Infra Downscaling | |
---|---|
Participants | |
Skills | Infrastructure |
Status | Planning |
Niche | Hardware |
Purpose | Infrastructure |
Tool | No |
Location | |
Cost | |
Tool category |
Maximize-Efficiency.jpg {{#if:No | [[Tool Owner::{{{ProjectParticipants}}} | }} {{#if:No | [[Tool Cost::{{{Cost}}} | }}
The space currently uses a lot of old, power hungry hardware and could be optimised by the addition of a small amount of money to refresh the infrastructure. This page stands to collect the ideas on how and what to do with performing this task.
Proxmox Servers
These systems take a lot of power (3 pieces of about 200-400W a piece) (confusion , precious , anotheroneikeepforgetting )
Usage
We run a copious amount of virtual machines on our Proxmox cluster
Replacement
Not sure if this is totally viable, but :
This quadcore CPU has full VT extensions, but is just 10W TDP! For 99% of our VMs, this is probably already a sufficient host. Motherboard has a max of 16GB (in 2x SODIMM DDR3 1866mhz)
Questions
- How is the reallife performance?
Netapp
The Netapp storage has been identified as a power consuming piece of hardware. It is also FAILING STEADILY so this is important to upgrade!
Usage
This machine is primarily used for NFS mounts across the infrastructure.
Replacement
- The0 has a 3u case/machine
with 12 disk hardware RAID controller(controller has broken ports) with brackets but without disks in it that might make a good replacement. - We could construct a JBOD and do software RAID - using the an ASROCK c2750d4i is a good starting point but it is kind of expensive.
- Maybe 3 old atom boxen with each 4 disks and something like ceph might also work?
- Or buy a new motherboard and use zaryas 4 port Areca controller
Questions
- What type of FS do we want? (ext4, XFS, ZFS)
- Can we do HA somehow? (ceph?)
- What OS?