Hyper-V and Active directory setup in a small buiness.

R

relink2013

n00b
Joined
Jun 17, 2015
Messages
14
Hello there, I manage a small family owned manufacturing company. I am looking to centralize our computer systems, and begin properly managing our setup before It ends up getting too out of hand.

We currently have 6 workstations, 2 of which are CAD stations. My hope is to use Hyper-V to virtualize all 6 of these.

What I am going to ask is probably a big order, so I dont really know what kind of answer to expect.

In a perfect world what I would like to do is this.

Hyper-V:
1. Have all workstations running on Hyper-V (everyone gets a laptop)
2. Have a small Cluster so that we can have as little downtime as possible.
3. The 2 CAD station VMs need decent graphics (but nothing ridiculous, my current setup uses a 660Ti and it does everything I need.)

Active Directory:
1. Simply put I want to be able to manage Windows user accounts, shared folder permissions, printer permissions, VPN, etc, all in one place. We have a Synology DS412+ as our file server, which I believe supports this.
2. I would like this to be separate from the Hyper-V host.

Ok, so my idea is this - and im sure ill be corrected if this is dumb -, we don't have a million dollar budget for this, so what I was going to do is setup a cluster of "consumer grade" hardware to handle Hyper-V, now of course I'm not talking i3s or anything. From my understanding I can cluster multiple servers together - so long as they are identical - and the load will balance across the cluster. Should one of the servers fail, shutdown, or whatever, the VM will automatically move to one of the other servers and the client user shouldn't even notice that anything happened...is any of this even correct?

Some other questions:

1. The biggest question, How much - estimated - bandwidth would be required to connect people via Remote Desktop to the Hyper-V server. I am estimating that each user would be running at 1080P 60hz 32bit color. (Just a guess from other remote desktops I have used would be about 25Mb/s per client or more. )
2. how much bandwidth would I need between nodes?
3. How do you go about correctly setting up the cluster?
4. What happens if a node actually needs to be replaced?
5. I don't know which option to go with for storage, iSCSI, SMB3, there are so many options, and I have no idea what kind of connection to use between the storage, and hyper-v servers.

I think I got all my thoughts out, I hope my post was coherent enough, I have many many other jobs around here so my mind is in a million different directions right now. I have looked up tons of information on all of this and think I have a pretty good handle on what to do, but finding information on some of the finer points -such as bandwidth- seems to be difficult. Not only that but finding a straight answer on some of the more technical aspects - such as the clustering - seems to be almost impossible. These are all thing that I could probably figure out through trial and error, and I probably will mess around on some of my home servers, but for work, I really can't afford to make any mistakes.

I have been coming here for years even though this is my first post. so I hope you guys can help shed some light on what I am trying to accomplish.
 
First show-stopper I see is. . .

The 2 CAD station VMs need decent graphics (but nothing ridiculous, my current setup uses a 660Ti and it does everything I need.)
Click to expand...

If you're going to be running all your workstations via "thin client" or RDP off of the Hyper-V server, you're not going to have any accelerated graphics at all. Unless I'm out of the loop or my info is outdated, there's no video card passthru feature in Hyper-V.
 
relink2013 said:
From my understanding using RemoteFX you can deliver hardware accelerated graphics upto DirectX 11 from a Hyper-V VM.

No OpenGL support aside from Software OpenGL 1.1, but I can make everything we run use DirectX so I'm not worried about the lack of OpenGL.

http://blogs.msdn.com/b/rds/archive...-remotefx-vgpu-on-windows-server-2012-r2.aspx
Click to expand...
I stand corrected. What's worse, I now remember reading about that. Which explains why I wrote my prior post with a buzz of hesitation firing in the back of my brain.
 
What is the budget, cause you will want a min 2 back end servers running everything for fail over / redundancy.

otherwise your 1 hyper-v box goes down and no one can do any work...
 
Lol, no worries. Also, Apparently In Windows Server 2016 they are - from my understanding - going to be adding in Hardware Accelerated OpenGL too.
 
there is no "budget" to speak of, It's hard to budget for something when I don't know what all is involved. But we are still fairly small, so I'm trying to keep it as low as possible.

