Civil 3D + Remote Desktop Services

[BlackBox]

On the heels of this thread, we recently decided to beef up our ability to remotely access our projects, rather than to supply CAD capable mobile workstations to users and have to implement a 'project check out' procedure, which spawned this post where I'm seeking input from others who also work remotely and can clarify some of the more technical aspects of resource planning, and implementation of a more complex RDS (if not VDI?) environment from an IT perspective.

Kindly forgive the duplication, as I feel I'm posing this here to a very different audience:

We work a lot of hours as a small firm, and are trying to lessen the load on users (myself included), by enhancing our ability to work remotely when needed.

Our current environment, after the upgrades I implemented last year (since we hadn't upgraded in +/- 10 years), consists of a new Dell R620 primary physical server running Windows Server 2012 R2, that hosts our Exchange, File/DC, and Terminal Servers via Hyper-V. I also have two legacy physical servers that have been upgraded by adding (in each) Windows Server 2012 R2, a second processor, 4 x 2TB HDD physical disks as RAID virtual disk, and a 2 port Gigabit NIC PCIe card (so each has 4 ports).

These legacy servers are currently slated to become secondary, and tertiary replica nodes for CYA, but now that we're considering upgrades to support a substantially greater remote access load, I've held off with that until we identify any changes needed for this task.

At the moment, a few of us remote into our Terminal Server via RDP, and then RDP into our physical workstations to do our actual work.

There's obviously several factors at play here, but we're experiencing a bit of lag when it comes to cursor movements in Civil 3D, resulting in bad OSNAP selection, etc. (minor issues). We're on subscription for Infrastructure Design Suite, but really only use AutoCAD (Engineers), Civil 3D and Raster Design (Designers).



For those of you who work remotely for CAD design, and do not work locally using a 'project check out' procedure, what systems are you using to gain access to your remote server, and what hardware/software configuration are you using?

More specifically:

Are you remotely accessing your physical workstation in the office, or has your in-office workstation been virtualized (VDI) so regardless, you're actually working off of the server both in office, and remotely?

How many users simultaneously work remotely through a given Terminal Server, and what phsyical server resources have been dedicated to that Terminal Server's virtual machine via Hyper-V Settings, etc.?

Does your Terminal Server virtual machine's physical server host have a GPU card installed? If so, what card are you using to support Civil 3D design?

What sort of bandwidth should we anticipate needing, both in terms of the physical connection to the office, as well as the throughput via security appliance, before gaining access to our Gigabit network via Terminal Server?

Cheers

[rkmcswain]

.... I'm seeking input from others who also work remotely....

We have a few users who do this on occasion, but not very often and certainly not as a regular practice.
They connect to the office network via VPN, and then remote desktop into their own workstation.
It's been a while since I've done this myself, but I seem to remember cursor lag being the main nuisance.

[BlackBox]

We have a few users who do this on occasion, but not very often and certainly not as a regular practice.
They connect to the office network via VPN, and then remote desktop into their own workstation.
It's been a while since I've done this myself, but I seem to remember cursor lag being the main nuisance.

I'm working with a few Dell & Microsoft folks for the past few business days, and they seem to think adding a GPU to the Terminal Server VM's physical host will rectify this, but I'm awaiting additional evaluation of diagnostics from security appliance throughput, latency, etc. (aka DPACK), before I am willing to authorize such.

I keep getting opposing feedback:

1. That the 'work' is being done on our physical workstations, which makes complete sense to me as we're RDP, and that the possible bottle neck is at the security appliance, our office bandwidth connection (+/- 25Mb/s; I have 60-75Mb/s at home [ISP just upped our package for free?]), and/or lack of resources for Terminal Server, as the entire office is on Gigabit network.

2. That the display 'work' is being done by Terminal Server (hence the GPU recommendation), which makes absolutely no sense to me. I RDP into all of our workstations for maintenance, etc. and I entirely notice a difference in both the speed at which I am connected, dialogs displayed, etc. and the quality of the graphics for CAD, Admin, and a couple of our old XP boxes that are still in use (software on them isn't available in Win8.1 flavor) - and that's within the domain, not just RDP from home.


The only reason this lowly MTA Server qualified guy sees for adding a GPU, is to implement VDI, where are CAD production workstations actually reside on the server, and our physical workstations become thin-clients, consuming a CAD-specific VM from server when we log in at the office, etc. More information is needed, which is where I turned to various forums for help in clarifying the issue.

Cheers