Caudena released some of its enterprise capabilities within MCP last year, and made them available to the consumer market this year at https://mcp.caudexcatena.com/.

We have received very good feedback from clients across the enterprise and consumer segments, they really like it.

Currently we don't have such plans.

Are you from CA or CT? :)

FBI and Europol will work with the same forest (unless they are using on-premise setup), but with different "patches".

Because we are on the 'unsafe' territory. And Rust doesn't even have a defined memory model. Rust is a little bit immature. We have some other services written in Rust though.

Thanks for the questions! We don't currently have plans to open-source it. For anything else, feel free to reach out at pa@caudena.com - happy to discuss further there. We'd like to keep this thread focused on the technical side rather than product discussions :)

Yeah, we figured people would compare it to PRISM :)

There are many possible optimizations, but they’re all highly specific to the particular problems you’re trying to solve.

We never targeted weakly-ordered architectures like ARM, only x86. We never used a wide variety of different processors. We are not developing the Linux kernel and are not into control dependencies, just relying on the fences and the memory model. There may be some CPU-dependent performance differences, like discrepancy because of NUMA or false sharing being noticeable on one processor, but not on another. RCU and hazard pointers are nothing new. For the disjoint sets we don't need them. For the forest patches and the tries we do. We are using TBB and OpenMP whenever possible and trying to keep things simple.

No.

"No" what? There are cases where it is traceable if someone uses it in a certain way, but if you do everything by the book, it's untraceable.

Ohh, you're absolutely right!

Assuming by undoing you mean splitting the cluster:

A linked list can be split in two in O(1). When it comes to updating the roots for all the removed nodes, there is no easy way out, but luckily:

- This process can be parallelized.

- It could be done just once for multiple clustering changes.

- This is a multi-level disjoint set, not all the levels or sub-clusters are usually affected. Upper level clustering, which is based on lower confidence level, can be rebuilt more easily.

If by undoing you mean reverting the changes, we don’t use a persistent data structure. When we need historical clustering, we use a patched forest with concurrent hash maps to track the changes, and then apply or throw them away.

We use a single instance for all clients, but when one CFD server processes new block data, it becomes fully blocked for read access. To solve this, we built a smart load balancer that redirects user requests to a secondary CFD server. This ensures there's always at least two servers running, and more if we need additional throughput.