But 90% of the workload that will be on this server is word processing, spreadsheets, and email. Clustering the servers together was why Im probably going to go with "consumer" grade hardware. It's not as good, but the load isn't that intense, and I would have more money to invest in 2 or possibly even 3 servers so I can get that protection. I definitely need that redundancy.

Also a Hyper-V cluster can perform load balancing as well correct?
 
I run two Hyper-V hosts and a separate DPM server. I've glanced at clustering, but instead I just try to make sure that either one of my Hyper-V hosts has enough RAM and capacity to take on the VMs of the other in case of hardware failure. Sure, we'd be down for the affected VMs while I do the transfer manually, but we're not splitting the atom where I work, and nobody would lose any money.

Enough about me. . .I could be wrong about all of this. . . but if you go with SMB3 via Windows Server 2012 R2, you're looking at a whole other cluster just for storage. That's on top of the Hyper-V hosts in their own cluster.

I once connected our Hyper-V hosts to a Synology device via iSCSI and it worked pretty well for hosting the storage of two or three servers. You should be able to get your Hyper-V cluster to use iSCSI and avoid the need for a separate cluster for SMB3. However, I'd be concerned about GbE network connectivity between the Synology device and Hyper-V hosts. . . you're going to max out GbE at 112 MB/s and a workstation nowadays with its own internal drive can easily do 80MB/s+. So you're probably looking at sub-workstation performance if more than one workstation is hitting the array hard.

Gotta run. Drupal updates just released *sigh*.
 
My first question would be why your want to centralize everything to that degree? With only 6 users, You are createing a point of failure, as well as a technical dependence upon yourself. Ask yourself this, "Do I really want to get called at 3am when someone can't get into their desktop?

That being said, I'd build 2 identical Servers with a bunch of RAM, solidly performing Disk IO, and a RemoteFX supported Video Card. And use Hyper-V Replica to keep all VMs in sync on the spare Server.in case of failure. At the same time, I'd keep a/the Domain controller in order for AD to survive an outage. Adding everything up, you're looking at a minimum of 2 Windows Server licenses, Windows 7/8/10 licensing on top of fairly beefy hardware. I don't see the economics of scale working well here.

Another alternative is to find a company offering DaaS (Desktop as a Service) that can accommodate hardware video acceleration as well and move everyone up into the "cloud". For 6 users, I'd see that as being more cost effective both from capital outlay as well as reducing your continued support of a monstrosity.
 
I think I've gotten some pretty solid advice so far.

@Nate7311, I have considered what you said before exploring this path. With the way people are around here, our computer situation in infuriating as it is. At least with this option it's all in one place, I can manage everything. from one spot. On top of that, people move around here so much that being able to just remote into your own "desktop" any time from any where in the building will be a huge plus for everyone, ill end up looking like Batman and SuperMan combined over here.

Now I do have a question regarding storage. Hurin mentioned iSCSI and Nate7311 mentioned building servers with thier own storage (at least that how I took it).

So which would be easier to manage?

A. Build the servers with their own RAID array and run the VMs using the servers local storage, and then use the Replica feature (ill have to do more research on this feature) to keep the 1st server cloned onto the second so in the event of a failure I just boot up the other I assume, and everything peachy again.

OR

B. Setup shared storage through iSCSI and use the automatic failover clustering?

From what I can see Option A sounds allot cheaper, and would require much much less networking between the 2 servers, and would probably perform better with less equipment to buy. but I would have to manually start the VMs on the 2nd server in the event the 1st one goes down. Option B sounds like it would perform much much slower unless I had a team of several 10Gb/s NICs (which are stupidly expensive), but I would gain the ability to have the server automatically failover in the event of a crash...So unless I missed something Option A seems like a much more viable option right now.

I really appreciate the help so far, Im trying to go step by step to figure out exactly what I needs to do this.

