Let’s kick off a community thread to share and discuss our system specs, whether they’re powering our native machines or virtual environments. It’s always intriguing to learn about the diverse setups we’re all using and to hear opinions on what the ideal configuration might look like.
Feel free to jump in with your own specs and thoughts. Here are a few prompts to get the conversation started:
System Specs Showcase: Share the specifications of your current setup. Are you running a native machine or a virtual environment? What hardware and software configurations do you have?
Performance and Usage: How well does your current setup perform for your needs? Are there any bottlenecks or areas where you wish you had more power?
Dream Configuration: If you could build or upgrade to your dream setup, what would it look like? Feel free to dream big here! What CPU, GPU, RAM, storage, and other components would you prioritize?
Real vs. Virtual Environments: For those running virtual machines, what drove your decision to opt for virtualization? How does it compare to running on native hardware in terms of performance and flexibility?
Budget vs. Performance: Balancing budget constraints with performance requirements can be challenging. How do you approach finding the right balance between cost and power?
Tips and Tricks: Do you have any tips or tricks for optimizing system performance? Whether it’s tweaking settings, using specific software tools, or hardware upgrades, share your expertise!
Future Trends: Where do you see system specs heading in the future? Are there any emerging technologies or trends that excite you?
Let’s keep the discussion friendly and informative. Whether you’re a tech enthusiast, a seasoned professional, or just starting out, everyone’s input is valuable. Can’t wait to hear about your setups
I think the power is fine but i don´t know if it would run better when it would be installed natively.
My current setup is a I5-8500T, 32gb ram, and a 1TB m2 ssd. The HAOS vm has 20gb and 4 cores assigned. With 12 watt average i feel this is pretty fair for the performance i get.
I run HA Core on Ubuntu + Docker. I tried several install methods (python, snap, VM) and found this is my preferred balance of performance and flexibility (I am a recovering sysadmin). I have about a dozen containers on this host, mostly for home automation, plus jellyfin.
Current hardware is a Dell Wyse 5070 thin client, 8GB RAM and 32GB eMMC flash. The “Pentium Silver” J5005 CPU is easily 2-3x faster than a RPi4 for less total cost ($100) while still ~4-6W idle (here in SoCal even a 12W always-on server can run $35+/yr in electricity). Average utilization is under 5%. This system has been rebooted maybe 3 times in 2 years, most of those because of my accidents/mistakes.
In general, VMs are a lot of overhead and duplication of services for a lightweight machine; they are better suited to enormous servers with lots of CPUs serving lots of different applications or users across different OSs. Containers are a good compromise offering some security isolation with easy upgrades. That said, I keep a Linux VM on my Mac mini for docker and tinkering.
While not specific to HA, video is usually the outlier that can quickly drive up costs. I have a number of cameras I feed into HomeKit (via Scrypted) and the J5005 did not do well rebroadcasting multiple ffmpeg feeds. While I have no desire to record 24/7 or detect people or objects, the trend is certainly toward CPUs with more hardware accelerated “AI” capabilities in this arena.
My main advice is to keep it simple and invest in your network and smart devices more than your server. HA is an amazing piece of software with remarkably low HW requirements. No amount of overbuilding the server will make up for cheap WiFi or spotty Z-wave networks. And before setting off to build a massive video surveillance system just because you can, carefully consider your requirements/risks vs costs for the processing, storage, cameras, wiring, and electricity involved.
1: Main host is a Dell Poweredge R740xd with a Xeon silver 4110, 128GB RAM and some SSDs and 10Gbit networking, running ESXi 8.0.2. Secondary host is an old Intel NUC Skull Canyon running ESXi 8.02 as well , I use this when doing maintenance on the main host. Network is mainly 10Gbit Ubiquiti. As storage devices I have a QNAP 1282T acting as my main storage and two older QNAPs (TVS 863 and 853 Pro) that I use for backup storage.
2: it performs very well. Backups peak at around 1GB/sec.
3: I’m happy with my current config, got enough compute and storage for years to come.
4: Flexibility, ease of management, zero downtime with live migrations, not wasting compute resources, ease of backups (whole machine), snapshots
5: My main server consumes less than 150W at peak so it’s really no big deal
6: If you use Zigbee in a virtualized environment, go for one of the ethernet based coordinators. That way you don’t need to lock your HA instance to the hardware with the USB stick installed.
7: My main worry is that after the Broadcom takeover of VMware I might need to move to another hypervisor, they are changing the subscription models drastically.
My Home Assistant server is a tiny PC Lenovo M72e it has a pretty old Intel Core i3-3220T processor, and 8 GB of DDR3L RAM and a new 100 GB SSD, with about 700 entities, 4 cameras, 16 add-ons, running non stop for about 1,5 year it has not crashed once.
The processor use is at 10%, ram use at 30% and disk use at 50%. That is still plenty enough to get you for another 3 to 5 years of using this equipment and not have to upgrade. The cost was around $75.
Right now I don’t think I’ll need to upgrade for the next couple of years.
I went with HAOS, I also have a second tiny PC that has OpenMediaVault for disks with movies and tv series that I watch via Kodi and using portainer I have pi-hole, qbittorent etc. I think for a server running 24/7 the HAOS way is the perfect fit for me.
I think you should also consider power consuption for your server, not only budget and performance. My Lenovo M72e consumes about 0,25 kWh a day so it hits a sweet spot between cost and performance.
No real trics, just to use USB2 ports for a Zigee/Zwave dongle.
Looking forward to see Thread/Metter evelve, but right now I pretty happy with my Zigbee networks.
wow, that is a crazy powerful setup. On point 7. I decided to buy a lifetime key for 8.0u2 from a key distributor, so I think that could be an option for you as well.
I’m running home assistant in docker.
I have old pc running i3-4130 processor with 28 gb ddr3 ram and 30 gb swap, nvidia quadro p1000 for gpu and google coral tpu.
I have currently 3 ssd with a room for 3 more. I use 512 gb ssd for my os and docker containers and two ssd for backup.
I’m currently running 32 containers.
PowerEdge R820 with dual Xeon E5-4607, 256GB of RAM and 8-8TB SSD SCSI drives in a RAID 10. 8-1Gbit network cards in a team. I use the server for Jellyfin, Sonnar, SABNZBd and other applications. HA is running on a virtual machine set with 12 cores, 32GB of memory and 100GB hard drive.
I have zero issues with perfomance and haven’t noticed any kind of slow downs for anything I do on it.
Kind of the setup I have now just newer server. The R820 that I have is from my companies old servers.
I always like virtual because they are easy to move between hardware and backup fully.
My setup was very budget for me because all the hardware was free
