I've not seen a thread ripper, too expensive...
And a couple of my build:
These are from when I ended up buying quite a few different x399 boards to try and find one that wasn't a mess for esxi when threadripper first released.
Some more processors, rome is blue, naples green:
Then the rack where they live at the moment, from top to bottom in this pic dual epyc 7452 rome 192gb of ram per processor each one only has one processor seated at the moment and space for a ton more ram, this is the ten year plan and has huge scope for upgrades, second is the same, 3rd is a darktrace appliance which is a network based artificial intelligence appliance, ill show you a screenshot later, its properly cool, then there is a dedicated backup appliance which is a storeonce 4500 24tb, it has an identical sister appliance at the colo. under that are a pair of FC Brocade switches and finally another EPYC chassis. I probably have a picture of the disk infrastructure as well somewhere. Put simply this is scalable infrastructure for the back end stuff. Disk wise it's connected to around 80 odd 10k rpm 600gb spindal disks in an EVA (enterprise vault array) which to be honest is the biggest bottleneck at this point and which another disk failed in again today. luckily 2 were delivered today and another 3 will turn up tomorrow so I am set for a few more failures, they tend to go on rebuilds when you put a load of extra stress across the disks, especially in larger disk groups.
Cant find any pics on the disk arrays but here is an example of one of my disk groups:
The DMS group has a failure atm but the setup can handle multiple simultaneous failures on any given tray at any one time up to a maximum of around, well, it's complicated and depends on provisioning how close the disks are to one another in physical proximity and some other variables but it is much more resilient than a traditional raid configuration. I have experienced 3 simultaneous failures on the same group same night, that was on a group of around 48 disks and no data loss, no downtime. EVA even today is a very good storage solution.
Inside an Epyc chassis, none of my chassis have any storage save for an SD card, storage is provided over 10gb FC.
Just for good measure, some darktrace active threat analysis and network traffic analysis of my network right now, I have had to heavily chop the images but now when I say it's cool I am sure you will agree:
All the little dots on the right hand side of the firewall are mostly people home working over the last 24 hours. I have multiple vlans/lans over multiple sites so the boxes teal and different shades of blue in the background are the additional fortigates located at places like my house and are shown inside the firewall. The main firewall is a fortigate as well with all of the good stuff enabled.
This basically works off of the back of my core switches which replicate traffic on all ports back to the appliance so the AI can work it's magic, it has active threat management so can stop traffic before it does any damage.
Caught it at a pretty good time as the network is running its backup and replication routines until around 5am.