On a -slightly- side note, Im going to be setting up some of this stuff in my home "lab" to mess around and get more familiar with how it works. Do the servers need to be identical in order to use the Replica feature? I have several computers that are powerful enough to run at least 2 VMs, but none of them use the same hardware. could I have one setup with an i7 and one with a Phenom II X6, or do they need to be identical for it to work? (note this is just for practice at home, for work I will be building 2 identical servers from brand new components)
 
Personally, I've always shied away from automatic fail-over clustering. The complexity and expense just hasn't been worth it to us.

If I were you, I think I'd just build two Hyper-V hosts with enough hard drive capacity so that either can host everything on its own, and just handle things manually. Unless we're talking about multi-TB datasets, you should be able to get your VMs over to the unfailed host in less than an hour or two (from backup or perhaps directly from the array on the failed host via some shenanigans, any thoughts on your backup solution?).

The second/backup/aux server could be a little less beefy than the first (though I'd get the same motherboard and CPUs, though not strictly necessary). Slower hard disks, more capacity-efficient array, less RAM, etc.

Incidentally, you'll want a separate domain controller VM running on each server. It's not strictly necessary (I've seen it work with a single domain controller hosted on the Hyper-V host, the Hyper-V host doesn't seem to mind that the dc is down as it boots) but it's a good idea.

I custom-built our Hyper-V hosts. One word of advice. If you think you'll only need 32GB of RAM on the server, get 96GB. If you think you'll only need 4TB in the storage array, get 12TB. 2 NICs? Get 4. :)

--H

P.S. Don't use replication on a domain controller VM.

P.P.S. I should mention that I haven't experimented with the built-in replication function though it's probably just a few clicks away from being functional on my two servers (I just recently finished moving everything back and forth between the two (and a spare 3rd machine that hosted a dc VM) while I wiped and reinstalled on my two primary servers. Hello Server 2012 R2!).
 
Last edited:
Sounds like good advice to me, this is seeming less and less difficult. :D

As for the Domain Controller -which I know very very little about- I was just going to have that be on it's own hardware on our network. would that not work?

The one thing I would have to make sure to setup though is the automatic replication, from my understanding you can even set it up to run every 30seconds?? This is because if we go down, I need to be able to get everyone back up quickly, and while loosing as little work as possible (preferably none)
 
Sounds like that workplace is pretty hardcore for being so small. =)

And if so, you really don't want only one domain controller (dc). If you lose your single dc, nobody can log into anything (assuming of course that your workstations are domain members).

So long as you're going to have two hyper-v hosts, you might as well have a dc on each. Domain controllers in such a small environment are pretty much "set and forget". . . they take care of themselves very well. And it's nice to have dcs "encapsulated" in VMs. But, you must be very, very careful never to use snapshots or other similar features with them, because doing so can mangle your domain (a dc that travels back in time may not be aware of it and mayhem ensues).
 
Some people will probably laugh. . . but for non-admins who are looking to become admins, Mark Minasi write really approachable, plain english, "let's talk about it and then show you how it's done" books for Microsoft Server OSes. Here's the latest for Server 2012 R2.

He goes over everything from DNS, to domain concepts and setup, to Hyper-V.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
I just ordered that book, wasn't too pricey, and it cant hurt to have.

So now I get why you mentioned running the DC on the hyper-V servers, makes sense now to have 2 of them. But how do the two DC not interfere with each other, while staying in sync as well?

Also how do I setup the two Hyper-V servers so that they stay in perfect sync with each other, but don't interfere with each other? I believe the term is "Split-Brain", how do I ensure that these things can work in harmony, and not interfere with each other? Also how do I make it so that the client computers even know to connect to Server 2 and not Server 1, without having to go and manually point them all to a different IP, and then how do they know to go back to server 1 once it's back online?
 
relink2013 said:
So now I get why you mentioned running the DC on the hyper-V servers, makes sense now to have 2 of them. But how do the two DC not interfere with each other, while staying in sync as well?
Click to expand...
Active Directory (which is what Microsoft domain controllers use) is built to have multiple servers active at the same time. So, the short answer to your question is: They don't interfere with each other because that's how it's designed to work. To be clear, these won't be clones of each other. It will be one dc that you create and set as dc1.yourdomain.com and a second dc that you build and run simultaneously as dc2.yourdomain.com. They constantly talk to each other, handle authentication requests as needed, and keep the AD domain in sync.

