Page cover
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Our Distributed Network

A drive has to answer immediately: list this folder, sort by date, count what you have used, rename, move, restore. A storage network has to keep bytes and return exactly those bytes to anyone who asks with the right fingerprint.

NeoDrive gives each job to the layer that can do it.

The index holds the structure. The private IPFS network holds the content.

Two layers, one product

You
  │  folders, names, drag and drop
  ▼
NeoDrive
  │  tree, quotas, permissions
  │  encryption in the browser
  ▼
Index                         Private IPFS swarm
names, folders, dates         encrypted bytes
favorites, trash, quotas      pinned on authorized nodes
owner, and the CID      ←→    served back by the accelerator

The index knows who owns the file, which folder it sits in, what it is called, how large it is, and the content identifier that points at the bytes. It does not hold the document.

The swarm holds encrypted blocks. It does not hold your file names, your tree, or your organization. A node that stores the block stores ciphertext, pinned so it stays, announced so the rest of the swarm can find it.

That separation is what makes a one-gigabyte rename instant. The index changes the name or the parent folder. The CID does not change, so nothing is written again on the network. The mutable facts of a drive sit above an immutable content address.

Upload

  1. You drop the file. The browser encrypts it, with a fresh initialization vector, so this copy is unique.

  2. NeoDrive streams the ciphertext to a Kubo node on the swarm. The node chunks it, stores it, and pins it. The application streams the body through. It does not buffer the whole file in memory on the server.

  3. The index records the file: name, folder, size, type, and CID.

  4. Within about half a minute, the node’s announcer reads its pin list and declares the new CID to the Private Indexer. From that moment the swarm, and not only your account, knows where the content can be read.

You see a progress bar, then a file in the folder. The four steps stay on our side of the screen.

Download

  1. You open the file. The index resolves its CID and checks that it belongs to you.

  2. NeoDrive redirects your browser to the Swarm Accelerator.

  3. The accelerator asks the indexer which nodes hold that CID, in score order. It fetches from the best one and, if that node is slow, from the next. A recent read may already be in cache.

  4. The browser decrypts. The document is readable again on your machine.

NeoDrive does not proxy the download. It decides whether you may have the file, then lets the gateway and the nodes move the bytes. The account layer stays small. Throughput stays with the network.

Delete

Deleting a file updates your account immediately. The row leaves the index, the quota drops, and the trash behaves like a trash. On the node, the pin is removed. Disk space returns when Kubo garbage-collects unpinned blocks, which is how IPFS repositories reclaim space. The indexer drops the announcement when its lifetime expires, unless the content is announced again.

What you see is the quota and the folder. The network catches up on its own clock, and your account does not wait for that clock.

A private swarm, on purpose

Public IPFS is an open network with a global directory, the DHT. That directory is the right tool for public content. Company files need a network with an admission rule.

The Neova swarm is private:

  • a swarm key that outside nodes do not hold;

  • public bootstraps removed, and the public DHT left unused;

  • a directory of our own, the Private Indexer, answering in one request;

  • a read path, the Swarm Accelerator, that ranks nodes by freshness, latency, and success.

Operators join this swarm. Their nodes announce to this indexer. Your files are not published on the public IPFS network.

Who does what

What you need

Who does it

Open a folder, sort, search, see a quota

The index

Rename or move without uploading again

The index

Keep the bytes

Nodes on the swarm

Know which node has a CID

The Private Indexer

Download it quickly

The Swarm Accelerator

Keep the document unreadable on a node

Encryption in the browser

Keep random nodes out

The swarm key

Last updated