So, like many IT organizations you may have already made the decision to deploy virtual desktops – you’re ready to move from a small pilot to full production. But a lot of questions (and possibly some guesswork) stand in the way – what does the end state architecture need to look like? How do you get there? How are you going to make sure that you can move quickly and seamlessly from proof of concept to scalable production? Accounting for sufficient server capacity, network bandwidth and performance, storage IOPS, and especially quality of experience at the end-user level -- there are a lot of factors to contend with. And how do you predict user behavior in a production environment, including the load they’ll collectively place on your infrastructure when they log into their brand-new virtual desktops on Monday morning? Read More »
Taking the Guesswork out of Deploying Virtual Desktops: Cisco + VMware Validated Designs for View 4.5
Today Cisco announced the Virtualization Experience Infrastructure (VXI). I wanted to take the opportunity to discuss this new system, and offer some thoughts on what it means, relative to desktop virtualization.
Cisco announced a new system of technologies today designed to help enterprises introduce desktop virtualization to their organizations. Mark Boslet’s story for News@Cisco provides a good primer on the phenomenon if you need it, but in a nutshell, virtual desktops shift computing applications and business information off PCs and into a data center. Users than access those applications and information through a secure Internet connection (via a private or public Cloud), in much the same way as we browse the Internet today.
Desktop virtualization isn’t a new phenomenon (if you’ve been in the workforce for a couple of decades, you might remember the green screen and mainframe varietal!) but it’s becoming more and more compelling as good broadband connectivity becomes pervasive and customers tap server virtualization technologies and wide area network acceleration to make applications perform better over WANs.
This past week, Thomas Scheibe (Director, Data Center Architecture) had the opportunity to co-present with VMware and NetApp at TechFieldDay on a broad range of Data Center topics.
Thomas is one of the leaders in our Solutions and Strategy Unit (SASU) that is responsible for creating Cisco Validated Designs (CVD). One of the topics discussed was the recent CVD on Enhanced Secure Multi-Tenancy and Thomas asked, “How many of you are familiar with the depth of technical content in a CVD?”
I’m somewhat disappointed to say that the show of hands was less than unanimous. Now this shouldn’t have come as a complete surprise to us, because in the past CVDs were primarily targeted at Network Administrators and the TechFieldDay audience is traditionally more focused on Servers, Storage and Applications. But considering that many of our Data Center solutions are no longer just focused on networking elements, we look at this as an opportunity to create awareness for the Architect and Administrator communities. We also look at it as an opportunity to solicit your input and feedback on how we can better deliver content that will help you design and deploy Data Center solutions that are becoming more complicated as convergence, virtualization, and automation blur the lines between IT organizations. Read More »
As the solutions marketing manager for Cisco’s Desktop Virtualization solutions, I want to use this opportunity to start a dialog around the trends we’re seeing across various IT organizations, and their efforts to embrace desktop virtualization. I thought we might start with posing a common question: As you’re designing your data center infrastructure to handle virtual desktop workloads, Is desktop virtualization really just another workload? Do you build a single consolidated, shared infrastructure to accommodate the usual server workloads alongside VM-hosted desktops, or do you handle these somehow differently? A VM’s a VM, regardless of what’s sitting on it, right? Consolidated… shared… elastic… this is the cloud infrastructure vision isn’t it? Whether you’re embracing VDI or App Virtualization, why should desktops be different? Maybe we should start with probably the biggest challenge and exposure associated with moving to virtual desktops: The End User
Quality of user experience and application responsiveness as impacted by a sub-optimal infrastructure still tend to be among the biggest impediments for virtual desktop implementations moving from proof-of-concept to production (that and the sometimes elusive path to expected ROI/TCO, which we’ll get into in another post). These are often the result of insufficient testing to replicate end-state loads on network, computing and storage. The results of a small pilot quite often don’t accurately predict what really happens when you multiply “The End User” by ten-fold.
So often with these projects, somewhere along the way, the combination of disappointing user experience, maybe compounded by unrealistic expectations results in the solution never getting off the ground. I’d like to say that there’s one solution that never fails. But let’s be honest – there are so many variables when you consider the infrastructure (compute, network, storage) as well as use cases across the constituents in your workforce, that there’s likely no single prescriptive approach to ensuring success Day 1.
This much we can agree on: your chances for success significantly improve when you commit to the right Day 1 infrastructure approach, tailored to delivering the best user experience possible, vs. hoping your current infrastructure is agile and elastic enough to accommodate the 300 users who don’t know what they’re about to step into Monday morning when they log-on to their new virtual desktop.
Here are a few questions to consider when trying to “build-it-right” on Day 1 :
- What would happen if you were to mix desktop workloads directly with enterprise application workloads? Isn’t the approach to updating, patching and securing desktops very different from the approach taken with business critical applications in the data center? It’s not hard to imagine A/V scans on desktop workloads impacting the performance of applications workloads residing among the same compute resource pool.
- What’s the profile of the compute and storage infrastructure? It’s well known that desktop virtualization can place a significant burden on memory and I/O before it does on CPU, except in the case of graphics-intensive apps. Therefore it makes sense from both an economics and “user-experience” perspective to ensure that the memory / CPU / I/O ratio is well suited to hosting virtual desktops. Likewise with storage -- virtual desktop IOPS can be extremely high, especially during boot and logon storms… this can account for larger than necessary storage costs… so doesn’t it make sense to ensure that the compute and storage infrastructure are designed and configured around the unique requirements of desktop workloads?
- What about security? The advent of virtual desktops gives IT a unique opportunity to dynamically create virtual workgroups that have access rights to certain resources and not others. This could possibly be achieved even when mixing desktops and application workloads, but how much more difficult would that be to manage and maintain, especially at the outset of moving your virtual desktops into full production?
Let me know what you think … and look out for some exciting news from Cisco on desktop virtualization by registering for our Collaboration Experience Launch on November 16th. In my next post we’ll provide more color to the argument here… and would love to include your thoughts.