Also how do I setup the two Hyper-V servers so that they stay in perfect sync with each other, but don't interfere with each other?
Click to expand...
Well, that sounds more like fail-over clustering. I don't know how you do the equivalent with just replica VMs. Someone else will have to speak to that.

I believe the term is "Split-Brain", how do I ensure that these things can work in harmony, and not interfere with each other?
Click to expand...
I've only heard of "split brain" in the context of DNS. But maybe it's used elsewhere.

Also how do I make it so that the client computers even know to connect to Server 2 and not Server 1, without having to go and manually point them all to a different IP, and then how do they know to go back to server 1 once it's back online?
Click to expand...
Assuming you're doing Hyper-V replicas, I would assume that the replication/failover service won't allow both to be up at the same time. Each of the replicas will use the same IP and be essentially clones of each other. Which Hyper-V host they're on is irrelevant in that case. It's all handled by IP-routing.
 
relink2013 said:
I just ordered that book, wasn't too pricey, and it cant hurt to have.

So now I get why you mentioned running the DC on the hyper-V servers, makes sense now to have 2 of them. But how do the two DC not interfere with each other, while staying in sync as well?

Also how do I setup the two Hyper-V servers so that they stay in perfect sync with each other, but don't interfere with each other? I believe the term is "Split-Brain", how do I ensure that these things can work in harmony, and not interfere with each other? Also how do I make it so that the client computers even know to connect to Server 2 and not Server 1, without having to go and manually point them all to a different IP, and then how do they know to go back to server 1 once it's back online?
Click to expand...

That's the entire purpose of Active Directory - they're self clustering. One will be your first DNS server, the second your backup, and they handle the rest (including telling each machine who to talk to).
 
Quick blurb on Hyper-V Replica: http://www.aidanfinn.com/?p=12147

Given the 2 routes that have been mentioned here. Hyper-V cluster with shared storage and Replica. I'll bet that it still comes back down to cost to implement. It almost always does. That being said given the heavy duty Apps that you are talking about, you're looking at some hefty servers to run them on. Adding in the disk IO (whether local or in a SAN) to run them well, makes the project more expensive. I'd guess that if you buy new hardware, this will be a $25k+ project. Is that in the budget? If it's not, then this is a hypothetical discussion :D.
 
Ok, back after a great fathers day weekend. Thanks to Amazons zippy shipping my book should be here today.

Active Directory (which is what Microsoft domain controllers use) is built to have multiple servers active at the same time. So, the short answer to your question is: They don't interfere with each other because that's how it's designed to work. To be clear, these won't be clones of each other. It will be one dc that you create and set as dc1.yourdomain.com and a second dc that you build and run simultaneously as dc2.yourdomain.com. They constantly talk to each other, handle authentication requests as needed, and keep the AD domain in sync.
Click to expand...

Looks like I really need to learn more about how Domain Controllers work. I have mostly been focusing on the Hyper-V side of things. The first thing that popped into my mind after reading that they wont be clones of each other, is if one server goes down then what happens?

Given the 2 routes that have been mentioned here. Hyper-V cluster with shared storage and Replica. I'll bet that it still comes back down to cost to implement. It almost always does. That being said given the heavy duty Apps that you are talking about, you're looking at some hefty servers to run them on. Adding in the disk IO (whether local or in a SAN) to run them well, makes the project more expensive. I'd guess that if you buy new hardware, this will be a $25k+ project. Is that in the budget? If it's not, then this is a hypothetical discussion .
Click to expand...

Believe me when I say that most of what we do around here is pretty light weight. These CAD systems don't have the resource usage that people think they do...well unless your Boeing or NASA anyway. I do most of my work in a 2D cad system called AutoSketch, and some of our newer projects are being done in Sketchup, or SolidWorks. However our drawings are generally of very small electric components, were not designing spaceships or anything, so the workload is small enough that I am able to smoothly work using SolidWorks on my 1st gen Lenovo Yoga, right now, now im sure if I loaded up a cad drawing of a 747 my computer would probably burst into flames. Other than that (which we only have 2 CAD operators, myself and the owner), the rest of the workload is mostly Microsoft Office, Quickbooks, and Fishbowl.

