Enterprise or Upstream (How to Decide) 🔗 **Source:** [Enterprise or Upstream (How to Decide)](https://www.youtube.com/watch?v=cVMS1kPRwVg) > [!note]+ Document Notes > tags: [networking, enterprise-infrastructure, sonic] > author: Adam Dunstan > source: YouTube > reference: https://www.youtube.com/watch?v=cVMS1kPRwVg > date created: 2024-12-19 20:30 > > # 🏷️ Title: Enterprise or Upstream (How to Decide) > <!-- deep-read:sections start --> ## 🧰 Action Inbox - [H1] Evaluate procurement process optimization before white box switching — save more on cost #procurement - [H2] Assess team skills for DIY vs partnership approach — significant commitment required #planning - [H3] Insist partners push contributions upstream — unlock community benefits #opensource ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ### 🔗 References - [Enterprise or Upstream (How to Decide)](https://www.youtube.com/watch?v=cVMS1kPRwVg) ## 📝 Summary - Adam Dunstan, independent consultant with extensive networking experience, discusses choosing between enterprise and upstream versions of SONiC - Key reasons for white box switching: operational automation, better tooling integration, simpler platform integration, new functionality development - Bad reasons: expecting unlimited choice (limited by ASIC vendors targeting high-volume customers), cost savings alone (procurement optimization more effective) - Upstream (GitHub version) provides community leverage vs enterprise versions that may not contribute back - Three approaches: operate only, partner for integration, or full DIY (requires significant team commitment) - Bottom line: Always push for upstream contributions to unlock community benefits and avoid vendor lock-in ## 📖 Main Content Adam Dunstan presents a strategic framework for deciding between enterprise and upstream SONiC implementations. Drawing from his experience at CenturyLink (saved $250M) and Mercedes-Benz (saved 25M euros), he argues that the primary value of white box switching lies in operational automation rather than cost savings alone. The ecosystem complexity requires understanding four layers: ASIC vendors (Broadcom, etc.), ODMs, software layers, and ongoing support. ASIC vendors drive feature availability by targeting high-volume cloud providers, potentially limiting choice for smaller deployments. Dunstan distinguishes between upstream (GitHub-hosted, community-supported) and enterprise versions (may be forks without upstream contribution). He advocates strongly for upstream approaches to leverage community development—potentially thousands of engineers vs. Cisco IOS's peak of 300-400 engineers. Three implementation approaches exist: operational-only, partnering for integration work, or full DIY requiring diverse technical skills including ASIC programming, kernel development, and hardware integration. ## 🗒️ Notes #### ⏱️ [00:00](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=0) — [H1] Speaker Introduction and Background - Key idea: Extensive experience from Bridge Communications through Avicii terabit routers to white box switching - Evidence: "I've been in this business for a long time... first of the terabit routers using supercomputer technology" #### ⏱️ [02:30](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=150) — [H1] Good Reasons for White Box Switching - Key idea: Operational automation is the primary driver, not traditional set-and-forget networking - Evidence: "Traditional networking guys will go now that it's up and running the routing protocols take care of it all... but the reality of that is way in the past" #### ⏱️ [04:15](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=255) — [H2] Platform Integration Benefits - Key idea: Network infrastructure should integrate with existing systems rather than requiring specialist platforms - Evidence: "Why do I need a specialist Network platform... most CIOs... would like it to kind of disappear and be part of my platform" #### ⏱️ [05:45](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=345) — [H3] New Protocol Development Capabilities - Key idea: White box switches enable custom protocol implementation previously impossible - Evidence: "I'm working with some guys who are building new protocols... we can actually put new protocols inside a switch pretty easily" #### ⏱️ [06:30](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=390) — [H1] Bad Reason: Expecting Unlimited Choice - Key idea: ASIC vendors target high-volume customers, limiting options for smaller deployments - Evidence: "They're chasing large volumes... building systems for large players Cloud providers... thousands and thousands of switches" #### ⏱️ [08:00](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=480) — [H2] Legacy Infrastructure Challenges - Key idea: Finding white box solutions for legacy 10/100 Mbps infrastructure proves difficult - Evidence: "Thousands of these 10 and 100 gigs 100 megabit switches... it was a real challenge to go and find a white box switch" #### ⏱️ [09:15](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=555) — [H1] Bad Reason: Cost Savings Alone - Key idea: Procurement optimization delivers larger savings than technology changes - Evidence: "At Century over a couple of years I saved them 250 million dollars... that wasn't because we did switches it's because we changed the way we did procurement" #### ⏱️ [10:30](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=630) — [H1] Ecosystem Complexity - Key idea: White box switching requires understanding supplier relationships instead of single vendor integration - Evidence: "You've got to go and dig into the place that I used to live in... understanding the suppliers to your vendors" #### ⏱️ [12:00](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=720) — [H2] ASIC Vendor Limitations Example - Key idea: Broadcom switches cannot do simple port-to-port forwarding due to architecture requirements - Evidence: "You really can't do this because of the way that the forwarding and control is instantiated... you've got to build Bridge domains" #### ⏱️ [14:15](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=855) — [H2] Upstream vs Enterprise Definition - Key idea: Upstream is GitHub-hosted community version, enterprise may be forks without contribution back - Evidence: "Upstream is the thing that's on GitHub... Enterprise version is something other than that" #### ⏱️ [15:45](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=945) — [H3] Vendor Claims vs Reality - Key idea: Some vendors claim SONiC support without upstream hardware support - Evidence: "There are vendors out there today where they say hey we support Sonic their Hardware support isn't an upstream" #### ⏱️ [17:00](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=1020) — [H1] Implementation Approaches - Key idea: Three levels - operate only, partner for integration, or full DIY - Evidence: "Am I just going to operate a network... partner with someone... or DIY" #### ⏱️ [18:30](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=1110) — [H2] Partnership Approach Example - Key idea: Even large telcos partner to avoid ASIC-level programming complexity - Evidence: "We did not want to go and have to get the broadcom code... we still went and partnered someone to do that" #### ⏱️ [20:00](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=1200) — [H1] Community Scale Potential - Key idea: SONiC community can leverage more engineers than traditional vendor development - Evidence: "Cisco Systems and iOS at its peak is perhaps 400 Engineers... not a huge number compared to the number of people that can be leveraged against the future of Sonic" #### ⏱️ [21:30](https://www.youtube.com/watch?v=cVMS1kPRwVg?t=1290) — [H1] Upstream Advocacy - Key idea: Insist partners and ODMs contribute to upstream to unlock community benefits - Evidence: "I believe you should insist that they push it to Upstream... if we don't unlock the community we won't have anything different" ## 📚 References - [Enterprise or Upstream (How to Decide)](https://www.youtube.com/watch?v=cVMS1kPRwVg) - SONiC GitHub Repository - Broadcom SDK Documentation - ASIC Vendor Platforms ## 🗣️ Paraphrase - White box switching's main value is operational automation enabling cloud-like provisioning workflows, not traditional cost savings - The ecosystem requires understanding ASIC vendors, ODMs, software layers, and support—choice is limited by vendors targeting high-volume customers - Three approaches exist: operate existing systems, partner for custom integration, or build complete in-house capability requiring diverse technical skills - Always advocate for upstream contributions to leverage community development rather than creating isolated enterprise forks ## 📰 Newsletter **SONiC Strategy: Choosing Your Path Forward** If you're evaluating SONiC for network infrastructure, Adam Dunstan's framework helps clarify the decision between enterprise and upstream approaches. The key insight: pursue this technology for operational automation benefits, not just cost savings. **Key Considerations:** - **Good reasons**: Operational automation, platform integration, custom protocol development - **Bad reasons**: Expecting unlimited choice (ASIC vendors target high-volume customers), cost savings alone (optimize procurement instead) - **Implementation paths**: Operate-only, partnership, or full DIY (significant commitment required) **Strategic Recommendation**: Always push for upstream contributions. SONiC's community potential far exceeds traditional vendor development teams. The ecosystem complexity requires understanding suppliers throughout the stack, but the automation and integration benefits can transform network operations for organizations ready to invest in the approach. Resources: [SONiC GitHub](https://github.com/sonic-net/SONiC), [Networking Documentation](https://docs.sonic-net.org/), [Community Forums](https://github.com/sonic-net/SONiC/discussions) ## ✅ Action Items #### DevRel - Developer advocacy: Create SONiC integration workshops and automation demos - Education: Develop curriculum covering ASIC ecosystem and vendor relationships - Developer experience: Improve upstream contribution documentation and tooling - Developer marketing: Highlight community scale advantages over enterprise approaches - Community management: Foster partnerships between enterprises and upstream contributors #### OSPO - Policy & Governance: Establish guidelines for upstream-first networking technology adoption - License Compliance: Review SONiC and ASIC vendor licensing requirements - Security Management: Assess vulnerability management across networking ecosystem components - Software Inventory: Track networking infrastructure dependencies and upstream relationships - Developer Support: Provide