What level of routing decisions do you expect to be made on a single / dual ethernet drop to a server card with no other ports?

Like this is cute and all, but kind of a weird target market.

Bunch 'O VLAN's to guest VM's without that routing consuming CPU cycles on the host hardware. RouterOS has a full API so you could automate everything.

The card actually has 4 "ports". 1 virtual port dedicated to management (via PCIe passthrough), 1 GBe port, and 2 SFP28 cages. Plenty for a cloud hosting provider. Both SFP28 cages, plus the virtual 1GBe port support passthrough via PCIe to the physical server.

That's just one use case. Another is a dedicated firewall for the actual physical server that's powering the thing, running at a full 25Gbps consuming zero CPU time on the server itself and zero U's of space... all for less than $199 street price. The card supports PCIe 8x passthrough, so the physical connection to the server is 64Gbps - way more than a single SFP28 cage can support.

Pretty slick...

And yet none of this is actually attractive from a cost perspective at this rate compared to other options that would actually be used in the same space is my point. Everything I can think of to do with this card can be done better and cheaper somewhere else.

And yes, I've done this stuff at very large scales.

Cheaper than $199 (probably around $175 street price is my guess), 0U's, no licensing or reoccurring fees, and consumes zero host resources? Even as just a physical firewall or hardware vpn endpoint in front of a colo'ed server, using one SFP28 in full passthrough... this thing sounds incredible.

I think people that feel this device is pointless just have not been in a situation where this device is exactly what they needed. It is a niche device, admittedly.

However, you have my attention - what device are you referring to that does this better and cheaper?

This is all better done with ACLs or direct programming on a broadcom chipset in a traditional scale top of rack switch or similar, or better done in the actual host CPU land because its something complicated that can't be done with simple ACL / TCAM programming.

Seriously, there isn't a good problem here that this solution solves that isn't better done elsewhere.

Top of rack requires space, which must be paid for monthly in a colo'ed environment, or reduces usable space (ie. revenue generating space) for hosting/cloud providers.

Done in actual host CPU consumes resources.

Direct programming on a chipset is a lot more complicated than running an off the shelf router.

Yes this is a niche product - but your proposed alternatives are silly in most cases where just slapping a zero U device into an unused PCIe port is far less complicated, less expensive, and easier to maintain.

This thing costs less than $200, a one time cost. Just reading the Broadcom documentation alone will cost your organization more than that...

So you don't plug servers into an existing switch? Do they just connect via magic or something?

New ones do, but 5 year old PCIe3 ones, such as the SolarFlare 7000 series which are still supported with drivers by Xilinx can be had for around $30 off ebay for home use.