Welcome to the Jukebox Podcast from WP Tavern. Nathan Wrigley interviews Saumya Majumder, lead software engineer at BigScoots, about a recent major Cloudflare outage and BigScoots’ advanced Cloudflare-powered WordPress architectures and performance solutions.

Who is Saumya Majumder
Saumya is lead software engineer at BigScoots specializing in high-performance WordPress engineering and advanced Cloudflare architectures. He builds large-scale systems—custom caching engines, migration tools, worker-based automations, and edge computing solutions—and oversees enterprise customers, internal WordPress projects, plugins, and developer-friendly hosting solutions.

How Saumya got into caching and CDNs
Saumya’s interest began in college tinkering with PHP and early WordPress. Seeing latency problems when pages had to travel long distances to origin servers, he and a collaborator created one of the early CDN-level page-caching solutions (Super Page Cache for Cloudflare), later acquired by Optimal. The idea: instead of only serving static assets from CDNs, serve full HTML pages from CDN PoPs so visitors get responses from a nearby node rather than the origin across the globe.

Cloudflare outage: what happened and why it matters
Saumya explains that the internet and platforms like Cloudflare are immensely complex. Cloudflare is more than a CDN; it’s a large platform with many interdependent components. Even with extensive safeguards, a low-probability edge-case bug or configuration issue can cascade into major outages. In the recent incident Cloudflare’s engineers initially misdirected diagnostics because some symptoms resembled a DDoS; propagation effects across Cloudflare’s global control plane then prolonged recovery. Fixing the root cause and propagating changes across thousands of PoPs takes time, especially when handling bursts of client traffic during recovery.

He stresses that outages are inevitable across any major provider—Google, AWS, Azure, GCP, or Cloudflare—and that enterprises contract SLAs that make outages costly for providers both in reputation and credits owed to customers. Cloudflare’s response practices—detailed postmortems and transparency—are notable. They analyze causes, publish what happened, and work to remove dependencies or harden systems to avoid recurrence.

How BigScoots mitigated the outage for customers
BigScoots uses Cloudflare as a reverse proxy and leverages its API. During the outage, BigScoots used the Cloudflare API to turn proxying off for affected customers so traffic went directly to BigScoots origins. Because the Cloudflare API remained reachable in that incident, BigScoots was able to rapidly switch traffic away from broken proxies and reduce downtime for customers whose origins could handle direct traffic. This approach works for traditional WordPress hosting setups proxied through Cloudflare; customers running entirely on Cloudflare-hosted platforms (Workers, Pages, etc.) couldn’t rely on that workaround.

CDN-level page caching and Cloudflare Enterprise
Saumya outlines the evolution from server-side cache files (advanced-cache.php, local page caches) to CDNs serving static assets, and finally to CDNs caching full HTML pages. CDN-level page caching reduces latency dramatically since a cached HTML response can be served from a local PoP rather than from the origin data center.

Cloudflare Enterprise enables higher cache hit ratios and advanced features:
– Tiered caching: PoPs consult upstream tier nodes within Cloudflare’s private network before going to origin, allowing many requests to be satisfied internally without returning to the origin. This reduces origin load and improves latency.
– Regional and global cache strategies: Enterprise plans give better long-term caching behavior and more control.
– Cloudflare’s internal backbone (private intranet) and many PoPs make fetching cached content across regions fast and efficient.

BigScoots Cache plugin and integration
BigScoots developed an in-house plugin, BigScoots Cache, to integrate with Cloudflare Enterprise and manage page caching behavior. Key capabilities:
– Control and fine-tuning: Set CDN cache TTLs, tailor cache rules, and manage nuanced behaviors per site.
– Intelligent cache purging: When a post is published or updated, the plugin clears the relevant page and related pages (taxonomies, archives, author pages), not just the single URL, avoiding stale content across linked pages.
– Hooks and extensibility: Developers can use actions and filters to customize caching logic (e.g., clear caches for pseudo-archive pages generated via shortcodes).
– REST API: BigScoots Cache exposes an API so external systems (e.g., headless backends or e-commerce platforms) can purge or manipulate cache programmatically.
– Managed support: For customers who don’t want to manage hooks and settings, BigScoots can implement custom cache-clearing logic or assist with migrations and configuration.

Portal-level controls and user-facing features
BigScoots exposes fine-grained Cloudflare Enterprise controls in its client portal, allowing users to:
– Toggle protections for login pages and apply DDoS/bot mitigations.
– Enable BigScoots’ proprietary hardening and speed optimizations.
– Configure image optimization, Rocket Loader, and other performance features.
– Block or challenge traffic by country or continent; combine block and challenge rules.
– Block AI or bad bots and manage WAF/challenge settings.

