The agility and scalability of virtual desktops enable use cases that are not possible with a physical desktop environment. However, introducing a virtual desktop infrastructure is complex. Time-scales can be long, resource requirements high.
In an effort to relieve the discomfort for customers and partners VMware have introduced a Rapid Desktop Program. This program looks to validate View Proof of Concept appliances to ensure that they meet criteria for performance and reliability. By removing the complexity of the “I”, an organisation can focus better on the assessment of virtual desktops and in turn deliver faster .
Pivotv3 are the first to release an appliance that has been validated by the Rapid Desktop Program. How does Pivot3’svSTAC VDI allow you to overcome common issues with VDI projects? Is this likely to improve take-up of VDI? And, this is an appliance, such devices are normally associated with big enterprise solutions – is this only a big enterprise solution?
Why do VDI Projects Stall?
Yet, desktop virtualisation is not the same server virtualisation. Still, the process of assessing and planning the move from ‘traditional’ desktop environment to a virtualised desktop environment is similar to that of server virtualisation. Very simply, you…
- Determine what it is you need
- Measure what it is that you have got
- Assess how that would be delivered in a virtual environment
- Implement and test a virtual environment
- Roll-out
- Monitor, re-measure and re-assess as required.
Simples?
Rarely.
There are a number of reasons that this method impacts VDI projects, two stand out:
Infrastructure Complexity – the nature of desktop use is complex and varied. You rarely ‘reduce’ infrastructure with VDI: more likely you increase it. Fundamentally, this is often because the core desktop OS being virtualized was not been designed to run in a virtualized environment. VDI projects typically require the introduction of SAN infrastructure, additional networking, as well as virtualisation software. This leads to a very high capital expenditure cost.
Support Multiplicity Given this additional hardware and virtualisation software, support costs can increase in order to support the SAN and changes in workspace and application delivery models. In addition, any outages in the back-end VDI infrastructure can impact a great number of people. If there are savings from operational expenditure, these can take time to come through, but are often difficult to realise, especially with smaller organisation sizes.
A third reason typically overlooked is strategic vision. Having a strategic level view for VDI is vital necessary for a successful deployment. Why? Because in understanding how virtualised desktops will fit and deliver to the business strategy shapes what is required and how those virtual desktops should be delivered, it answers the question of what part they will play in the overall enterprise desktop strategy/IT service delivery plans.
These issues are not unsolvable: but they are not always properly addressed. Not addressing these issues early in the project cycle is a fundamental problem for VDI projects.
Pivot3 VDI Value Proposition
Pivot3 have designed vSTAC VDI to reduce infrastructure complexity by delivering an optimised scale-out SAN; host pre-configured virtual servers that are required to deliver a VDI environment; and enabled hypervisor options for high availability. Pivot3 claim that with their appliance, VDI service can be configured in 30 minutes, with around an hour’s worth of training.
This of course, allows for fast delivery – but a VDI project is for life, not just for Christmas. Pivot 3’s vSTAC design is intended to provide linear elastic scaling while reducing storage tuning. Pivot3’s goal is not just to simplify storage, but eliminate it as a VDI obstacle – giving you a lower cost of entry for VDI but also a predictable option for dynamically scaling out the service in the future.
There are alternative integrated infrastructure solutions available, an important factor for the vSTAC VDI is that each device be cost effective, self-contained and have a validated environment. In collaboration with VMware, the device configuration has been validated to support over 100 virtual desktops based on VMware’s own Viewplanner measurement tool. An appliance like vSTAC VDI has not been available before.
P3 – Pivot3 Production Pilot
Pivot3 have put vSTAC VDI at the heart of their P3 rapid desktop production pilot program for VMware View. Each vSTAC VDI appliance will be tuned to deliver a high-performance VDI experience combining 20 threads of compute power, 96G of RAM, 3TBs of storage, along with a 200G Solid State storage tier.
The vSTAC P3 program includes:
- Two vSTAC VDI appliance SKUs:
- Pivot3 vSTAC VDI “P3 starter appliance”: A self contained ready-to-deploy starter appliance for 50 to 100 desktop production pilots which includes pre-configured trial licenses of VMware View, vCenter Server, Microsoft Windows 2008 Server
- Pivot3 vSTAC VDI “standard appliance”: The ability to dynamically and automatically scale out a VDI environment by adding additional appliances each with a capacity of more than 100 desktops;
- 30-day P3 pilot agreement;
- One-day onsite assistance; and,
- Phone support during 30-day period.
