Server internal USB ports question

Ok I have a Dell R620 and a Dell R510 that I have been working my ass off to make work. Its been a massive pain in my ass. Truthfully, I want the 620 as a compute node to do video and code, and I want the R510 as a VM host / storage server. Each of these I believe have an internal USB port. I am watching an LTT video about their new servers they are testing with 1PB of flash and they mentioned you install an OS to a USB stick, and then chuck it in a server to run.

So ok question, would it be a good idea to use a bigger USB stick? My 620 has an SD card slot… can I use that? Can I boot off the SD or USB and use the other medium for usr/bin?


Wouldn’t bother with the SD card, those things like to die so bad

If your server has a slim optical, I would get a laptop hdd caddy and throw a 2.5 SSD in it

Those USB ports were also used for HW license keys

The internal USB ports and SD card slot are intended for situations where you have no other use for local drives. For example, to boot-up an ESXi instance that will connect to an iSCSI SAN as its datastore.

If you’re going to have local drives anyhow, there’s every reason to boot-up from those instead. Common USB thumb drives (and SD cards) aren’t intended for full-time use like that. Their performance isn’t great, and you should expect them to fail after a relatively short amount of use.

1 Like

Yes you can use an SD card or USB stick or even usb hard drive/SSD as large as you want to. The only exception is if it is a dual SD module (which is optional on R720 not sure if it is on a 620) I think you’re limited to 32GB cards for the mirror function to work.

Either option works but also relies on flash memory that isn’t as durable as an SSD. When I started out messing with rack servers I used a regular USB stick on an internal port to boot ESXi on my R710 and it worked flawlessly for a couple years all the way up till I replaced it.

1 Like


Just curious because I’m a prick like that, what if I specifically stored like the boot table on it? Do I have in system access to these ports? EX: Save Iso’s to SD and boot off / update from?

My 620 does but I am enough of a shit to actually want to use it. As well, the only caddy I have I actively use with my thinkpad.

If its blue its still USB 3 tho right? I can’t really see into the plug to really check…

Actually in my case its that I have 2 drives that are usable with this machine, it requiring registered dell drives for some damn reason ( I wish I could just chuck 2 normal 320GB SATA6G drives in it grumble grumble ), and at that the only ones I can afford are used 300GB drives that I believe are known to fail, or rando ones wendell sent me that he literally found in the bottom drawer of a desk and was surprised they even worked.

In my case its the fact that an SD card or a USB stick is more reliable than the disk array at all. As GBBoy said above, put a caddy in it to get around the bay issue as it requires a corporate key server I don’t have, being corporate trash. My 510 I can just tell it ignore everything and scan all the hardware every boot and make a whole new inventory. Then it works fine, though I am still learning about RAID so I haven’t really done anything with it yet.

You know what, if I don’t have any other option to get the 620 to boot and I really get into my youtube channel, I’m going to be using my laptop less and will need less eed for a scratch drive, fuck it if I need it I need it I guess.

I only have 4 slots on the one machine, and the other has 2 internal “slots” that questionably work. I worry about useable storage for things, personally, as well as overall reliability. If I am going to get the life left out of the drives I have, I don’t want to burn it on bootups. Waste of potential reads or writes left, thats how I look at hardware.

Spending 8 bucks once every 2 years is a lot cheaper than spending 500 bucks when I don’t have it. Thats the price of having enterprise hardware in a house IMO XD

Joke post incoming about the potential of jbod SD adapters in servers for a big boot drive

Meh, regular backup of the boot disk and spend 5-8 bucks every 2 years?

Big fat meh from me

I love LTT, but based on their previous experiences with data-loss, I’m not entirely convinced LTT is a reliable source of server/storage best-practices :wink:

Why would you want to use a USB drive as a low-performance, low-reliability, single point of failure? No SMART, no TRIM, no guarantees on quality of wear-leveling (same for eMMC).

If you can’t get standard SATA drives working on the Dell SATA controller, sounds like you should be thinking of replacing it with something that works correctly.

You could always PXE-boot it and use root-on-NFS from a LVM2 logical volume on an RPi’s SD card :wink:

Eh you have a point

I am used to power gaming trash…… i may have also just fudged up something

Sounds like you might just need to update the firmware. Dell stopped requiring their own branded drives years ago.

What HBA/RAID controller(s)?

Dunno have to check