Oscuro AI (OAI) FAQ's

Although not all encompassing, here are the most frequent questions we receive.

FPGA vs GPU

My vendor says FPGA systems are garbage, difficult to manage, not that fast, not as good as GPU's and we need to learn a lot more skills sets.

A: Your vendor is lying or uneducated. There's no special skills sets required to manage a solution that has on-board FPGA's. Oscuro AI FPGA's are already programmed for the services and require no intervention, modification by clients. Programming FPGA's in with Verilog or VHDL is also no longer a heavy lift either. The performance of FPGA's cannot be denied by vendors who neither support, them nor field them.

Is Oscuro AI and 8A and do you sell to Defense? If not an 8A do you work with a Prime?

A: Oscuro AI aims to bring performant storage to International clients. Although our technology proved tremendous advantage to the War-Fighter in multiple domains ( too many to mention really) we decided the DoD is not for us.

We are neither an 8A entity nor will we seek to be. We do not have partnerships with Defense Primes in the US. We are currently developing Defense partners outside of the US through existing US agreements like AUKUS. The process of selling to US Defense will always be mired in procedural tape, IP risk and other variables which would delay or hinder our Corporate goals.

Why no OS?

Splunk HATE

Your team says Splunk is not a fan of yours, why is that?

A: Because we reduce Splunk footprint by 60%, licensing by 50%, increase ingest by 1000%, and reduce power, cooling, and server count, and they don't like that. Cisco and Vast have partnered NOW to bring performant storage to Splunk, when for many years Splunk employees told customers they didn't need "that much speed".

Do you have an overly complex Splunk environment that's costing you millions per year but can't ingest what they promised? Invite us in to meet with your team and the Splunk team and let's see what happens. Ask them if they will support you deploying Splunk on faster architecture than Vast and get ready for the answer you will get.

Somewhere, somehow, the OEM's forgot that YOU are the customer.

Your team says there's no OS. Why is that? Why not do what the other OEMs did?

A: Because we want to bring performant storage to everyone without the "vertical apps" game all the OEMs are playing now. We prefer you use KVM (closest to BM) RHL6,8 and then run K8's, Hyper-V, VM's with minimal hit in virtualization to the IOPS or throughput.

Basically bring your own tools, leverage your enterprise licensing contracts and build with the tools you ACTUALLY need without orchestrators, storage controllers and other bottle-necks in the vertical application stack the OEMs feed you.

Are you looking for, or open to partnerships?

A: OAI absolutely belives that partnering is key to success. We empower our partners to deploy our architectures in a myriad of industries. Currently, we have a few partners lining up to deploy or architecture to their respective industries and engrained with their technical solutions.

Although we may not be "actively looking" we are always open to unique discussions and collaborations, just email 1 of us.

Certification

My OEM says you're not certified on their platform so why should I buy your gear?

A: OAI doesn't have to certify on any OEM architecture. OAI provides performant storage which is faster and cheaper than the OEM's. All OEM's work with NVME. The difference is this, are the NVM drives they sell you proprietary. If the drives are proprietary then you have locked yourself into 1 vendor. Oscuro uses COTS NVM drives to avoid vendor lock-in. We also utilize X86 Servers with 24 cores. All COTS, thus no need to certify on anyone's architecture.

Never heard of you guy

I asked all my vendors and NOBODY has ever heard of you, why is that?

A: Because your vendors don't want you to know we exist. Cisco, Nutanix, Dell etc they all know who we are. The big OEM's need to exist. Where would we be without Cisco on Networking. But, small, medium and large business don't have the budget Governments have, and they want the capability to stay competitive in the AI race. Our model breaks there's, thus don't mention that team.

The IOPS tool

I don't buy into the high IOPS number you all claim, throughput is what counts.

A: OK, put down the stick. IOPS are the measure and TP is second to that. You're USED to OEMs telling you they offer high IOPS and lying. We don't lie, 18 Million IOPS with sub 4ms latency. We offer both with no vertical application stack to buy, manage or CONTINUE to pay for.