Saumya Majumder, lead software engineer at BigScoots, joined the Jukebox Podcast to discuss a recent major Cloudflare outage and how BigScoots designs high-performance WordPress architectures using Cloudflare Enterprise. Saumya specializes in high-performance WordPress engineering, edge computing, worker-based automations, custom caching engines and migration tools. He oversees enterprise customers and develops scalable, developer-friendly hosting solutions.

Why outages happen
Saumya explains that large internet services like Cloudflare are immensely complex systems with many interdependent components. Small, rare edge-case failures can cascade into large outages despite tests and reviews. When a critical lower-layer component fails, higher-level services that depend on it can be affected. He emphasizes that all large providers—Cloudflare, AWS, Google Cloud, Azure—can and do experience outages. These incidents are costly in terms of reputation and service credits to enterprise customers, so providers work hard to remediate and learn from failures.

Cloudflare’s scale and response
Cloudflare is more than a CDN; it provides a broad developer platform and many security and networking services. Because of its scale, fixes and configuration changes must propagate across a very large, global control plane and edge network, which can take time and create a ripple effect as traffic and state stabilize. Saumya praises Cloudflare’s transparency—postmortems that explain root causes and remedial steps—and their ongoing work to remove single points of failure (for example, previous work to remove dependencies on external services like GCP for KV).

How BigScoots mitigated the recent outage
During the outage, Cloudflare’s public-facing services like Turnstile and many websites (including X and ChatGPT) were impacted. BigScoots had automated responses in place: their tooling used Cloudflare’s API to turn off the Cloudflare proxy for affected customers, routing traffic directly to BigScoots’ origin servers until Cloudflare recovered. Because the Cloudflare API remained available, this rapid failover reduced actual customer downtime for sites hosted at BigScoots that normally rely on Cloudflare as a proxy. Saumya notes that such mitigations work when Cloudflare is used as a proxy; sites fully hosted on Cloudflare Workers or using Cloudflare as their origin would still be affected differently.

CDN-level page caching and history
Saumya describes how traditional WordPress caching worked—server-side caches (advanced-cache.php) storing HTML on origin to avoid reprocessing. Static assets were later moved to CDNs, but Saumya and a collaborator pioneered CDN-level HTML page caching (the Super Page Cache for Cloudflare plugin, later acquired). The idea: serve full HTML from the CDN PoP closest to the user, avoiding round trips to a distant origin and drastically reducing latency.

Cloudflare Enterprise and tiered caching
Cloudflare Enterprise unlocks advanced caching behaviors. Key benefits include high cache-hit ratios, longer-lived cache entries, and tiered/regional caching. Saumya explains tiered caching: PoPs are arranged in tiers so a lower-tier PoP (near the user) will first ask upper tiers within Cloudflare’s private backbone if a resource is cached, before querying the origin. This internal Cloudflare intranet (private backbone) is faster and avoids origin requests, lowering load and improving global response times.

Private connectivity and CNI
BigScoots runs their own data centers and has set up direct physical connections (CNI) to Cloudflare, effectively plugging BigScoots’ network into Cloudflare’s intranet. Instead of Cloudflare fetching from origins over the public internet (which can be slower and subject to congestion), Cloudflare can pull content over the private fiber connection directly to BigScoots’ origin. This reduces latency further and improves reliability and performance for cache misses.

BigScoots Cache plugin and platform features
BigScoots developed a proprietary plugin—BigScoots Cache—that integrates deeply with Cloudflare Enterprise caching while exposing fine-grained controls to users and developers:
– Intelligent cache purging: when content is published or updated, the plugin purges the affected page and related pages (archives, taxonomy, authors) automatically.
– Hooks and REST API: advanced users can customize TTLs, purge logic, and integrate cache control into external workflows (e.g., clear cache from a Laravel backend when inventory changes).
– Managed assistance: BigScoots will implement custom snippets or purge logic for customers who can’t or don’t want to configure advanced behavior themselves.
– Portal controls: BigScoots exposes a broad set of toggles for Cloudflare Enterprise features—WAF/login protection, image optimization, Rocket Loader, blocking/challenging bots, geo-blocking, and fine-tuned security rules—so customers can configure or ask support to configure protections and performance settings.

Who benefits
While BigScoots offers advanced features useful to agencies and enterprise customers, Saumya stresses that many smaller site owners also benefit. Controls are presented simply in the portal for common tasks (e.g., block or challenge traffic from a specific country), and managed support can implement changes for less technical customers. BigScoots provides systematic onboarding for enterprise migrations, aiming for zero-downtime migrations and performance optimization as part of managed packages.

Operational mindset and engineering trade-offs
Saumya reiterates that engineering is about managing trade-offs: it’s impractical to engineer for every 0.00001% edge case at all times, so providers prioritize. When outages occur, providers analyze and remove root causes and dependencies to reduce recurrence frequency. He encourages realism: outages are part of running large distributed systems; the goal is to mitigate impact, be transparent about incidents, and learn to prevent similar problems.

Closing
The episode explains how global infrastructure failures can ripple across the web, how Cloudflare’s scale contributes to both the impact and the complexity of fixes, and how BigScoots leverages deep Cloudflare Enterprise integration, private networking, and custom caching to reduce latency and improve reliability for WordPress sites. Saumya emphasizes transparency, continuous improvement, and managed services to help customers at all levels benefit from advanced caching and security features. For more details and links related to the episode, visit wptavern.com and search for this podcast episode.

