Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Dear Datacenter owners

RURKRURK Member

Dear Datacenter owners, I am reaching out to you as a trusted expert in data center infrastructure. I currently have a project that requires deploying a specialized cluster (RedTeam research) and I would like to discuss the possibility of hosting it in your facility.

The cluster consists of three servers with the following hardware specification:
CPU: 2x Intel Xeon E5-2697 v3 (14 cores / 28 threads each) → 56 cores, 112 threads per server
RAM: 256 GB DDR4 per server
GPU: NVIDIA RTX 5000 24GB GDDR6 per server
Storage: NVMe SSD ≥ 2 TB per server (for fast I/O)

In addition to the hardware, we have specific networking requirements:

Internal connectivity: A dedicated 10 GbE switch (copper or fiber) connecting the three servers. We would like to manage this internal switching infrastructure ourselves – meaning we need full administrative access to the switch or at least a private VLAN with our own routing and ACL capabilities.

External access / test environments: We need a gateway server or a software-based NFGW (Network Function Gateway) that will act as a controlled egress point. This gateway must allow the cluster to reach the internet and/or other target environments (for legitimate testing purposes) under our strict control – with the ability to configure NAT, firewall rules, and dynamic routing on the fly.

We are prepared to handle all configuration and maintenance inside the cluster. From your side, we would need rack space, power, cooling, basic L2 connectivity, and the possibility to interconnect our own L3 switch (or a dedicated VRF). Also, assistance with provisioning an external gateway (e.g., a public IP address for outbound traffic) would be appreciated.

Could you please evaluate the feasibility of this setup? I would also be interested in approximate timelines, rack rental costs, bandwidth pricing, and the options for a managed 10G internal network.
And for cheap of course.

Comments

  • Ian_Dot_TechIan_Dot_Tech Member, Patron Provider

    Hello,

    PureVoltage operating multiple datacenter suites, can assist with this setup. We have years of experience with tailored setups.

    I will shoot you over a DM, I hope to work with you.

  • @RURK said: RedTeam research

    Or more accurately: abuse

  • AlyxAlyx Member, Host Rep

    I'm really curious of what kind of "RedTeam research" requires this setup 🤣

    Thanked by 1averagedatahoarder
  • itzsenuitzsenu Member
    edited August 23

    "redteam research" requiring your own l3 switch, full nat/firewall control, and a controlled egress gateway ? ( asking for 3x dual xeon servers with rtx 5000s, dedicated 10g switching, and full administrative network control ). thats c2 infrastructure and "for cheap of course" asking for nation state lab infra at homelab pricing.

    Just ask what that RedTeam research even is. Nice budget

    Thanked by 2sbenchid tux
  • DonnMoDonnMo Member

    dear diary

  • @RURK said:
    The cluster consists of three servers with the following hardware specification:
    CPU: 2x Intel Xeon E5-2697 v3 (14 cores / 28 threads each) → 56 cores, 112 threads per server

    Can someone explain how this 2x math works?

  • @TimboJones said:

    @RURK said:
    The cluster consists of three servers with the following hardware specification:
    CPU: 2x Intel Xeon E5-2697 v3 (14 cores / 28 threads each) → 56 cores, 112 threads per server

    Can someone explain how this 2x math works?

    It works the same as 9+10=21

  • Another contender for top 10 most suspicious LET posts of 2026 :p

    But assuming it's legit, 24GB VRAM + 256GB DDR4 doesn't seem like an ideal combo for AI. Usually CPU offload demolishes pp/s & tok/s. And if you're planning a shared model across all three of these over the network links (even 10gb/e ones) it will be even slower. But maybe it could work out with the right mixture of experts model. Let us know what kind of speeds you can get out of it.

    Thanked by 1unsafetypin
  • forestforest Member

    @WyvernCo said: But assuming it's legit, 24GB VRAM + 256GB DDR4 doesn't seem like an ideal combo for AI.

    Could be for hash cracking. That doesn't need much VRAM at all.

    Thanked by 1WyvernCo
  • RURKRURK Member

    Right now we need 90GB vram and 250GB ddr5

  • The way this post is worded is very odd

  • @RURK said: 250GB ddr5

    250gb RAM for each server?

  • RURKRURK Member

    @JasonM said:

    @RURK said: 250GB ddr5

    250gb RAM for each server?

    yes, correct

  • @RURK said: Right now we need 90GB vram

    @bbmmsvr4u had 16gb or 24Gb as i remeber.....

    Thanked by 1bbmmsvr4u
  • ehabehab Veteran

    only @emgh can perform inspections ... he will require drawings first.

    Thanked by 2barbaros emgh
  • it's a relevant ask, hopefully it's relevant pay too

  • JoshRJoshR Member, Patron Provider

    @RURK Budget? Location? How many IPv4? IPv6?

  • SaragoldfarbSaragoldfarb Veteran, Megathread Squad

    @Alyx said:
    I'm really curious of what kind of "RedTeam research" requires this setup 🤣

    I think it's a typo and it was meant to say "RedTube".

    Thanked by 2Alyx tux
Sign In or Register to comment.