Build Advice Any overkill or bottlenecks for data processing rig

[rob2019]

Hello. Thank you for taking the time to read this.

This is my first build and I am not very knowledgeable of hardware.

I would like to build a machine that can handle about 4 VMs and can be used for intensive data processing. The machine will not be used for games or video editing. Only using 1 monitor mostly, maybe 2 at most.

Here is what I am considering so far:

CPU: Intel i5 9600k
MB: MSI MPG Z390 Gaming Plus
RAM: Samsung 32G DDR4 2666Mhz (16 x2)
Graphics: Intel HD Graphics 610 (Do I need something better? This is the default with the box.)
SSD: Samsung M.2 970EVO Plus (500G x 2) (May opt for just one and use external drive for storage)
PSU: Antec EAG PRO 750W 80+ Gold
Cooler: DEEPCOOL GAMMAXX 300 FURY
OS: Windows 10 Pro
Monitor: LG 24inch

This machine is not for games, but since the processing could be very intensive, I choose a CPU and MB with game specifications. Is the RAM speed sufficient? Do I need to go with G.Skill 3200Mhz? Probably won't need to overclock much if any, but with multiple VM instances running, I want to get the best performance as possible.

Since VMs can't access the graphics card anyway, I was thinking the standard HD Graphics 610 card will be enough. However, will having this graphics card degrade the performance of the other hardware?

I think there might be some overkill in my setup, but I am mostly concerned about bottlenecks, like the graphics card issue.

The setup above in usd would be about 1350. Would like to keep it below 1500.

Thank you for your time and consideration.

 

[kanewolf]

For VMs and data processing, an i7 with hyperthreading can be beneficial. A 750W power supply is not required. A 500W would be plenty.

 

[rob2019]

For VMs and data processing, an i7 with hyperthreading can be beneficial. A 750W power supply is not required. A 500W would be plenty.

Thanks for your reply. After I sent my post, I was just thinking that I should go with an i7 8700 instead of the i5 9600k. Do you think I need to get a separate graphics card? Or is the integrated hd 610 graphics unit enough? Thanks for your time and advice.

 

[faalin]

What kind of data processing are you doing? what software are you using. Is this just a side project or are you making money off it? how important is the data you are using?


After the core unit you have enough resources to maybe allocate 1-2 VM's

At a minimum for the VM you will need 1 CPU core, 4-8GB of ram, and 128GB of HDD space. If your running windows 10 on the VM's you will also need a license key for each VM.



It may be cheaper to pick up a second hand HP Gen8 server. Last year i picked up a HP DL380e Gen 8 server, dual e5-2470 cpu giving me 16 cores 32 theads, 32GB of ram, 2 250GB SSD in raid 1 for OS and 8 3TB drives in raid 6 for storage. In all it was still under your $1500 price limit and you could save money by not getting 24TB of storage.

 

[rob2019]

What kind of data processing are you doing? what software are you using. Is this just a side project or are you making money off it? how important is the data you are using?

Thanks for the advice. No money involved at all. Basically it's a lab set up to get data sets like logs off the network, process them, build and add to dbs and then send results to the network. It can be intensive, but not intensive 24/7, only maybe 2 hours total a day.

After the core unit you have enough resources to maybe allocate 1-2 VM's

At a minimum for the VM you will need 1 CPU core, 4-8GB of ram, and 128GB of HDD space. If your running windows 10 on the VM's you will also need a license key for each VM.

I could do the jobs with just 1 or 2 VMs, but I like to separate job types (1 VM = 1 particular job), so I was thinking that I might use more.

VMs will be running Debian non-interactively.

It may be cheaper to pick up a second hand HP Gen8 server. Last year i picked up a HP DL380e Gen 8 server, dual e5-2470 cpu giving me 16 cores 32 theads, 32GB of ram, 2 250GB SSD in raid 1 for OS and 8 3TB drives in raid 6 for storage. In all it was still under your $1500 price limit and you could save money by not getting 24TB of storage.

That's a good idea, and if I need to scale more in the future that seems like a good route to take. Basically, I want to keep it simple for now because I don't have a lot of experience. My plan is to run Windows 10 Pro and do my normal activities while running the VMs run on Hyper V.