Saumya Majumder, lead software engineer at BigScoots, joins the Jukebox Podcast to discuss a recent major Cloudflare outage, how large infrastructure hiccups ripple across the web, and the WordPress performance solutions BigScoots builds—often in close integration with Cloudflare Enterprise.

Background and focus
Saumya specializes in high-performance WordPress engineering and Cloudflare-powered architectures. Early in his career he worked on caching problems and co-created the Super Page Cache for Cloudflare plugin (later acquired). Over time he has built custom caching engines, migration tools, worker-based automations, edge computing solutions, and managed enterprise customers and internal WordPress tooling at BigScoots.

Cloudflare outage: what happened and why it matters
Saumya explains that large platforms like Cloudflare do far more than basic CDN work; they are complex stacks with many interdependent systems. A single unexpected change—such as a configuration file that doubled in size—can cascade and look like a DDoS event. That triggers investigations and mitigations that may initially be misdirected, prolonging downtime. Even after applying a fix, propagation across Cloudflare’s many points of presence (PoPs) and dealing with the resulting traffic spikes mean recovery takes time.

He stresses that outages are inevitable across providers—even well-engineered systems at Google, AWS, or private data centers can fail. Big companies mitigate risk with SLAs and financial credits for affected customers, so incidents are taken very seriously. Saumya praises Cloudflare’s transparency in publishing detailed postmortems and notes that they learn from each incident to reduce recurrence.

Mitigations BigScoots used
Because Cloudflare offers many features (WAF, Turnstile, Workers, hosting), the impact on an individual customer depends on which Cloudflare services they rely upon. For sites hosted at BigScoots but proxied through Cloudflare, the BigScoots team used Cloudflare’s API to turn off proxying rapidly when the outage started. That routed requests directly to BigScoots origins and kept many sites online while Cloudflare recovered. Saumya notes that this option wouldn’t help customers who host entirely on Cloudflare (e.g., on Workers or Pages), where the outage affects the hosting layer itself.

CDN-level page caching: the evolution
Saumya describes the shift from server-level HTML caches (advanced-cache.php style) to CDN-level page caching. Traditional caches store rendered HTML on the origin server so subsequent requests don’t rerun PHP and database queries. But latency remains when users are geographically distant from origin. BigScoots and peers pioneered serving full HTML from the CDN so users receive cached pages from a nearby PoP instead of the origin—dramatically reducing response times.

Cloudflare Enterprise and tiered caching
Cloudflare Enterprise enables high-assurance, global page caching with high cache-hit ratios. Saumya explains Cloudflare’s tiered-cache model: lower-tier PoPs first check nearby or regional cache tiers within Cloudflare’s private network (not the public internet). If a lower tier lacks content, it consults upper tiers before going to the origin. This intra-Cloudflare tiering reduces origin load and increases cache hits, improving speed for users worldwide.

Private connectivity and data center strategy
BigScoots runs its own data centers and has direct physical connections to Cloudflare (CNI). Instead of Cloudflare fetching origin content over the public internet, these private links allow Cloudflare to pull content directly and faster from BigScoots’ origin network. Saumya notes most hosts lease infrastructure from cloud providers and lack the physical access needed to create such private interconnects. BigScoots’ approach—owning hardware and establishing direct links—enables lower latency and more reliable origin fetches.

BigScoots Cache plugin and control features
BigScoots developed a proprietary plugin, BigScoots Cache, to manage Cloudflare-based page caching and give site owners fine-grain control. Key capabilities include:
– CDN-level page caching integration with Cloudflare Enterprise for high cache-hit ratios.
– Intelligent cache purging: when a post is updated, the plugin clears not just that page but related taxonomy, archive, and author pages.
– Hooks and APIs: advanced users can use hooks or BigScoots Cache REST API to trigger cache purges from external systems or custom workflows.
– Admin controls in the BigScoots portal: toggles for login protection, hardening, image optimization, Rocket Loader, and bot management.
– Geo and country controls: block or challenge requests from specific countries or continents via simple portal settings.

Managed support and enterprise onboarding
BigScoots provides managed services, including migrations with zero downtime, onboarding for enterprise customers, performance optimization packages, and bespoke engineering work. For many customers—agencies and enterprises—BigScoots offers hands-on configuration and can implement custom snippets or logic to handle edge cases (e.g., theme-specific pseudo-archives). Documentation exists for advanced users, but managed support handles most customer needs.

Conclusions
Saumya highlights that while incidents like Cloudflare’s outage are disruptive, they’re part of operating at scale—and large providers typically respond transparently and iterate to avoid repeats. BigScoots’ integration with Cloudflare Enterprise, combined with its own data center infrastructure and tooling, aims to maximize performance, reliability, and control for WordPress sites. The company emphasizes options for both novice users (simple toggles and managed support) and advanced users (hooks, APIs, fine-grained controls) to tailor caching, security, and delivery to their needs.

Welcome to the Jukebox Podcast from WP Tavern. Nathan Wrigley speaks with Saumya Majumder, lead software engineer at BigScoots, about a recent major Cloudflare outage, how such failures affect the web, and the WordPress performance and hosting strategies BigScoots uses—especially in conjunction with Cloudflare Enterprise.

