What shipped.

Comparison pages, use-case pages, guides, video watch pages and this changelog

Eight comparison pages built from vendors' own published pages with retrieval dates, five new solution pages, six guides and explainers, a proof-of-concept page, watch pages for both videos with transcripts and chapters, and a machine-readable site map at /llms.txt.

Compare alternatives →Guides →Videos →

Browser uploads write straight into the package bundle

Files uploaded from the browser are written directly into the package's bundle at their final offsets as batches arrive, instead of into one staging file per uploaded file. On a 420,000-file folder the finish step went from about five minutes to under ten seconds.

Large directory transfer →

Six upload batches in flight, and a steadier speed readout

The browser keeps six upload batches in flight instead of three. On the same 420,000-file folder and the same home connection, the measured upload rate rose from under 1 MB/s to between 3.7 and 5.4 MB/s. The transfer panel now updates its current speed once a second over an eight-second window instead of several times a second.

Node addresses come from the node registration only

A node no longer needs to know its own public address. Transfer specs, Web Edge URLs and diagnostics use the URL registered in the control plane, so a changed IP needs only that registration updated, or nothing at all when nodes are registered by DNS name. The admin connectivity check warns if a node still advertises a different host, and its hints no longer assume any particular cloud provider.

Administration and nodes →

Every file in a folder is read exactly once

Top-level folders are assigned to browser read workers by name rather than by position, which fixes a case where a folder could be skipped when the listing order changed between workers.

Folder reads run in Web Workers; cheaper per-request checks on the node

The browser reads a selected folder in four workers in parallel and sends each batch as a single request body, keeping the page responsive during very large uploads. On the node, the per-request origin check reads a session summary instead of loading the whole session, which removed the main per-batch cost for sessions with hundreds of thousands of paths.

Packages uploaded from the browser are stored as one bundle

Multi-file browser uploads are stored on the node as a single bundle object, the same format RIPTON Desktop uses, instead of one file per upload. File-heavy packages are read sequentially on download and no longer create hundreds of thousands of small files on the node.

Large directory transfer →

Archive downloads read from local storage with read-ahead

When the edge node is also the storage node, archive downloads read files from local disk with bounded read-ahead instead of fetching each entry over HTTP from the node itself. On a cold cache, archive builds on our test node went from about 1,000 files per second to about 2,800.

Download progress polls are a few hundred bytes

The progress endpoint used by the transfer panel returns session counters instead of the full path list. For a 420,000-file download that response was 81 MB per poll; it is now under a kilobyte. The control plane reconciles Web Edge download sessions from the same counters.

Storage cleanup runs in the background with progress

Deleting storage items on a node with over a million files no longer times out. The node runs the cleanup as a tracked job, the admin page follows its progress, and directories emptied by the run are removed in the same pass.

Large packages complete correctly after upload

Packages with more than a thousand files could stay in 'uploading' after a successful Web Edge upload because file states were updated one page at a time. Completion now marks every file in one statement.

Streamed downloads no longer stop at 60 seconds

Long archive and single-file downloads through the edge were cut off by a server write timeout. The timeout is lifted for streamed responses.

Browser downloads appear in the transfer panel

Archive and direct-file downloads handed to the browser show as rows in the transfer panel with progress and a rate, polled from the edge node with the download ticket, instead of only in the browser's download list. The edge's own storage fetches are exempt from the public rate limiter.

Batched small-file uploads

Browser uploads send small files in batches of up to a thousand with one bulk lookup on the node per batch, replacing one request and one query per file for file-heavy folders.

Licensing: permanent Free plan, 14-day trials, offline activation, tier-change preflight

The control plane issues permanent Free licences and 14-day trial licences, accepts offline licence activation for air-gapped installs, checks a tier change before it is applied, and counts licensed standby node capacity for failover.

Pricing →

Pipelines, reusable schedules, SIEM export, node failover

Pipeline orchestration APIs, reusable transfer schedules, audit event export to SIEM targets, and high-availability node failover, each with its admin console page.

Automation →