvements over the last couple of years and I actually personally consider it today in many respects to be ‘Nr.2′ in the industry only behind vSphere from a hypervisor and management functionality point of view.
So maybe this is nothing but a ‘marketing oversight’, but the overall ecosystem challenges and the carefully crafted but sometimes ‘suffocating’ Microsoft relationship clearly has the potential to be an inhibitor as much as it is a strategic accelerator for Citrix overall.

Andy

XenServer 6 – New Features

Cloud:

  • Self Service Manager – This new feature (superseding Lab Manager) enables you to build self-service environments for “private clouds.”  Self-Service Manager includes support for both XenServer as well as VMware vSphere
    -  simple virtual appliance and web-based UI.
    - offers multi-tenant support for creating VM “service catalogs”
  • vApps- Virtual Appliance support.  Within XenCenter 6.0 you can create multi-VM virtual appliances (vApps), with relationships between the VMs for use with the boot sequence during Site Recovery.  vApps can be easily imported and exported using the Open Virtualization Format (OVF) standard

Microsoft System Center integration:

  • Starting with the XS 6.0 release, you will have the option of managing XenServer hosts and VMs with System Center Virtual Machine Manager (VMM) 2012.  For more information refer to the Microsoft System Center 2012 beta page
    - System Center integration is available with a special supplemental pack from Citrix (installed on XenServer hosts), which is targeted for general availability when System Center 2012 ships later this year.

XenDesktop:

  • GPU pass-through (formerly experimental support) – With the 6.0 release, a physical GPU can be assigned to a VM so the applications running in the guest can leverage GPU instructions (“GPU pass-thru”).  This provides significant TCO benefits for the XenDesktop HDX 3D Pro technology used for delivery of CAD and other graphical applications via virtual desktops

Other (virtualizationmatrix-related) updates

  • Host RAM support has been increased to 1 TB
  • VM vCPU and vRAM support is increased e.g. up to 16 vCPUs and 128 GB RAM for Windows; increased Linux vCPU and vRAM support levels vary by distro

Posted in Blog, Matrix Updates, XenServer | 11 Comments »
Tagged: , , , ,



VMware or Red Hat to conquer the public cloud? – or – The balancing act of being “open”

Posted on June 13, 2011

by +

Having just returned from the European Red Hat Partner summit and the VMware vForum where I presented, it took me a while to digest the “openness” of it all …
The key messages conveyed in both events were (un?)surprisingly similar, considering that we have a major opens source software company on one side and a more traditional “business” model on the other.

Being proprietary rocks…
Let’s be straight – one could argue that in an ideal world (for selfish, money-making businesses without ethics) there would be no open source, being proprietary rocks!
After all making money by attracting and “retaining” clients (I’m deliberately not saying “locking them in”) is ultimately the goal of every business – and that actually applies to VMware in the same way as Redhat (let’s not mix up the ‘open source community’ with Redhat as a BUSINESS) …
Now that would obviously completely ignore the power and dynamics of an open technical community but more importantly that’s not in the interest of the consumer…
Public cloud promises to empower the consumer – so they will increasingly be looking for choice … no capital dependency, outsourced, pay per use service operation models enable you (in theory!) to switch providers like I just switched my energy and gas supplier to EDF last week – go to a comparison site, find the best deal and “click” … done (obviously not reality today with cloud). 

Public cloud can only exist on open source … or …?
What both events made crystal clear is that increasingly many “traditional” businesses will be forced to have a foot in both camps in order to balance customer demand for open choice with a business model allowing them to make money and retain customer “affinity” (or would we otherwise have  URLs like this …?http://www.microsoft.com/opensource/;-P

There was a bold statement by a speaker at the Red Hat summit:  “Public cloud can only live on open source!”
I was initially inclined to agree but then thought this through again and adjusted it mentally to what I believe to be more appropriate: “public clouds need to live on INTEROPERABLE source”…
Open source should of course help to facilitate this but if I just end up with a bunch of non-intuitive, non-integrated code, with undocumented APIs and outlandish image formats then the fact that its open source doesn’t help me at all.
So I am not saying that I don’t believe in open source, quite the opposite, all I’m saying is that the “open source” stamp on its own is not good enough and as a consumer of resources (not a developer) I would absolutely consider a proprietary solution as long as it is intuitive, cheap, with well-documented APIs and – that is the key – inter-operable with other public providers.

The Public cloud is only as good as the “connectors” to it  - Key Battle 1: Hybrid Connectors
VMware very much owns the majority of today’s virtual enterprise footprint and with that has a critical control point in the private cloud. The Public cloud is a completely different story with over 80% being OSS based and VMware yet hardly to be seen.
Especially for VMware it must be of utmost importance to provide a ‘best of breed’ connector between existing vSphere infrastructures and public vCloud Director resources before others provide this linkage to other (non-VMware) public platforms. So I expect a lot of focus on vCloud Connector functionality from VMware (in the same way as on ‘Concero’ from Microsoft).
For me VMware’s ‘must win’ play is to entice Service Providers to take advantage of the existing vSphere footprint “Hey look, all your customers already have VMware, the only thing you need to do is to provide public vCloud Director resources for them to burst out to – we provide the connector, it’s as simple as this!”