These controls are designed to be simple for non-technical users (pick a country and click) yet deep enough for agencies or developers to customize via hooks and APIs.

Direct interconnect (CNI) and private backbone connections
BigScoots runs its own data centers and has direct physical interconnects with Cloudflare (referred to as CNI). This private optical-fiber connection means that when Cloudflare needs to fetch content from a BigScoots origin, the request can travel over a dedicated secure channel rather than the public internet, reducing latency and improving reliability. Not all hosts can do this—providers that lease infrastructure from public clouds lack the physical control required to create such private links.

Benefits for WordPress sites
For WordPress users, the combined BigScoots + Cloudflare Enterprise setup delivers:
– Much faster page loads and improved Core Web Vitals due to CDN HTML caching and tiered cache behavior.
– Lower origin load and better scalability thanks to high cache hit ratios.
– DDoS protection, bot management, and country-level access control.
– Developer tools (hooks, APIs) for custom cache workflows and integrations.
– Fully managed onboarding and migrations for enterprise or agency customers, including zero-downtime migrations, performance optimization services, and engineering support.

Who benefits most
While small sites and “mom-and-pop” shops can use and benefit from these features (and often do, through simple portal toggles), the depth of customization—hooks, REST API, and advanced Cloudflare options—particularly suits agencies and enterprise customers that need fine-grained control, high performance, and managed services.

Conclusion
The recent Cloudflare outage is a reminder that even the most sophisticated internet infrastructure can experience failures. Cloudflare’s scale and complexity make it a powerful platform, but edge cases and propagation effects can trigger large incidents. BigScoots leverages deep Cloudflare integration, in-house caching technology, direct interconnects, and managed services to mitigate outages and maximize performance for WordPress users. Their approach combines enterprise-level caching strategies and user-facing simplicity so both developers and non-technical customers can take advantage of the speed and protections Cloudflare offers.

Saumya Majumder, lead software engineer at BigScoots, joins the Jukebox Podcast to discuss a recent major Cloudflare outage, how global infrastructure failures happen, and how BigScoots uses Cloudflare Enterprise and its own engineering to improve WordPress performance and reliability.

Background and role
Saumya specializes in high-performance WordPress engineering and Cloudflare-powered architectures. He builds large systems—custom cache engines, migration tools, worker automations, edge computing—and manages enterprise customers, internal WordPress projects, plugins, and developer-friendly solutions at BigScoots. His early work included creating Super Page Cache for Cloudflare (later acquired), an early CDN-level page caching solution that moved full HTML caching to the edge.

What happened with the Cloudflare outage
Cloudflare is far more than just a CDN; it’s a vast, complex platform with many interdependent systems. Saumya explains that when one critical component fails—often due to low-probability edge cases—the consequences can cascade rapidly. The recent outage was aggravated by an unexpected increase in the size of a configuration file, which propagated across Cloudflare’s network and looked like a DDoS, leading engineers to initially chase the wrong direction. Fixes must propagate across thousands of points-of-presence (PoPs), and stabilization can take hours as new traffic bursts and global propagation occur.

He emphasizes that outages are inevitable across all major providers—Cloudflare, AWS, GCP, Azure—and that companies take them extremely seriously because of SLAs and the financial and reputational consequences. Cloudflare’s strengths include transparency and detailed postmortems; they document root causes and mitigation steps and aim to remove dependencies (e.g., earlier reliance on external services like GCP for KV storage) to prevent repeats.

How BigScoots mitigated the outage for customers
Because BigScoots hosts its own infrastructure and integrates with Cloudflare, it could use Cloudflare’s API to toggle proxying behavior when the outage started. By turning the Cloudflare proxy off via API for affected customer domains, BigScoots allowed requests to go directly to origin servers, keeping sites online even while parts of Cloudflare were degraded. This approach works when Cloudflare is acting as a proxy for externally hosted sites; it wouldn’t help if a customer’s site is hosted entirely on Cloudflare Workers/Pages, since those rely directly on Cloudflare’s infrastructure.

CDN-level page caching and BigScoots cache
Saumya recounts how traditional WordPress caching kept HTML cached on the origin server (advanced-cache.php) and CDNs only served static assets. His early work focused on caching entire HTML pages on the CDN so visitors worldwide could receive full-page responses from a nearby PoP—dramatically reducing latency compared to origin-based caching.

Cloudflare Enterprise enables high cache-hit ratios with tiered caching. Tiered cache means lower-tier PoPs check upper-tier PoPs within Cloudflare’s private network before going to origin. If an upper tier has the page, the lower tier fetches it over Cloudflare’s intranet instead of the public internet, reducing origin requests and improving latency. Enterprise features include longer cache retention, tiering, and regional optimizations that boost hit rates versus lower-tier plans.

