April 14 outage: what happened, and what’s next

At 7:15 a.m. PT on April 14, one of Webflow’s CMS database clusters silently exceeded an undocumented capacity limit on its cloud provider’s engine. Although the provider’s console showed only 1.74 TiB used—about 1.35 % of the reported quota—the engine had internally reserved space for roughly 66 million files, eventually filling the full 128 TiB allocation. When engineers tried to reboot the node, it entered a crash loop because no logical storage remained. The failure cascaded across the platform, rendering the Dashboard, Designer, hosted sites, form submissions and APIs inaccessible for the majority of customers. Webflow’s rapid response—publishing a status page within nine minutes, convening a crisis call, and removing the offending shard—brought 96 % of traffic back online by 9:21 a.m. PT, with the remaining 4 % restored by 9:01 p.m. PT, and no CMS data loss reported.

The incident underscores a persistent challenge for SaaS and no‑code providers: hidden infrastructure constraints can undermine reliability even when surface metrics appear healthy. Webflow’s reliance on a single cloud vendor’s database engine mirrors the architecture of rivals such as Wix, Squarespace and Shopify, all of which must balance rapid feature rollout with deep observability. The lack of visibility into logical storage consumption reflects a broader industry trend where abstracted cloud services hide critical limits, prompting companies to build supplemental monitoring layers. Webflow’s post‑mortem reveals that its prior alerting only tracked reported usage, not the underlying allocation, a gap that many fast‑growing platforms have only recently begun to address.

Going forward, Webflow is expanding its shard count, upgrading the database engine to raise the hard limit to 256 TiB, and installing circuit‑breaker logic to isolate failing shards. These measures should mitigate the risk of a single node throttling the entire service, but they also raise questions about scaling costs and the complexity of managing a more fragmented data layer. Customers should monitor Webflow’s status page for any lingering latency or partial failures as the new sharding scheme rolls out, and watch for any impact on API rate limits or backup windows during the transition. The episode also serves as a reminder that even well‑funded SaaS firms can be blindsided by undocumented cloud limits, making transparent provider‑level SLAs and internal telemetry a competitive differentiator.