Welcome to the Jukebox Podcast from WP Tavern, hosted by Nathan Wrigley. In this episode Nathan speaks with Saumya Majumder, lead software engineer at BigScoots, about a recent major Cloudflare outage, how global infrastructure failures ripple across the web, and the WordPress performance and mitigation strategies BigScoots has developed — many built around deep Cloudflare Enterprise integration.

Saumya’s background and focus
Saumya specializes in high-performance WordPress engineering and Cloudflare-powered architectures. He builds large-scale systems including custom caching engines, migration tools, worker-based automations, and edge computing. At BigScoots he oversees enterprise customers, internal WordPress projects, plugins, and developer-friendly solutions aimed at speed, reliability, and scale.

How Cloudflare outages happen (and why they matter)
Saumya explains that Cloudflare is far more than a conventional CDN; it’s a complex platform with many interdependent systems. When a low-probability edge-case bug appears deep in the stack it can cascade into widespread disruption. Even companies with robust redundancy and review processes can miss rare interactions that trigger failures. Outages are expensive: enterprise SLAs can require significant credits when uptime targets are missed, so providers scramble to fix issues and limit impact.

Cloudflare’s transparency and learning
Saumya praises Cloudflare’s transparency. When outages happen Cloudflare publishes detailed postmortems explaining root causes and remediation steps. They also iterate to remove brittle dependencies (e.g., prior outages tied to external services or data center failures) so the same failure modes are less likely to recur. That said, zero failures is unrealistic for any large system.

BigScoots’ mitigation during the outage
BigScoots had an advantage: while many Cloudflare features (such as Turnstile authentication or Cloudflare-hosted apps) were impacted, BigScoots could still use Cloudflare’s API. They automated failovers that toggled Cloudflare proxying off for affected customer domains, routing traffic directly to BigScoots’ origin servers until Cloudflare stabilized. This reduced customer downtime for sites hosted on BigScoots but proxied through Cloudflare. Saumya notes that if you host entirely on Cloudflare Workers or Cloudflare Pages, the Cloudflare outage would be more disruptive because those services and their control planes were affected.

CDN-level page caching: what it is and why it matters
Saumya and a colleague pioneered CDN-level HTML page caching for WordPress long before it became common. Traditional server-side caching stored rendered HTML on the origin server; static assets might be served from a CDN, but HTML still required origin involvement. CDN-level page caching places HTML in the CDN edge, so users fetch fully rendered pages from an edge PoP near them. This reduces latency dramatically for global audiences and reduces origin load.

How Cloudflare Enterprise improves caching
Cloudflare Enterprise brings higher cache hit ratios and advanced tiered caching. In a tiered architecture, a local PoP first checks its cache, then queries an upper tier (within Cloudflare’s private backbone) before hitting the origin. This internal network (Cloudflare’s intranet) is faster and avoids public internet congestion. Higher cache hit ratios mean fewer origin fetches, more consistent low latency, and improved Core Web Vitals for users worldwide.

BigScoots Cache plugin and developer features
BigScoots built a proprietary plugin, BigScoots Cache, to tightly integrate WordPress with Cloudflare’s page caching. The plugin:
– Leverages Cloudflare Enterprise caching and tiering to maximize hit rates.
– Provides fine-grained cache controls and TTL tuning per page or content type.
– Implements intelligent purge logic: when content updates, it purges related pages (e.g., taxonomy and archive pages) automatically.
– Offers hooks and documentation for advanced users to customize cache behavior or integrate with external systems.
– Exposes a REST API for programmatic cache clearing, enabling integrations (e.g., backend systems, e-commerce flows) to purge caches on content changes.

Direct Cloudflare peering and CNI
BigScoots runs its own data centers and has invested in a private connection to Cloudflare (CNI). Rather than relying solely on the public internet for origin fetches, Cloudflare can fetch content from BigScoots over a private, high-bandwidth link. This reduces latency and improves reliability for origin pulls. Such direct peering requires physical data center control and isn’t feasible for hosts that only lease infrastructure from larger cloud providers.

Customer-facing controls and managed services
BigScoots exposes many Cloudflare Enterprise controls in its portal with fine-grain toggles for performance, security, and access control:
– Block or challenge traffic by country/continent.
– Hardening and bot management options.
– Image optimization and speed-related toggles.
– Login protection and WAF settings.
These settings are designed to be accessible for non-technical users while offering hooks and APIs for advanced customers and agencies. BigScoots also provides managed support, onboarding, zero-downtime migrations, performance optimizations, and bespoke engineering for enterprise and agency clients.

Audience and use cases
While BigScoots serves “mom-and-pop” sites, these Enterprise features and the BigScoots Cache plugin are especially valuable for agencies and enterprise customers with many sites, high traffic, or strict uptime and performance requirements. Saumya emphasizes that many smaller customers still benefit from easy toggles (e.g., blocking traffic from a problematic country) without needing deep technical knowledge.

Final notes
Saumya reiterates that complexity is inherent in large internet services and that outages, while serious, are part of operating at scale. Cloudflare’s transparent handling and continuous improvements are positive signs. BigScoots focuses on combining deep Cloudflare integration, private peering, intelligent caching, and managed support to reduce downtime, improve speed, and give customers control.

For more information and links mentioned in the episode, visit wptavern.com/podcast and search for the episode with Saumya Majumder.