MongoDB ft. Dev Ittycheria - How an Early Pivot Catalyzed an Open Source Movement

Sequoia Capital

Original Platform Vision and Early Pivot (2007-2009)

Dwight Merriman, Kevin Ryan, Elliot Horowitz founded company originally called "10gen"
Initial vision: Platform as a Service (PaaS) with modern stack for cloud computing era
Built data layer called "mongodb" as subsystem within platform
Problem identified: Traditional databases (Oracle, Sybase) couldn't handle internet-scale applications
- PayPal and YouTube outages caused by database failures
- Required complex "sharding" and database tuning by specialized DBAs
MongoDB designed as document-based NoSQL database vs traditional SQL row/column structure
- Document model analogous to invoices containing all transaction data
- More flexible for adding new fields vs rigid relational schemas

Company released PaaS beta after 12 months, positive user feedback
Google App Engine launched simultaneously with similar concept
Dwight realized scope too large for startup: "10 years to build, billion dollars needed"
Key insight: Users specifically praised the database component
Decision to scrap entire platform business and focus solely on database
Elliot (CTO) initially shocked but agreed after one hour consideration
Kevin Ryan pragmatically supported if both technical founders agreed
Emotional impact: Laid off non-database team members, threw away application layer code
Started writing drivers for all popular programming languages

Launched MongoDB publicly in 2009 as open source
Grassroots marketing approach:
- Attended meetup groups in multiple cities
- Demonstrated MongoDB integration with Python, other languages
- Expanded to conferences in SF, SV, NYC, Europe, Asia
Intersection with "NoSQL" movement timing
Big companies (Amazon, Microsoft, LinkedIn, Netflix, Google) had internal NoSQL solutions
Sequoia invested late 2010 when company had 12 employees
Strong developer adoption, growing downloads month-over-month
Company renamed from 10gen to MongoDB in 2013

Revenue run rate ~$40M in mid-2014
Board decided to hire experienced public company CEO
Dave Ittycheria joined as CEO from venture investing background
Had previously founded company, led through IPO and acquisition
Company struggling despite momentum: missing plans, dysfunctional leadership, ineffective go-to-market
Strategic challenge: Customers using MongoDB but no ongoing relationship post-deployment
Identified need for cloud-managed service to remove operational burden

Board debate in December 2015: "do now or do later" for cloud service
Tom Killalea (former AWS executive) recruited to board despite initial skepticism
Concern about "jack of all trades" positioning vs AWS single-use-case approach
After meeting engineering team, Tom convinced of platform approach advantages
Decision to build Atlas cloud service with June 2016 launch deadline
"Startup within startup" approach with dedicated resources
Challenges:
- Building operational capabilities for database management
- Site reliability engineering skills
- New go-to-market model for SaaS vs on-premise
- Sales compensation changes
- Customer signup flow and conversion funnels

Atlas launched June 28, 2016 at MongoDB World conference
Started at zero customers, $100M company run rate
Dave isolated Atlas metrics to prevent being "drowned out" by main business
Went public in 2017 during business model transition
Public investor concerns:
- Atlas only 2-3% of revenue at IPO
- Competition fears with hyperscalers
- No analogous successful companies
- Enterprise adoption questions for cloud databases
Amazon DocumentDB launched 2019, stock dropped 15% on announcement
Management positioned as validation of document database market
Atlas growth trajectory: 10 years to $100M revenue, 5 additional years to $1B

Originally used AGPL (Affero GPL) license - more restrictive than standard GPL
Concern: Hyperscalers could "strip mine" MongoDB open source code
Fear of "game over" scenario if hyperscalers offered Atlas-like services
Board meeting in Las Vegas during AWS re:Invent 2018
Debated Server Side Public License (SSPL) adoption
SSPL requirement: Organizations offering MongoDB service must either:
- Obtain commercial license, OR
- Make their own source code publicly available
Risk: Customer backlash, community rejection, not officially OSI-sanctioned
Decision took "months and months" due to high stakes
Announcement October 2018 with mixed community reaction
Open source zealots rejected non-OSI license
Customer education campaign required
Atlas business grew faster post-license change

Revenue approaching $2B annually
Atlas represents 70% of MongoDB revenue, >$1B run rate
Customer requests driving platform expansion:
- Search capabilities alongside database
- Unified data platform vs multiple point solutions
Return to original platform vision but earned through database success
Developer data platform offering multiple tools and capabilities
Cross-deployment model: on-premise, cloud, cross-cloud, edge computing

Try for free