HCHTech
Well-Known Member
- Reaction score
- 4,157
- Location
- Pittsburgh, PA - USA
Wall o' text - sorry!
I have an 10-employee accountant client that uses the Walters Kluwer / CCH software suite. They currently have two servers:
An Intel Hyper-V host with a single guest, the DC. New in 9/18, running Server 2016. 1 x Xeon E5-2620 (6-core plus HT, so 12 available cores to allocate), 32GB RAM, and a 500GB RAID 1 SSD array for the OS drive and a 500GB RAID 1 SSD array for the VM drive. We normally would have done a RAID 10 array for the VM drive, but server-grade SSDs were expensive and they beat us up on price a bit to get this deal done, so RAID 1 (and 32Gigs of RAM instead of 64) saved us enough money to get them to say yes.
The DC Guest is allocated 24GB of RAM, 10 cores of the CPU and 400GB of disk space (200GB in use)
Then, we have a standalone server, dedicated for the WK software. It was new in 10/16. It has an Opteron 6320 processor, 32GB of RAM and 5 x 900GB SAS 10K spinners in a RAID 10 with a hot spare. That gives us 1.8TB of usable storage, of which, 200GB is used. This server is running Windows Server 2012 R2.
So we have one new-school and one old-school server. We inherited the older server and built and installed the newer one when we got this client (replacing an ancient SBS 2011 Rig) in 2018. We fully expected that we would replace the WK server with an additional VM when it reached the end of it's life.
Enter the software vendor, who sensed we were too happy with our situation and decided to stop support for Server 2012 at the end of this year (despite the fact that Server 2012's EOL isn't until 10/23).
So we have a choice of upgrading the 2.5 yr old standalone server to Server 2016 (WK doesn't support 2019 yet), or upgrading the 9 month old Intel server with more RAM, more storage and perhaps an additional processor, then creating a new Server 2016 VM for the WK software.
Obviously, I would prefer to do the upgrade on the Intel box and just decommission the older server, but my client very reasonably wants to see the costs either way since he is still smarting a bit from buying 2 servers within 24 months. (Plus, he's an accountant, so always wants all of the details)
1. If we choose to upgrade the 2012 server, we'll need to purchase a license for Server 2016, and essentially backup the data, wipe the disks and install everything from scratch, then restore the data, rejoin it to the domain and repoint all of the workstation installs to the new server. Considering this is a mission-critical application for them, we would have at most a Friday-evening through Monday-morning window to accomplish the job. Oh, and the WK folks unhelpfully do not offer weekend support. Their support also ends at 7pm my time on weekdays.
2. If we choose to upgrade the Intel server, we can add the new components in a single weekend visit, then we could create the new VM remotely at our leisure, AND, we wouldn't have to purchase a license for Server 2016, since we've only used 2 installs (Host plus DC Guest). This way, we could get the new server all configured and then revisit over a weekend to re-point the workstations to the new WK server, then decommission the old one. Much less pressure.
#2 is easier for us, and gets a better end result, but costs more for the client. #1 is cheaper for the client, but obnoxious for us since we can only work after hours on the project, and will have the pressure of time to get everything up and running by Monday morning (oh, and did I mention that WK doesn't have weekend support?)
Because I have to justify my recommendation, and show the dollars both ways, I'm a bit worried. I don't want to lean on the "#1 is too hard" reason for recommending option 2, even though that's a big part of the reason. It's obvious the #2 is the "right" way to go, but I'd love him to reach that conclusion on his own without me having to talk him into it.
Am I missing anything? Are my conclusions correct? What would you do with this situation?
I have an 10-employee accountant client that uses the Walters Kluwer / CCH software suite. They currently have two servers:
An Intel Hyper-V host with a single guest, the DC. New in 9/18, running Server 2016. 1 x Xeon E5-2620 (6-core plus HT, so 12 available cores to allocate), 32GB RAM, and a 500GB RAID 1 SSD array for the OS drive and a 500GB RAID 1 SSD array for the VM drive. We normally would have done a RAID 10 array for the VM drive, but server-grade SSDs were expensive and they beat us up on price a bit to get this deal done, so RAID 1 (and 32Gigs of RAM instead of 64) saved us enough money to get them to say yes.
The DC Guest is allocated 24GB of RAM, 10 cores of the CPU and 400GB of disk space (200GB in use)
Then, we have a standalone server, dedicated for the WK software. It was new in 10/16. It has an Opteron 6320 processor, 32GB of RAM and 5 x 900GB SAS 10K spinners in a RAID 10 with a hot spare. That gives us 1.8TB of usable storage, of which, 200GB is used. This server is running Windows Server 2012 R2.
So we have one new-school and one old-school server. We inherited the older server and built and installed the newer one when we got this client (replacing an ancient SBS 2011 Rig) in 2018. We fully expected that we would replace the WK server with an additional VM when it reached the end of it's life.
Enter the software vendor, who sensed we were too happy with our situation and decided to stop support for Server 2012 at the end of this year (despite the fact that Server 2012's EOL isn't until 10/23).
So we have a choice of upgrading the 2.5 yr old standalone server to Server 2016 (WK doesn't support 2019 yet), or upgrading the 9 month old Intel server with more RAM, more storage and perhaps an additional processor, then creating a new Server 2016 VM for the WK software.
Obviously, I would prefer to do the upgrade on the Intel box and just decommission the older server, but my client very reasonably wants to see the costs either way since he is still smarting a bit from buying 2 servers within 24 months. (Plus, he's an accountant, so always wants all of the details)
1. If we choose to upgrade the 2012 server, we'll need to purchase a license for Server 2016, and essentially backup the data, wipe the disks and install everything from scratch, then restore the data, rejoin it to the domain and repoint all of the workstation installs to the new server. Considering this is a mission-critical application for them, we would have at most a Friday-evening through Monday-morning window to accomplish the job. Oh, and the WK folks unhelpfully do not offer weekend support. Their support also ends at 7pm my time on weekdays.
2. If we choose to upgrade the Intel server, we can add the new components in a single weekend visit, then we could create the new VM remotely at our leisure, AND, we wouldn't have to purchase a license for Server 2016, since we've only used 2 installs (Host plus DC Guest). This way, we could get the new server all configured and then revisit over a weekend to re-point the workstations to the new WK server, then decommission the old one. Much less pressure.
#2 is easier for us, and gets a better end result, but costs more for the client. #1 is cheaper for the client, but obnoxious for us since we can only work after hours on the project, and will have the pressure of time to get everything up and running by Monday morning (oh, and did I mention that WK doesn't have weekend support?)
Because I have to justify my recommendation, and show the dollars both ways, I'm a bit worried. I don't want to lean on the "#1 is too hard" reason for recommending option 2, even though that's a big part of the reason. It's obvious the #2 is the "right" way to go, but I'd love him to reach that conclusion on his own without me having to talk him into it.
Am I missing anything? Are my conclusions correct? What would you do with this situation?
