Hi All!
I am a (mechanical) engineer with a small contact engineering firm who has, unfortunately, become the defacto CTO for our merry band. We have 10 engineers in house and 3-5 outside. This is likely to be a long post, so I'll thank you in advance for muddling through it...
OK, A little history about me and the company so you all have some perspective. I joined the company 3 years a go when there were only four of us and we were working off of 4 workstations with USB external drives mapped as network shares. We got a real backup when we were scared, or every couple of weeks, with ever came first. Realizing that this was not a very durable condition, I built our first server. This is an Essentials 2012 box with a 4 disk SATA hardware raid 5 (on the motherboard controller) and a 3 GB internal disk as back up. Off site back-up is accomplished by rotating USB disks. It's a pretty capable little machine with a 6 core E5 Xeon and 32GB Ram, redundant power supplies and is UPS protected. This worked great for a couple of years and 6 more users and we never taxed the server at all. Typically under %10 CPU and %25 memory. Then we got offsite employees.... All over the world. So, I configured Anywhere Access with a VPN and RemoteApp for a Timekeeping database and away we went. Except that now my little server need to be up 24/7 and I can hardly schedule a shutdown without affecting production, and if something really fails, were &@#$ed. End of the history segment.
The current server handles the following: 1) Domain controller, AD Services 2) Anywhere Access 3) SQL Express Server 4) Autodesk Vault Server 5) File server.
So, what I am thinking is we need to build another matching server, install full 2012R2 configure it as a Hyper-V Host, and deploy the aforementioned workloads as VMs. Then Take the original server down, install full 2012R2, configure it as a Hyper-V host and add it back in to the environment. This is where I get lost. From What I have read, to build a failover cluster requires a SAN. This could be an iSCSI box on the cheap side, but I don't see how to create a redundant iSCSI installation... I could also go direct attached SAS with a dual expander JBOD and clustered heads running full 2012R2, but from what I understand, I would need THREE JBODs to be able to take one down for maintenance, or I'm back to a single point of failure at the JBOD.
I don't need to support hundreds of users, I have a hard time seeing us ever break fifty. I don't need extreme performance, but an improvement will help the boss feel good about spending the money. I need to be able to take any machine offline for maintenance without bringing the whole infrastructure down. I have a budget of around $6K. It will cost approx. $1600 to duplicate the original sever and I have 4 2012R2 licenses available, but no CALs. figuring $1200 for CALs, that means I have about $3K for new Hardware. BTW, we are %100 1GE network, and have just one 16 port switch.
What are my options? Do I need to go the cluster route, or is there a cheaper way to do this? can you do redundant iSCSI? Where does Hyper-V replica fit in? Storage spaces seems to play a role here too, but seems super picky (read super expensive) about SAS hardware. Holy S**t Batman, this is complicated.