Red Hat CloudForms Architecture

Of course it’s not that simple but there is a chance to develop a public VMware cloud ecosystem around vCD in this way – again before someone else offers seamless alternatives (more than just Amazon’s VM Import). But will it be enough to only provide linkage to public VMWare vCD resources? IMHO absolutely not. I am very curious to see how much VMware will enable connectivity to other public provider platforms going forward … Again, it will be a fine balancing act but I’m convinced that it won’t be successful otherwise.

Red Hat has indicated that CloudForms (their IaaS platform) can indeed manage VMware though their DeltaCloud driver and – while currently positioning CloudForms for private and hybrid – their vision (of course) is for DeltaCloud to be the top-level public layer linking into private (or public) VMware clouds …

Who will be the next Microsoft? Key Battle2: PaaS
Now – here’s the real battle for Cloud control (or better ‘ecosystem control’) … Who will provide the application platform for the volumes of future cloud-based applications? Who will control the ecosystem of future application suites? Who will be the next Microsoft? Again, it’s all about control points and the pain of moving. If switching public cloud providers could really be as easy as switching utility providers, switching your application platform (as ISV) is rather like moving house! Being open is a great value proposition here and it’s not just the OSS providers who have realised this …

Red Hat recently announced their hosted “OpenShift” PaaS platform which essentially allows developing and running Java, Ruby, PHP and Python applications and comes in 3 different editions. From 1) “Express” (free) which provides a runtime environment for simple Ruby, PHP and Python apps over 2) “Flex” for multi-tiered Java and PHP apps with more options (like mySQL DBs and JBoss middleware) to full control with the “Power” edition supporting “any application or programming language that can compile on RHEL 4, 5, or 6″ and enables to deploy apps directly on EC2 and (in the near future) to IBM’s SmartCloud.

VMware had before announced their own open (yes open) “Cloud Foundry” PaaS project, it has incarnations as fully hosted service (currently in beta), as open source project (CloudFoundry.org) or a free single PaaS instance for local development use. 
This is quite an ingenious move IMHO which could not only foster the adoption of this critical layer for VMware (away from e.g. MS Azure, Google’s App Engine or Amazons’ Elastic Beanstalk) but will also be helpful in countering competitive ‘lock-in concern’-campaigns.

Professional Suicide
Yes, I honestly believe that KVM has a good chance to become hypervisor of choice for public cloud. However … that is unlikely to be the control point… . So which management platform(s) will take that all important crown …? I don’t want to hazard a guess, there are many …and that is part of the problem, the open source “communities” would need to become more of a COMMUNITY if they want to succeed. ESX could not be beaten with 7 or 8 different (but weak) flavours of Xen and that was just a single OSS project splintered by commercial offerings … in the same way the sea of OSS based cloud controllers with eucalyptus, openstack, cloudstack, deltacloud, opennebula will face  focussed (more proprietary) heavy-weights like Microsoft, Google and Amazon – with different commercial models and lots of money to throw at this.
The increasing number of OSS management solutions and “open bodies” will also make VMware less nervous than intended as long as they indirectly compete with each other …

In the current mess of FUD and openness, a somewhat splintered OSS ecosystem versus the lure of “best of breed” with vendor lock-in attached, the key challenge for the poor consumer and IT provider will be to figure out the developing control points and the dependencies they buy into … e.g. is a multi-hypervisor management layer the way “out” of vendor lock-in or the lock-in to the other vendor’s higher layer management solution…? ;-)

BUT (and it’s a big “but”) I would argue that anyone not strategically looking at these open solutions is at best ignorant or – e.g. if you are a service provider yourself – more likely long-term professionally suicidal … yes, in an ideal world everyone wants ‘today’s best of breed’ but more critically you have to maintain your negotiation potential through the ability to switch and if only for that reason alone you need to keep your options open!
And while I spoke to many completely different clients at both events, that was a common concern raised by most of them.

Industry endorsement like the recent OVA announcement will help to give Red Hat and KVM the needed credibility and weight … let’s see how well the vision will be translated into tangible solutions.

 Andy

- CloudForms (IaaS) is in beta with availability planned for fall 2011
- OpenShift ‘Express’ and ‘Flex’ are available as beta and ‘Power’ will be available later this year

PS Before you dismiss this as the ramblings of a Linux geek in sandals and bum bag - I am not and I don’t own any … – but yes, I do ramble on occasionally … ;-)

Posted in Blog, KVM, public cloud, Red Hat, vCloud Director, VMware | 8 Comments »
Tagged: , , , , , , , ,