- Prices starting at $35,000
In addition, the P3 program will provide IT professionals, prospective customers and channels with the ability to:
- Test drive Pivot3’s VDI solution in a live hosted environment from Pivot3’s website for up to 300 desktops, and…
- Configure their own VDI environment, complete with ROI, using the VDI configurator tool also on Pivot3’s website.
What is interesting here is that the program includes a “complete environment”. The starter appliance has all Microsoft components preconfigured, one of the few requirements for a PoC is to for you to create the first image. It could be argued that similar services already exist: Citrix made a number of changes way back in XenDesktop 5 to improve the PoC>delivery development cycle and now have VDI-in-a-box with the purchase of Kaviza to better target SME customers. But, both are software solutions: vSTAC includes the hardware, the associated Windows servers and the hypervisor optimisations. Quest are developing Project Liberty which suggests a similar offering – but on Microsoft Hyper-v: but that project is still in development.
VMware have sought, with the Rapid Delivery Program to deliver VDI appliances to ease delivery, and Pivot3 have been the first to deliver.
To be fair, “Get Storage wrong” and VDI doesn’t work. Storage is a factor in VDI stall, but not the only factor.
The P3 program is geared towards provisioning quickly. The vSTAC appliance will be pre-opitimised and configured. Initially it will link with p3 website to provide sizing notes, but it isn’t beyond the reason to include those services as part of the PoC. Having an appliance model will allow for more predictable scaling and reduce the learning curve for the desktop team – who can focus on the desktops rather than hypervisors and SANs. In turn this the allows reduction in the need to manage multiple environments improving support and reliability. All good.
However, virtual desktop delivery is not just about workspace image provision. “In-a-box” solutions can skip the complexities of Microsoft file and directory service integration; profile management and application and data migration. While the appliance has Active Directory, and storage capacity – will that be able to accomodate user personalisation and file storage in a new centralised service? There is also a question of whether an appliance model can fully manage the pain point of scalability.
The cost of the vSTAC appliance, when it starts shipping in November will start at $350 per desktop load. Each appliance supports @100 users – $350/desktop which approaches the “from a $1/day” virtual desktop instance quoted by DaaS providers such as Desktone. Yet each appliance increments your user count in blocks of 100: a move from 100 to 120 users may appear a daunting cost. This is where DaaS providers can offer a more dynamic service.
Yet, perhaps the most obvious component missing is the wider strategy vision. For example there is no implication or measurement of how applications your organisation uses will work and be licensed in a vSTAC world. Virtualization stall can be more about the business and organizational impact the technology exerts than about the technology itself. In vSTAC Pivot3 have great appliance – but there is still scope for a PoC to needs tools from vendors like Centrix software/Liquidware Labs to help assess application usage; and/or AppDNA/ChangeBase to evaluate how those application will be transferred to a different OS or virtualised.
After reading the Pivot3 SPECs on other sites, the Pivot3platform doesn’t appear to do anything special. It is simply introducing limitation to VMware that make VMware so great. Since Pivot3 storage OS is embedded on thier host, it seems that you wouldn’t be able to update your host, fail a host, or even reboot a troubled host without degraded storage. I think HPs solution, even though it isn’t perfect, still allows you to do these things without the risk. Pivot3 design of Binding ESXi to a host, introduces problems and promotes the loss of data. It will be intesting when Pivot3 gets the support call that someone just lost their Exchange server.
Thanks for taking the time to respond.
I’m unsure I follow tho’. Here I’m discussing what I view as benefits of Pivot3’s vSTAC VDI appliance. An appliance focused, tested and configured to offer the hardware to build a VDI environment on. While I’m confident that it’ll do what it says on the tin, I’d query the process that is required to migrate an existing environment from “what it is you have now” to “an P3’d VDI service” –
Yet, you talk about Exchange and HP services. Do HP have a similar VDI appliance? Do HP have a similar VDI appliance that also includes Exchange? Or, are you trying to apply different use cases to this appliance? Are you looking to compare against P3’s CloudStack or Databank technologies?