BigScoots built BigScoots Cache, a proprietary plugin and API that integrates with Cloudflare Enterprise to:
– Leverage CDN-level HTML caching with high hit rates.
– Provide intelligent cache purging: when a post is updated, the plugin clears not only the post page but associated taxonomies, archives, author pages, etc.
– Offer hooks and filters for advanced control (TTL overrides, custom purge logic for nonstandard pages like shortcode-driven listing pages).
– Expose a REST API for programmatic cache control, useful for headless setups or external apps.

BigScoots portal and admin controls
BigScoots exposes fine-grained Cloudflare Enterprise controls through its portal, allowing users (including non-technical customers) to toggle features and apply security and performance policies:
– Country/continent blocking and challenging
– Image optimization and Rocket Loader
– WAF and login protection toggles
– Blocking or challenging bad AI bots and other threats
– Proprietary hardening options beyond Cloudflare defaults

These controls are designed to be easy for simple tasks (e.g., block traffic from a specific country) while still allowing advanced users and agencies to fine-tune behavior. BigScoots also provides managed support and will implement custom hooks or snippets for clients who need bespoke purge behavior or performance tuning.

Private backbone integration (CNI) and performance
BigScoots runs its own data centers and private cloud infrastructure, unlike many hosts that rent cloud instances. This allowed BigScoots to create physical connections—private optical fiber links—to Cloudflare (CNI). Instead of Cloudflare pulling content from origins over the open internet, Cloudflare can fetch data via BigScoots’ private network connection, reducing latency and avoiding public internet congestion. This direct peering and backbone integration is a major performance advantage but requires physical infrastructure ownership or colocation.

Who benefits
While BigScoots supports a range of customers, the advanced caching, thorough integration with Cloudflare Enterprise, and managed services are especially attractive to agencies, enterprise sites, and teams that need guaranteed reliability, fine-grained cache-control, and hands-on onboarding. That said, many smaller customers benefit too through simple portal toggles and managed responses to common problems like traffic spikes or abusive bot traffic.

Operational approach and philosophy
Saumya stresses that engineering is iterative and that caching and edge solutions are ongoing processes. Some sought-after features (edge DB-protected caching, login-user caching) involved years of research and development. BigScoots prefers to automate mitigation (e.g., API-driven failovers), provide transparent reporting, and offer managed services for clients who don’t want to handle low-level details. He also applauds Cloudflare’s transparency in outages and their ongoing efforts to remove single points of failure.

Summary
Outages happen even at the largest infrastructure providers; complexity and rare edge cases make perfect uptime impossible. The important responses are rapid mitigation, transparency, and architectural changes to reduce recurrence. BigScoots mitigates third-party outages by combining origin control, automation (API-driven proxy toggling), proprietary caching integrations, direct network peering, and a suite of user-facing controls and managed services built around Cloudflare Enterprise to maximize WordPress performance and reliability.

Nathan Wrigley interviews Saumya Majumder, lead software engineer at BigScoots, about Cloudflare’s recent outage, how such outages affect the web, and the performance and reliability solutions BigScoots builds for WordPress sites.

Background
Saumya specialises in high-performance WordPress engineering and Cloudflare-powered architectures. Over time he has built caching engines, migration tools, worker-based automations, and edge computing solutions. He supports enterprise customers, internal WordPress projects, and developer-friendly scalable hosting solutions.

Cloudflare outage: what happened and why
Saumya explains that large internet services like Cloudflare are more than CDNs: they are complex platforms with many interdependent systems. When a low-probability edge case occurs—such as an unexpected configuration or file size change—it can ripple through layers and trigger widespread failures. Fixing the root cause and then propagating repairs across hundreds of points of presence (PoPs) and an internal network takes time, and the mitigation itself can be complex due to traffic surges and propagation delays.

He stresses that outages are inevitable in complex distributed systems—no provider (Cloudflare, AWS, GCP, or others) is immune. Enterprises rely on SLAs and expect rapid remediation, but even so, edge cases and hidden dependencies can cause incidents. Saumya praises Cloudflare’s transparency: the company publishes detailed postmortems that explain what happened, what was fixed, and what changes will prevent recurrence. He notes Cloudflare’s track record of learning from incidents and removing problematic external dependencies.