My main goal here is to:
A. Use Hyper-V to centralize all workstations so that they can:
1. Be accessed from any computer in the building. (We currently have people who move around allot and need me to move their desktop every time they move)
2. Have a clone so that if something goes wrong, no big deal. (it's cheaper and easier than giving everyone 2 computers)
3. Easy to upgrade all "worstations" at once, and be able to appropriately (and even dynamically) allocate system resources the the users who need them.

B. Use Active Directory to:
1. Be able to centralize all user accounts (this can get sticky, especially when someone gets let go and decides to change the password on their computer before they leave)
2. Beable to control permissions across the network, including our Synology Server. (I currently have to finely control permissions per user across several different servers right now, and it tends to get a little messy.)

C. Finally I would like to shut everything off to the outside world except our phone system, and then set up a VPN so that the people who need to work at home, can work just as if they were sitting in their office.

Than after all this is setup and running, my next steps are a centrally managed Anti-Virus/Malware system (this is currently a huge pain because everyone likes to just ignore the virus scans and updates.) and then eventually setup an exchange or some other kind of shared calendar system. We currently use the CalDAV server that's built into our synology, but It just doesn't seem to be able to handle the workload, it constantly drops events, and needs to be rebooted. It's also a ridiculous waste of time to go around to every computer in the building and manually connect them to all 7 of our calendars.
 
You really need to read up on AD. They just take over for eachother as needed (gross simplification).
 
Well I've got a 2" thick book sitting right next to me that covers everything I should need to know about Windows Server 2012 R2...at least to get me started on what I want to setup.

Im keeping this tab pinned so Ill be sure to keep checking this thread, and ill post back if I have any questions, or successes. When I get this going I plan on doing a whole write up on how I did it, it's a great way to remember things, while helping others.

I really appreciate everyone's help, Google is great, but it will only get you so far, sometimes you just need to ask people who know.
 
And do- you'll learn a lot about AD when you do something wrong, and have to fix it later. That's one of the fastest ways to learn - and search technet when you hit problems. Despite how often many of us have laughed at MSFT over the years, when it comes to their server products, they take them VERY seriously, and the community on it is spectacular.
 
back up here a second,,,, unless they are doing very simple stuff cad is going to be a major problem.
What cad programs are they using and how heavy are they using it?
Does the cad package itself offer support for being on a virtual machine?
Have others been successful at getting it to work?
Are they doing large renders?

What I have done in the past is set up dedicated high end systems in the data center and had the cad users rdp or vnc to them one per system.
Some acceleration features in some cad products don't work with rdp but they do with vnc you will need to test it out.
 
What cad programs are they using and how heavy are they using it?
Click to expand...
1. AutoSketch (old 2D system), SolidWorks, Sketchup
2. Not heavily

Does the cad package itself offer support for being on a virtual machine?
Click to expand...
1. AutoSketch (yes but the whole program has been discontinued since 2009 anyway)
2. SolidWorks (yes)
3. Sketchup (yes, provided there is OpenGL acceleration, which is supposed to be coming in server 2016)

Have others been successful at getting it to work?
Click to expand...
1. AutoSketch (doubt anyone has tried, doubt it will be any trouble)
2. SolidWorks (Yes)
3. Sketchup (Aside from the lack of hardware OpenGL, yes)

Are they doing large renders?
Click to expand...
No

What I have done in the past is set up dedicated high end systems in the data center and had the cad users rdp or vnc to them one per system.
Click to expand...
Considering the low number of cad operators here I have actually considered doing just that.
 
Does anyone have any recommendations on bandwidth for each connection? More specifically a way for me to calculate the bandwidth myself?

