Radom Streams of Consciousness during “USE IT OR LOSE IT” vacation: Remote PC is ^&%$!* AWESOME and why isn’t everybody talking about it?
So I am sitting at in the coffee shop at my hotel while I am on the Oregon coast using some of my “use it or lose it” vacation before the year ends. Last night, I literally had the greatest pint of beer in my entire life (India Pelican Ale from the Pelican Bar and Grill in Pacific City Oregon) and have just noticed that the same bar I drank it at is opening for Breakfast at 8 AM. I am wondering if I have a drinking problem because I have now rationalized that it is 11AM (early lunch) on the East coast and I could easily blame drinking this early on Jetlag. Like a lot of IT workaholics, I am really trying to get better at this whole “vacation” thing. At any rate, I thought I would sit down and read Jarian Gipson’s post on Remote PC and try to “get hip” to it as I myself am pretty excited about it.
For the record, I am NOT a VDI guy, that in and of itself is no longer badge of shame and it has been nice to see the virtualization community become more tolerant of those who are not jumping for joy over VDI. That said, I think VDI is totally cool but it is very hard to justify paying more for desktop delivery and trying to sell OPEX savings to CIO’s who worry about the next Quarterly stock call. Selling OPEX in the world of publicly held companies is a tough row to hoe. Then, in May, I read Jarian Gibson’s Blog about Remote PC to which I immediately asked “Can I haz it?”
Now I am excited, this works better than traditional VDI for SO MANY reasons, let’s take 1000 Flex/Teleworkers.
The 1000 Telworker Scenario:
Say you have to set up a telework solution for 1000 remote users. Typically this involves the procurement of 1000 laptops and sending the users home with them then building out the back end infrastructure to support either XenAPP or XenDesktop.
Sending users home with a Laptop and providing VDI Access:
So I am doing some brief estimating but I am assuming a laptop costs around $1000 and supplying 1000 end users with one puts in into the project a cool 1 million dollars out of the gate.
Project Cost so far: $1,000,000
Supporting the back end infrastructure:
A quick rough estimate of VDI memory/core requirements I would say you would need at least 20 servers to accommodate 1000 XenDesktop users. At around $10K per server you are looking at another $200,000 in back end hardware costs (not to mention licensing)
Project Cost so far: $1,200,000
So, in addition to the licensing debacle that you get to go through (one I have since thrown my hands up in disgust over) with Microsoft and the set up of the infrastructure you are 1.2 million into this deployment. We could switch to XenAPP (Now we’re talkin!) to save a little more. If you use VMWare (I don’t want to get involved in the hypervisor holy war) than you are going to have more cost as well.
So with XenAPP, I think you should be able to get by with 9 servers (30 users per 8GB VM w/2 vCPUs). At 9 servers you are looking at $90,000 and you are looking at round $1.09 million for your project. Nice savings but you are still stuck with building out the back end infrastructure.
Remote PC Scenario:
With the remote PC Scenario, we get a chance to actually take advantage of BYOD (bear with me here) and take advantage of the cheap PC models that are out there. We can replace the initial investment from a $1000 Laptop to a $400-$600 desktop (bulk PC purchases get this kind of leverage). This presents an opportunity to reduce that cost from $1 million to $400K-$600K. (Let’s use $500K as a baseline)
Now you’re talking the “language of love” to your CIO, CAPEX. Not only have you reduced the initial procurement costs but you do not need to build out the same amount of back end infrastructure. In the Remote PC scenario, you have your DDC’s brokering the connections but the XenAPP/XenDesktop farms are completely gone or maybe one or two XenAPP Servers for apps like ArcGIS, SAS and CAD.
I have spent hours extrapolating CPU Cores and RAM to try and come up with a user density, in all likelihood you have several thousand cores and several terabytes of RAM sitting at desks and in cubicles that can now be tapped into for remote access using Remote PC.
Why this would work?
While working to set up the teleworking solution at a previous employer we noted a few things. First, after making a seven figure investment in laptops, we found that only 20% of them (that’s a generous number) actually connected to us remotely. The remaining users insisted on using their own equipment. Let’s take my case for example, at my desk at home, I have the “Crackpot Command Center” going with four monitors and a ball busting six core 16GB system (As any true geek would). So, when I want to connect to work, am I supposed to unplug everything and connect my keyboard, mouse and ONE MONTIOR (seriously?) to my Laptop? Maybe two monitors if I have a docking station? No freakin’ way!
Even non-geeks have their setup at home already and I doubt they have a data-switch box to switch back and forth so a teleworker can either work from the kitchen table OR they can UNPLUG their monitor and plug it into the docking station or laptop? The fact is, this is just not likely and the same end user would likely prefer to just use their equipment. This is something I witnessed first-hand to the complete shock of management.
In addition to the BYOD paradigm or UYOD (Use your own device) paradigm you also maintain support continuity. The first time we discussed VDI with our server group my team looked at me like I was crazy. First off, desktop management is 20 years in the making and there are old, established methods of supporting it. A complete forklift of the status quo is much more difficult than just provisioning desktops on the fly.
One of the issues with VDI was the inability to get people to understand that your 5 person Citrix team cannot support 10,000 desktops. Even more, they did not put 5 to 10 years into their IT careers to go back to supporting desktops. I personally am not overly excited to deal with Desktops after 16+ years in this industry and neither are most of the server admins I work with. The inability to integrate XenDesktop/View/ VDI in general, with the incumbent support apparatus at an organization is a significant, and in my opinion, often overlooked barrier to adoption. Your Citrix team likely is not THAT excited about doing it and the desktop team is completely intimidated by it. We go from imaging systems to the “Corporate Image” to setting up PVS, configuring PXE, DHCP Scopes, DHCP failover logging into the hypervisor….etc. Folks, it’s a Dekstop, the level of complexity to do a large scale deployment is far more advanced and much less forgiving than imaging laptops as they come in the door. Advances in SCCM integration for XenDesktop were very welcome and a timely feature but ultimately Remote PC delivers continuity of support as it is little more than an agent installed on the existing PC. The same people who support the PC today can continue to do so, server admins are not being asked to become Desktop Admins and the only thing that changes is that you are extending your infrastructure into the cloud by integrating the DDC and the Access Gateway allowing users the same consistent experience regardless of where they are working from.
You know what, I could BE a VDI guy if:
• I don’t have to put Windows 7 images on my million-dollar SAN (I LOVE Atlantis but it is not safe to assume your Citrix team can put the squeeze on your storage team)
• I don’t have to strong arm Server Admins to do Desktop Support
• I don’t have to buy a 2nd Windows License (or deal with Licensing)
• It can be made consistent enough that the incumbent Desktop team can support it
Holy crap! I’m out of excuses…I think I could become a VDI guy…
Hey CCS! I bet you can even install the Edgesight agent?! (They’ll get the joke) What’s not to like here? Yes, VMware, HP/Dell/Cisco might be a little bent for awhile since you won’t need as much hardware/Hypervisor software and Microsoft might find themselves chagrined as they cannot gauge you for more licensing costs but in the end, you get to simply extend your enterprise into the cloud without drastically changing anyone’s role. This also allows organizations to wade into VDI instead of stand at the end of the high dive at the public pool while Citrix, VMWare and Gartner chanted “JUMP, JUMP, JUMP!”
Isn’t that what we wanted when all this started?
Thanks for reading