BigScoots’ mitigation and approach
When Cloudflare’s public-facing services (e.g., Turnstile captcha or other features) were affected, BigScoots used its own systems and APIs to reduce customer impact. For customers proxied through Cloudflare, BigScoots automated a temporary unproxying via the Cloudflare API: traffic was routed directly to BigScoots’ origin servers until Cloudflare restored normal operation. This mitigated login and access problems for many WordPress sites that otherwise relied on Cloudflare features for authentication or front-line services.

Saumya notes that for applications fully hosted on Cloudflare’s worker or edge platform, the API workaround wouldn’t help—those workloads would be affected by Cloudflare’s outage. The important point is architecture matters: whether Cloudflare is used as a proxy/CDN in front of an origin or as the primary host changes recovery options.

Why Cloudflare is widely used
Despite the risks, Saumya emphasizes why many major sites and enterprises use Cloudflare: it provides a vast global backbone, many features (CDN, security, developer platform, serverless/workers), and performance advantages others don’t match. The scale and feature set are what attract customers, and Cloudflare’s rapid post-incident remediation and transparent communication also build trust.

CDN-level page caching: the BigScoots approach
Saumya recounts the evolution from server-side caching (advanced-cache.php and local HTML files) and static-file CDNs to the idea of serving full HTML pages from a CDN. The benefit is obvious: a user in Sydney need not have a request cross an ocean to the origin if the HTML is cached at a nearby PoP. Saumya and a colleague developed early CDN-level page caching (a plugin later acquired) to serve page HTML globally, not just static assets.

Cloudflare Enterprise opens advanced caching capabilities. BigScoots uses Cloudflare Enterprise to provide global page caching with high cache hit ratios, tiered caching, and regional tiering. Tiered caching means local PoPs consult higher-tier PoPs within Cloudflare’s private backbone before going back to the origin. That reduces origin requests and keeps responses fast by fetching cached content internally within Cloudflare’s network rather than over the public internet.

Private interconnects and performance
BigScoots runs its own data centers and has direct physical connections to Cloudflare (CNI / private interconnect). Instead of Cloudflare fetching content from origin over the open internet, these private links let Cloudflare reach BigScoots’ infrastructure over a dedicated, high-bandwidth path. That reduces latency and avoids public internet congestion, improving the speed of cache fills and origin responses.

This direct connectivity is possible because BigScoots owns and operates its infrastructure. Many hosts that lease infrastructure from hyperscalers can’t establish equivalent private links to Cloudflare in the same way.

BigScoots Cache plugin and portal controls
BigScoots built BigScoots Cache, a plugin and platform integration that ties WordPress to Cloudflare Enterprise caching and offers fine-grained control:

– CDN-level HTML caching: Pages can be cached at PoPs worldwide, reducing latency for global users.
– Intelligent purge: When a post is published or updated, BigScoots Cache clears not only that page but related taxonomy, archive, and author pages as needed, improving cache consistency.
– Hooks and extensibility: Advanced users can use actions/filters to customise TTLs, purge rules, or target pages that are not standard WordPress archives (shortcode-generated listings, etc.). Documentation exists for developers; BigScoots also writes snippets or custom code for customers who need help.
– REST API: BigScoots Cache exposes an API to clear cache programmatically, useful for headless setups or external systems (e.g., Laravel backends) that need to trigger purges.
– Portal toggles and enterprise controls: The BigScoots control panel exposes many Cloudflare Enterprise settings for customers—country/continent blocking or challenging, image optimisation, Rocket Loader, hardening protections, bot and AI-bot blocking, WAF/login hardening, and more. These controls aim to be accessible for non-experts (simple toggles) while also supporting advanced customisation.

Target customers and managed services
BigScoots serves a range of customers from small sites to agencies and enterprise clients. For higher-scale customers and agencies, BigScoots offers systematic onboarding, zero-downtime migrations, performance optimisation services, and managed support. If customers don’t want to tinker with advanced settings, BigScoots’ support team can configure the portal options, tune caching, or implement custom purge logic on their behalf.

On engineering philosophy
Saumya describes caching and large-scale system design as ongoing work: there are always more edge cases to solve and new improvements to pursue. Recent BigScoots efforts include edge and login-user caching protections, considered long-term R&D projects.

Final points
– Outages happen even at top providers; what matters is response, transparency, and learning.
– Architecture choices determine mitigation options: using Cloudflare as a proxy in front of an origin allows origin-directed failover strategies; fully relying on edge hosting is different.
– Combining Cloudflare Enterprise with private interconnects and a host-managed caching layer can yield significant performance and reliability advantages for WordPress sites.
– For customers, many advanced features can be managed via a UI or by BigScoots’ managed services, with developer hooks and APIs available for deeper integration.

For more details and links related to the conversation, search the episode at wptavern.com/podcast.