training on networking ecosystem complexity and vendor relationships - Community Engagement: Support SONiC upstream contributions and community participation - Strategic Alignment: Align networking automation goals with broader platform integration objectives #### Community - Maintainers: Support SONiC maintainers with testing resources and feedback - Contributors: Create onboarding programs for networking engineers new to open source - Sponsors: Fund upstream SONiC development and community infrastructure - Ecosystem: Coordinate between ASIC vendors, ODMs, and software contributors ## 🧪 Hypotheses - Organizations pursuing SONiC primarily for cost savings will achieve less value than those focused on operational automation - Upstream SONiC adoption will accelerate as community development capacity exceeds traditional vendor engineering resources - ASIC vendor feature roadmaps drive white box switching capabilities more than software layer innovation - Partnership approaches will dominate over DIY implementations due to ecosystem complexity requirements ### 📊 Methodology - Survey SONiC adopters to compare cost savings vs automation benefits achieved - Analyze GitHub contribution patterns to measure community development velocity vs traditional vendor release cycles - Study ASIC vendor product announcements correlation with SONiC feature availability - Interview organizations to categorize implementation approaches and success factors ### ❓ Questions - How can smaller organizations access white box switching benefits without high-volume ASIC vendor targeting? - What governance models best ensure enterprise contributions flow back to upstream communities? - How do partnership agreements balance proprietary customization needs with upstream contribution requirements? ### 🔮 Future Research - Comparative analysis of SONiC community development velocity vs traditional networking vendor engineering capacity - Economic modeling of procurement optimization vs white box switching implementation costs - Technical assessment of ASIC vendor roadmap influence on open networking feature development ### 💡 Implications - Practice: Network engineers need broader ecosystem understanding beyond single-vendor relationships - Policy: Open source networking policies should mandate upstream contribution requirements - Security: Distributed ecosystem creates more complex vulnerability and update management requirements - Education: Networking curricula must include ASIC vendor relationships and community development models - Funding: Investment in upstream communities may yield better returns than enterprise support contracts ### 🧠 Learning Styles Assessment Reading level estimate: Grade 12-14 (technical content with industry jargon) #### 👀 Visual learners - Present: None identified in video presentation - Improve: Ecosystem architecture diagrams, vendor relationship maps, decision tree flowcharts #### 🔊 Aural learners - Present: Full video presentation with expert narration - Improve: Podcast-format discussion with Q&A segment #### 🗣️ Verbal learners - Present: Technical definitions, industry examples, analogies to traditional networking - Improve: Glossary of ASIC vendor terminology, procurement vs technology comparisons #### 🧩 Kinesthetic learners - Present: None identified - Improve: SONiC deployment labs, vendor evaluation checklists, ecosystem mapping exercises #### 🧠 Logical learners - Present: Clear decision framework, cost-benefit analysis, systematic approach comparison - Improve: Quantified ROI models, decision matrices, risk assessment frameworks #### 🧑‍🤝‍🧑 Social learners - Present: Conference presentation format with community context - Improve: Discussion forums, peer experience sharing, community contribution guides #### 🧘‍♀️ Solitary learners - Present: Self-contained presentation covering complete decision framework - Improve: Self-assessment questionnaires, individual planning worksheets, reflection prompts #### 🌿 Naturalistic learners - Present: Ecosystem thinking, supplier relationship patterns - Improve: Industry lifecycle analysis, technology evolution patterns, market dynamics ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Video has spoken content but no visual elements described - Improve: Audio descriptions of any slides or visual elements, structured transcript with headings #### Keyboard navigation - Present: Standard YouTube player controls - Improve: Chapter navigation, searchable transcript interface #### Contrast and readability - Present: YouTube's default player interface - Improve: High contrast player options, customizable text size for transcripts #### Text resizing and reflow - Present: Responsive YouTube interface - Improve: Dedicated transcript page with text customization options #### Media and captions - Present: Audio content available - Improve: Professional captions, downloadable transcript, multiple language options #### Forms and interactions - Present: Basic video controls - Improve: Accessible video bookmarking, note-taking interface integration #### Standards reference - Video content needs WCAG 2.2 AA compliance for captions and audio descriptions ### 🌍 DEIB Assessment #### Cultural breadth - Present: US-centric examples (CenturyLink, Mercedes-Benz) - Improve: Global networking infrastructure examples, international regulatory considerations #### <!-- deep-read:sections end -->