Here are my guesses. If I went with shared storage, and 6 VMs

Storage to hyper v : 10Gb (at full load this should provide each VM with SATA 1 speeds)

Hyper v to main network (where clients connect): 2Gb (using rdp I have been monitoring my bandwidth usage and have yet to see it go over 5Mb/s, so 2Gb/s should give me way more bandwidth than I need.

Live migration: I have no clue
Replication : I have no clue
Heartbeat: I assume this can be just about anything
Management: again, does it really matter?
Any other connections that are required: no clue

I have tried and tried to find any recommendations, or guidelines on how to calculate bandwidth requirements and I simply can't find anything.
 
rdp does not allow opengl acceleration however I had some luck with vnc allowing it.
You can test it with your current cad stations.
I strongly recommend dedicated cad machines.
Be careful because many vnc installers are infested with malware.
Run malwarebytes after installing it and only download from the projects website not mirrors.
 
Let's take a look at this from a couple viewpoints. First, let's discuss the practicality, and then, we can discuss the possibility.

So first, practicality. I've looked into this solution a few different times for customers who have wanted to do the same type of thing, even with less-demanding (basic) applications and do basically a VDI solution for their computers, using basic computers or laptops at the end-user point to remote desktop to the server. However, the cost to do this is going to be SO much more than you'd be spending on NICE (and even duplicate backup spare hardware) individual computers for end users instead.

Let's say you want to get two decent servers to run what you are talking about. Given the processing, RAM, and storage capacity needs you are talking about, local storage is the way to go. Don't go with a SAN and iSCSI as that just further complicates your network, increases the cost by several factors, and creates an inverted pyramid of doom. Two servers, local storage, and you can replicate between the servers as well if you want. The biggest thing you have to take into consideration here is a separate system just for storage of your backups, as backups are your most important thing.

Two servers with the performance capabilities you are needing (and some room to grow) are probably going to run you about $5,000 per server (give or take.) Given the number of virtual machines you are talking about, you probably want to purchase at least two licensees of Windows Server 2012 R2 Standard, which will give you a total of four WS12R2 VMs. You should be able to use Windows 7/8.1 licenses for your other individual virtual machines for your end users to remote desktop into for daily use. All total you're looking at about $2,500 for licensing with the additional user CALs. On the rest of the hardware side you're going to need some decent network bandwidth between servers and the rest of your core network, so at minimum directly-connected 10GbE network adapters and cables, $1,000 total, and ideally some 10GbE capable switches, $1,000 for one of those as well. You're going to want storage and software for backing up all of that as well, we'll say a good $3,000 for that.

So just for your server infrastructure you're looking at $17,500. You still have to purchase and support individual desktops/laptops for the end user to use in remoting to their virtual desktop, so you are still looking at an average cost of $500 per computer there, so your total infrastructure is around $20,500. That's a minimum cost for this sort of configuration.

Let's compare that with a standard setup with individual desktops and a single (backed up) server for domain/storage, etc. A single server with a single license of Server 2012 R2 Standard, probably around $5,000. Add your backup for about $2,000 (less server, less backup.) Your server infrastructure cost is now only $7,000 approximately. If you were comparing with the budget from the above option, that leaves you with $13,500 to spend on end users computer systems, or approximately $2,250 which will be more than enough to get you very nice desktops/laptops PLUS have spares available on hand for replacement if needed. You also would see better performance (local will almost always beat remote) and less single-point-of-failure issues and complexities.

Now that we've covered the practicality of it all, though, we can discuss the possibility.

Yes, what you are describing is possible to do. As the core of your work is going to be relying on your server infrastructure though you want to really put your money there to be sure it is done right. After all, since all your eggs are now in one basket (even if there are two servers) you have to be sure those baskets are big enough for your eggs even a few years down the road. Get something with business-class warranty support. Last thing you want is a major bug causing you to be waiting around for replacement hardware for a week or trying to research by yourself why these drivers are causing you to blue screen twice a day. You also have to consider your network needs (higher demand via remote access and higher throughput demand for transferring data for backups and between servers for replication.) The other thing I haven't seen mentioned much here is backups. Even if you have replication going, data and VMs can get corrupted. It's not pretty and it's not easy to fix when it happens. Replication is not really backup, it's just a way to get you up and going faster if something on one server goes down. So you need to keep in mind that all of this needs a good backup system in place.
 
