I want to build a file server to move my company away from Dropbox. We run our entire company through Dropbox and file explorer. We currently have 20 office employees that need access daily to do their jobs. There is a total of 9TBs of data and approximately 1.6 TB of data that is shared withal 20. For the 9TB of total, only 3 people need access to it.
What is the minimum hardware requirements that are necessary to replace Dropbox with a server?
Cores, threads, generations, DDR4 or 5
Also, any recommendations on OS, Hypervisor, or alternative ideas are welcome
Nobody can answer that without pulling it out their a**. Well except dropbox engineering staff, and their info would be useless anyway.
You should be asking what kind off-the-shelf software or open source solution you are going to use and then looking at their documentation for sizing guidance.
You do know what you are going to use, right? Or is your org yolo enough to develop something in house ?
If havent reached that step yet, stop now and start researching. You are two steps ahead where you should be and without a plan
Is a sure way to shoot yourself in the foot and knee.
Are all the employees local or are some of them remote/behind a VPN?
If it is mission-critical (your OP makes it sound like it is), then something that can gracefully handle a controller and drive failure. Synology SA3400D - Budget option
Pure X20 - Needs some additional server(s) to actually be a NAS
If you know your way around M$ Server, then look into DFS. In case MS Server is new to you, ignore this entirely.
Is there a video series or book/anything that you guys recomend that teaches the basics? I am having a hard time understanding what is even being talked about. (Example- When installing NIC cards I had to figure out what a Subnet Mask was and how to use it. I never saw that building pcs previously.)
Thank you for these. I started looking into them last night.
You are 100% correct. I am on this journey currently. That is why I asked about the hardware, because I am trying to build a homelab that will mimic the new office file server. Any homelab videos/links you recommend?
Your assumptions are all correct. I am just getting into servers. I have built over 20 personal Pcs from parts and have a good understanding of hardware, but servers are completely new to me. My experience gave me a false sense of confidence moving to the enterprise world. Servers are on a completely different level than personal PCs. Just setting up 40GB NIC card was an experience.
Seems like I aimed just a little bit higher.
We would need remote access and no vpns
I appreciate you guys not laughing me out of the forum and giving me a starting point for this journey. I really like this community and hope to one day contribute.
There are definitely many pitfalls when it comes to servers, so homelab a lot.
For example you may build a file server with RAID and use SMR disks - everything looks fine for years, but then you get a disk failure and your rebuild takes too long and you lose another disk and poof… data gone.
There is a lot of affordable homelab options nowadays, older enterprise hardware, old hardware you have around your house, cheap low power SBCs or any combination of those.
Build something, than yank one of the disks and see what happens, yank that crappy 20$ PERC H310 you flashed with IT mode LSI firmware and try to recover, see what happens with 10$ knockoff Intel NIC, DIY a smart home server, figure out Docker, build, break, repeat…
Yes it is.
And while all sort of norms and best practices require your company to have backups, guides, handbooks and manuals, the real test is for the apprentice to pull the wrong cable accidentally on a Friday afternoon.
No, by reading in between the lines what you asked, you are not yet remotely ready deploy something of this scale on your own. If its meant to be production grade. If its its meant to be shit and giggles deployment, ignore criticism parts below.
Your stated goal is to replace third-party service with on premise hardware and software stack.
You will have to research what you actually need in high level functionality wise, then find what can fulfill those and finally then architect real world solution around that.
Within budget or constraints of existing infrastructure, if any.
You are skipping all the above and jumping to last last and least import point, hardware. If you don’t know what you are trying to build, hardware is irrelevant.
As a personal recommendation you need few years as junior in small well lead to get the grasp of basics or internalized understanding of all above. Seeing entire system architected or at least spec docs and them turning them into reality is critical to understanding this process.
And it definitely is a process.
Eagerness is good, but there are no magical courses that condense years of wide spread learning into a book or two.
There should be, but aren’t. IT is still too young and dynamic field for that, we really could use programs churning out actually certified system integrators like you need to be for task above.
EDIT: removed so many f*** typos it wasn’t funny.
