Figured I'd detail my homelab here, especially as I make a few upgrades.
Physical Hardware:
ESX1 - Xeon E3-1265L v3 / 16GB ECC UDIMM / 80GB S3500 SSD (OS) / Intel X520-DA2
ESX2 - Xeon E3-1265L v3 / 16GB ECC UDIMM / 80GB S3500 SSD (OS) / Intel X520-DA2
ESX3 - Xeon E3-1265L v3 / 16GB ECC UDIMM / 80GB S3500 SSD (OS) / Intel X520-DA2
ESX4 - Xeon E3-1265L v3 / 16GB ECC UDIMM / 80GB S3500 SSD (OS) / Intel X520-DA2
NAS1 - Norco 4220 / Xeon E5-2630L v2 / 128GB ECC RDIMM / 80GB S3500 SSD (OS) / 5x Fusion-IO 1.2TB (RAIDZ NFS share to ESX) / Intel X520-DA2 / H310 IT v20
NAS2 - Norco 4220 / Xeon E5-2630L v2 / 128GB ECC RDIMM / 80GB S3500 SSD (OS) / Array1 - 9x8TB RAIDz2 / Array2 - 9x4TB RAIDz2 / Intel X520-DA2 / H310 IT v20 / H310 IT v20 / H310 IT v20 / Intel X520-DA2
Brocade ICX6610-48 with full licensing (courtesy of STH forums)
ESX HA Cluster:
ESX1-ESX4 are in a HA ESX cluster of four nodes, running around 20 VMs (detail to follow). I can safely tolerate a 2node failure with 4hosts. All storage is on NAS1 across the Fusion-IO array. NFS speed tests around 1GB/s R/W for my VMs.
I utilize a VMUG Advantage yearly subscription for licensing.
I run a virtual vcenter server, but it does not run inside the HA cluster, instead running on NAS1 which is actually a ESX box with a virtual NAS appliance. It is stored on the same Fusion-IO array as the other VMs
VM details:
I have ~5 server 2012 boxes running different services, including DNS, DHCP, AD, WSUS, BlueIris
I have 1 server 2012 box running "applications that let me maintain my plex library", including MediaCenterMaster (highly recommend)
I have 13 Ubuntu 18.04 server running plex. When I began sharing my plex library other users, I ran into an issue where two friends sharing the same plex server were watching the same TV show, and it was messing up the watched/unwatched for each user. I now have a separate Plex VM for each user with whom I am sharing. Those VMs are allocated with 4vcpu, 1GB vmem, 200GB vdisk, with each VM limited to 2 remote streams (capped at 4Mb 720p). All of the Plex VMs are also running PlexPy to give me a little more visibility into Plex performance
Hardware location:
My crawlspace. I have two dedicated 15A circuits with an APC 1300 UPS on each, with 2 ESX nodes and 1 NAS plugged into each.
2019 plans:
Array reconfig
I have 6x6TB HGST drives on the array to copy the data on the 9x8TB array to in preparation for:
I am expanding my main array to include all of my media, instead of my current 9x8TB and 9x4TB arrays. In my current array setup, I would need 9 more drives to expand my array of 4TB or 8TB drives, which was poor planning on my part. My plan is to copy all of the data on my 9x8TB (~50% full) to the 5x6TB temporary array, then recreate my main array as a ZFS array with 5 devs of 3 drives each. This will provide me with two important things: 1) I can tolerate 5 drive failures without loss of data, and 2) I can now expand the array just 3 drives at a time. I believe I will end up with about 72TB of usable space.
My plan is to then use the 6TB drives and 4TB drives in an unraid box at an offsite location with a complete copy of my data.
Physical Hardware:
ESX1 - Xeon E3-1265L v3 / 16GB ECC UDIMM / 80GB S3500 SSD (OS) / Intel X520-DA2
ESX2 - Xeon E3-1265L v3 / 16GB ECC UDIMM / 80GB S3500 SSD (OS) / Intel X520-DA2
ESX3 - Xeon E3-1265L v3 / 16GB ECC UDIMM / 80GB S3500 SSD (OS) / Intel X520-DA2
ESX4 - Xeon E3-1265L v3 / 16GB ECC UDIMM / 80GB S3500 SSD (OS) / Intel X520-DA2
NAS1 - Norco 4220 / Xeon E5-2630L v2 / 128GB ECC RDIMM / 80GB S3500 SSD (OS) / 5x Fusion-IO 1.2TB (RAIDZ NFS share to ESX) / Intel X520-DA2 / H310 IT v20
NAS2 - Norco 4220 / Xeon E5-2630L v2 / 128GB ECC RDIMM / 80GB S3500 SSD (OS) / Array1 - 9x8TB RAIDz2 / Array2 - 9x4TB RAIDz2 / Intel X520-DA2 / H310 IT v20 / H310 IT v20 / H310 IT v20 / Intel X520-DA2
Brocade ICX6610-48 with full licensing (courtesy of STH forums)
ESX HA Cluster:
ESX1-ESX4 are in a HA ESX cluster of four nodes, running around 20 VMs (detail to follow). I can safely tolerate a 2node failure with 4hosts. All storage is on NAS1 across the Fusion-IO array. NFS speed tests around 1GB/s R/W for my VMs.
I utilize a VMUG Advantage yearly subscription for licensing.
I run a virtual vcenter server, but it does not run inside the HA cluster, instead running on NAS1 which is actually a ESX box with a virtual NAS appliance. It is stored on the same Fusion-IO array as the other VMs
VM details:
I have ~5 server 2012 boxes running different services, including DNS, DHCP, AD, WSUS, BlueIris
I have 1 server 2012 box running "applications that let me maintain my plex library", including MediaCenterMaster (highly recommend)
I have 13 Ubuntu 18.04 server running plex. When I began sharing my plex library other users, I ran into an issue where two friends sharing the same plex server were watching the same TV show, and it was messing up the watched/unwatched for each user. I now have a separate Plex VM for each user with whom I am sharing. Those VMs are allocated with 4vcpu, 1GB vmem, 200GB vdisk, with each VM limited to 2 remote streams (capped at 4Mb 720p). All of the Plex VMs are also running PlexPy to give me a little more visibility into Plex performance
Hardware location:
My crawlspace. I have two dedicated 15A circuits with an APC 1300 UPS on each, with 2 ESX nodes and 1 NAS plugged into each.
2019 plans:
Array reconfig
I have 6x6TB HGST drives on the array to copy the data on the 9x8TB array to in preparation for:
I am expanding my main array to include all of my media, instead of my current 9x8TB and 9x4TB arrays. In my current array setup, I would need 9 more drives to expand my array of 4TB or 8TB drives, which was poor planning on my part. My plan is to copy all of the data on my 9x8TB (~50% full) to the 5x6TB temporary array, then recreate my main array as a ZFS array with 5 devs of 3 drives each. This will provide me with two important things: 1) I can tolerate 5 drive failures without loss of data, and 2) I can now expand the array just 3 drives at a time. I believe I will end up with about 72TB of usable space.
My plan is to then use the 6TB drives and 4TB drives in an unraid box at an offsite location with a complete copy of my data.