You bring up some very valid points. This is still an avenue I would like to pursue at some point, but I am thinking that just letting people remote into their own individual workstations if they need to be away from their desk will be the way to go for now.

BUT, now for the fun part, the company bought me an HP ProLiant ML310e Gen8 v2 for my own personal use, with the strict understanding that I learn everything I can about SharePoint, Active Directory Directory Services, and exchange.

Of course I figured I'd learn about Hyper-V while Im at it, so I ordered a bigger PSU, and I have an HD7870 waiting to be installed just so I have some GFX to play with as well. I threw in a 1TB HDD to start with, and upped the ram to 16GB, considering this is at home, my wife and I will be the only ones using it, eventually (if needed) ill up to 32GB.

So I've got some really fun work ahead of me.
 
relink2013 said:
You bring up some very valid points. This is still an avenue I would like to pursue at some point, but I am thinking that just letting people remote into their own individual workstations if they need to be away from their desk will be the way to go for now.

BUT, now for the fun part, the company bought me an HP ProLiant ML310e Gen8 v2 for my own personal use, with the strict understanding that I learn everything I can about SharePoint, Active Directory Directory Services, and exchange.

Of course I figured I'd learn about Hyper-V while Im at it, so I ordered a bigger PSU, and I have an HD7870 waiting to be installed just so I have some GFX to play with as well. I threw in a 1TB HDD to start with, and upped the ram to 16GB, considering this is at home, my wife and I will be the only ones using it, eventually (if needed) ill up to 32GB.

So I've got some really fun work ahead of me.
Click to expand...

I currently have all my home services runnong on a ML310e G8 V2 as well. They are quite nice home servers and you can do a lot with them.

For a small company, going with a local Exchange server is definitely not going to be the best route. It will be much more cost effective, much more resilient, and much less complex to go with a hosted Exchange server, such as Office 365. Active Directory is a good place to start though, setting up your host server for Hyper-V and a VM for your domain controller/DNS/DHCP services.
 
For a small company, going with a local Exchange server is definitely not going to be the best route. It will be much more cost effective, much more resilient, and much less complex to go with a hosted Exchange server, such as Office 365.
Click to expand...

Unfortunately it's the only route. We have recently enacted a strict "no-cloud" policy. Currently the only thing we have hosted off site is our email, and that will be changing as soon as we can ensure we have the ability to keep our own email server up 24/7.
 
Ok so after heavily weighing my options, and about a 2 hour discussion with boss man, we are going to look into going with a hosted solution, but only for one year, then we're going to migrate everything down to our own servers.

We are currently about to begin moving the entire company to a new facility, while at the same time growing, taking on new business, etc, so it's a very vital time for us and very hectic as well. So I made the call to go with 365 for a year and by then I should have a firm grasp on how to setup and manage these servers, and we will have a much bigger budget as well.

Can anyone give me any info, or experiences, opinions etc on the hosted exchange, share point, etc? I'm actually having trouble finding info on those specific parts. Microsofts website just wants to go on and on about one drive.

Also I assume I would still need to host the directory server here on premise?

Lastly, does anyone have any tips on the best way to create a domain name to connect with all these services? The reason for this is I'm trying to think ahead. If I create a domain name the forwards to the proper addresses and everyone here uses that domain name to connect to exchange, share point, etc, then when the time comes to bring our data down into our own servers all I will need to do is change the address the domain name points to and everyone "should" be able to continue working as if nothing happened, and I won't have to go re setup everyone's computers to connect to the new servers.
 
You should keep Email and SharePoint in the cloud. Less maintenance and licensing costs to worry about. Otherwise, there is the Hybrid Exchange model.

Your on premise site, you need Active Directory domain controllers and Active Directory Federated Server. You can setup Single Sign On. User authentication will synchronize between your office and Office365. There is lots of MS documentation on setting this up, as well as DNS records transition.
 
You must log in or register to reply here.
Back
Top