Who Saumya Majumder Is
Saumya leads high-performance WordPress engineering and Cloudflare-powered architectures at BigScoots. He’s built systems ranging from custom cache engines and migration tools to worker-based automations and edge computing. He oversees enterprise customers and internal WordPress projects, building scalable, developer-friendly solutions.

Cloudflare’s Outage: What Happened and Why Failures Occur
Saumya explains that while people often think of Cloudflare only as a CDN, it’s a huge, complex platform with many critical dependencies. Global outages occur not because of malice, but because a single low-probability edge-case failure in a massive system can cascade. Fixing the root cause takes time; propagating fixes across Cloudflare’s many Points of Presence (PoPs) and control planes is nontrivial, and the recovery itself can produce traffic bursts and require further stabilization.

He emphasizes that outages are inevitable across major providers—Cloudflare, AWS, GCP, Azure, etc.—and come with financial and reputational consequences due to SLAs. Cloudflare’s response to outages tends to be transparent and thorough, publishing detailed postmortems and then working to remove fragile dependencies (for example, previous issues involving KV storage tied to third-party services).

BigScoots’ Mitigations During an Outage
BigScoots mitigated customer impact during the outage by using the Cloudflare API: they turned off Cloudflare proxying for affected domains so traffic went directly to BigScoots’ origin servers. Because Cloudflare’s API remained available, BigScoots automated rapid failovers to bypass the proxy layer until Cloudflare stabilized. Saumya stresses this approach only works when the origin hosting is separate from Cloudflare-hosted compute (e.g., if a site is fully hosted on Cloudflare Workers/platform, that API workaround wouldn’t help).

CDN-Level Page Caching: Origins and Benefits
Saumya recounts early work he and a colleague did on CDN-level page caching, producing a plugin (Super Page Cache for Cloudflare) later acquired by another company. Traditional server-side caching stores generated HTML on the origin server, which still forces requests from far-away clients to traverse long distances to the origin. CDN-level page caching moves that HTML out to the edge so users worldwide receive content from a nearby PoP, significantly reducing latency and origin load.

Cloudflare Enterprise and Tiered Caching
Cloudflare Enterprise enables BigScoots to deliver CDN-level page caching with very high cache hit ratios. Saumya explains tiered caching: PoPs are organized into lower and upper tiers within Cloudflare’s private network. If a lower-tier PoP lacks a cached asset, it first checks upstream tiers within Cloudflare’s intranet (not the public Internet). If another tier has the content, it serves it quickly; only when no cache exists does the request go to the origin. This architecture reduces origin requests and improves global performance.

BigScoots’ Direct Interconnect to Cloudflare
BigScoots runs its own data centers and has built a private interconnect (CNI) to Cloudflare. This private fiber connection avoids the public Internet when Cloudflare needs to fetch content from BigScoots’ origin, reducing latency and variability. Saumya notes that most hosts don’t run their own data centers and thus can’t make such physical connections—BigScoots’ setup lets Cloudflare fetch origin content over a private backbone for faster, more reliable origin pulls.

BigScoots Cache Plugin and Fine-Grained Controls
BigScoots developed a proprietary plugin, BigScoots Cache, to orchestrate Cloudflare page caching and allow fine-grained control. Features include:
– Intelligent cache purging: when publishing or updating content, the system clears not only the page but related assets (taxonomies, author archives, linked pages).
– Hooks and APIs: developers can use filters, actions, and a REST API to programmatically purge or adjust cache behavior (useful for custom applications or e-commerce).
– Advanced controls in the BigScoots portal: customers can toggle settings for login protection, bot blocking, image optimization, Rocket Loader, and more. Settings can block or challenge requests from selected countries or continents, apply proprietary hardening rules, and manage WAF/bot behavior.

Target Customers and Managed Service Model
While the platform offers advanced options for technical users, BigScoots positions these features to be accessible for all customers. Documentation explains hooks and APIs for developers, and managed support teams will implement custom rules and snippets for customers who prefer not to handle the technical details themselves. BigScoots provides systematic onboarding for enterprises and agencies, handles migrations with zero downtime, and offers performance optimization packages and engineering services that include custom code work.

Key Takeaways on Reliability and Complexity
– Outages happen even at top providers; the engineering challenge is mitigating and learning from them.
– Transparency and post-incident learning are important—Cloudflare’s detailed postmortems and corrective actions are valuable.
– Having origin hosting separate from a CDN/proxy allows mitigations (e.g., disabling proxying via API) during CDN outages.
– CDN-level caching and tiered cache architectures drastically reduce latency and origin pressure when implemented correctly.
– Direct private interconnects between host and CDN improve speed and consistency but require physical data center infrastructure.
– Fine-grained controls, managed services, and developer APIs let hosting providers tailor caching and security behavior for diverse WordPress needs.

Conclusion
The conversation underscores that the internet’s ability to work at global scale depends on complex systems that will sometimes fail. Hosting and CDN architectures that combine edge caching, tiered cache logic, private interconnects, and well-crafted management tools can both improve performance and reduce user-visible impact during incidents. BigScoots’ work with Cloudflare Enterprise highlights one approach to delivering faster, more resilient WordPress hosting through tight integration, automation, and managed expertise.

For more details and links, see the original episode notes at wptavern.com/podcast.