For many years, DeltaV Remote Client has allowed users to access DeltaV Terminal Servers via Remote Desktop Protocol. This is advantageous for the following reasons:
1. Lowering hardware costs
2. Minimizing workstation setup/configuration time
If users need 30 DeltaV operator workstations, instead of
A. buying and setting up 30 physical XP or Win7 computers
they
B. get 4 terminal servers, 30 thin client RDP appliances, and host 7-8 sessions on each server.
Sounds great, except now that virtualization is upon us, doesn't it make more sense to host 30 virtual workstation instead of 4 virtual terminal servers?
With virtualized Win 7 workstations, the hardware advantages disappear, and the setup/configuration advantages can be mitigated by good administration of the workstation virtual images. Above all else, some of the particulars of the deltav application in the terminal services environment can be avoided (picking sessions, shared BOI settings, 1 terminal server effects 7 workstaions etc.).
I'd love to hear opinions / feedback on the pros and cons of these technologies. I saw a lot of information on virtualization at Exhange, but not as much on remote client.
I think the hardware advantages for the client side would be similar in both cases (RDP to a Terminal Server Session or thin client to virtualized Windows 7 workstation). For each, 30 devices would be required on the client side.
On the server side, I would agree that the now supported direction of using virtual workstations will eliminate some of the quirkiness of the terminal server solution - not having to pick sessions and individualized settings being big. I've used the RDP solution mainly for flexibility with engineers and avoided it for operators in most cases. With Emerson fully supporting virtual operator stations moving forward, I will definitely purse the option.
In reply to Michael Moody:
From articles I've read, the balance to strike when deciding between virutalization and terminal services lies in the end use experience. Terminal services works out great when serving out applications; virtualization when serving out entire operating system environments. DeltaV workstations are currently closer to the latter, (although it really doesn't have to be that way! - Operators need a handful of applications; if anything the entire OS being delivered via RDP offers more complexities than if Operate and BOI were deployed as terminal applications). I agree, TS/RDP for engineering/professional stations is fantastic.
The other advantage to virtualizing workstations is the upcoming capabilities for failover of workstations to the redundant host should the primary host become unavailable. Remote Client Terminal servers cannot provide similar failover options (especially when licensing of sessions must be accounted for upfront or else manually during a failure of the terminal server).
I am a big proponent of virtualization.
The first reason is the diverging of simultaneous hardware and software upgrades. With a virtualized solution, you may upgrade your host server hardware and guest software independently.
Second, DeltaV 11.3.1 (and prior) must run on a 32-bit operating system which limits the maximum amount of RAM available to 4 GB. If you have three or four terminal sessions on a server, then they must all share those limited resources. Virtualization allows you to assign the 4GB maximum to each virtual machine, should you so desire and if performance warrants it. DeltaV v12 will support a 64-bit operating system, but currently virtualization best addresses the 4GB limit.
Finally, a virtualization architecture can be constructed that leads itself to extreme high availability. If your terminal server fails, then you lose three or four sessions simultaneously and until you repair your server. If your virtualization solution contains a failover cluster, then a complete host server failure would take the guest virtual machines offline only long enough for the other cluster member(s) to boot them up and bring them back online.
In reply to Ben Bishop:
Virtualization will also allow for a nearly instant online upgrade of your workstations. You can clone your existing workstations, and upgrade them in an offline system. Then the upgrade is as quick as powering off the old image and booting the new image.
You also move away from the limitations of how much data the iFix service can process. On a terminal server, all of your operator interface data is coming through a single computer, running into possible issues if you have a large number of client sessions all looking at a large number of different data sources.
All in all, with support for online virtual operator stations, I think the use of terminal servers for permanent operator stations has specific use cases, but in general may not be as good a solution as virtualization.
In reply to Brian Atkinson:
Ok, I'm hearing support for my position. What about virtualizing a terminal server? Any support for doing this?
I worry it would pose too many layers, doesn't get away from terminal services environment issues for dedicated workstations, and may or may not be included in the failover function.