Andrew –
Your article is full of good content about Pivot3. However, what is not covered is the underlined risk that Pivot3 adds to the table. Pivot3 appears to be using a SATADOM to put ESXi on. Furthermore, this appears to be where they are hiding their storage software. If two of these devices die, in an array, you just lost ALL of your VMs with no type of recovery. Do you want to risk your data on a device that has a failure rate higher than a consumer flash drive?
In regards to DR, Pivot3 does not appear offer data replication, so they would only be able to reside on VMware hostbased replication for backup. Furthermore, they would require a complete array outage for any type of updates. Last but not least, according to their current document, they have no upgrade paths. So, if you upgrade thier RAIGE (vSTAC OS), you would take a complete data loss on your array as they are not cross compatible.
The point here is that your article is favoring a product without looking at all of the pros and cons. Since this is ALOT Of data that could potentially be lost because of Pivot3 lack of development, it would only make sense to look at the potential issues when posting such a favorable article.
S3cur31, I follow what you’re saying but I’m struggling to follow the reasoning.
What P3 offer is a VDI *appliance*. In such instances I’m less likely fussed on data replication and I’m not at all bothered about doing an upgrade of vmware. I’ve bought an appliance – I expect it to do that. I don’t need/want to know what’s going on under the hood. That’s why I bought an appliance.
The risk of losing your VMs? These are desktop instances I’d expect those VMs to be data-free/lite. If you’re hosting a lot of data within the VMs I’d be questioning the strategic vision rather than the device. But then that’s a vision question: worthy of an article in its own right.
It is a fair point that if you move to VDI, and VDI in-a-box/appliance are no exception, to highlight if you’ve only one box the impact of that box’s failure is high. But that isn’t limited to P3’s solution. Citrix’s VDI-in-a-box solution suffered from a code malfunction that caused users some issues. So from a ‘pros/cons’ issue I see that as a generic issue: again, worthy of its own article quite likely.
I’d ask again – you originally talked about Exchange and HP services. Do HP have a similar VDI appliance? Do HP have a similar VDI appliance that also includes Exchange? Or, are you trying to apply different use cases to this appliance? Are you looking to compare against P3′s CloudStack or Databank technologies?
The issue is the architecture. Lets get back to the basics, would you trust your CEOs desktop to be run on a flash drive? We all know that the failure rate is high. So, lets take it a bit higher, would you trust your companies desktops to be ran on a flash drive? Using Pivot3’s solution, this is what your doing. It appears they have loaded their storage OS and VDI components on a SataDOM per box. This means that if a satadom fails, so does your storage or esxi host. We all plan for VMware’s High Avaliability(HA) to recover the VMs right? The last time I checked, HA doesn’t recover storage node failures. So if you are looking at a simple single box solution or looking to grow them out, I would kindly recommend that you look for an alternative at the moment. Even VMware is trying to fix this age old issue with storage on the same box as the host. The fact of the matter is that it isn’t ready to production if the data that you have on it is critical to your business.
The correct way to do this is in the disk and not cheap flash drives…
I just came acrose this discussion and would like to help clear up any confusion. I am the CTO at Pivot3 so you can consider me informed but maybe just a little biased 😉
S3cur31 is correct that ESXi and the Pivot3 vSTAC OS are stored on a SATADOM. He is incorrect in characterizing them as cheap flash drives. The SSD in this product are SLC flash (not the cheap MLC) and have an MTBF of over 2 million hours.
His second point that HA does not recover storage node failures is also correct but immaterial. vSTAC OS does recover from storage node failures. vSTAC OS presents the storage from all appliances to ESXi as a single virtualized iSCSI target. If an appliance failes for any reason (including S3cur31’s woried about SATADOM failure), vSTAC OS will preserve all of the storage and HA will recover the desktops on other appliances.
S3cur31 was also worried about two SATADOMs failing and loosing all of the data. This is theoretically true but very low probability. Any system that can survive a single failure (but not two) like RAID 5 have a failure probability that can be quite easily calculated. It turns out that the chance that a second SATADOM fails while the first one is being repaired is around 0.001%. Normal RAID 5 systems based on rotating disk drives have a much higher failure probability. Our calculations and field experiance show that drive failures are much more common than SATADOM failures. Because of this we offer RAID levels that protect five simultaneous drive failures.
I hope this helps clear up any confusion