The tech community is actively discussing the economics of cloud computing versus on-premise infrastructure, sparked by 37signals' recent revelation of their successful cloud exit strategy. While cloud computing has been widely embraced as the default choice for many businesses, community discussions highlight that the emperor's new clothes might not be as magnificent as once thought, particularly for companies with predictable workloads.
The Real Economics of Going Cloud-Free
A key insight from the tech community reveals that 37signals' $700,000 investment in Dell systems represents a relatively modest infrastructure footprint - possibly just two full racks of servers. With modern colocation costs potentially running as low as $10,000 per month, the economics become particularly interesting when compared to their previous $3.2M annual cloud expenses.
Breaking Down the Numbers
- Previous AWS costs: ~$3.2M annually
- New hardware investment: $700,000 (one-time cost)
- Estimated colocation costs: ~$120,000 annually
- Current cloud expenses: $1.3M (temporary, due to AWS S3 contract)
- Projected five-year savings: >$10M
The Missing Pieces
Community discussions point out an important consideration absent from 37signals' public narrative: the human cost of managing on-premise infrastructure. However, the company's CTO David Heinemeier Hansson addressed this concern, noting that they maintained the same team size post-transition, contrary to initial speculation about needed workforce expansion.
The Cloud's Sweet Spot
The discussion has sparked a broader debate about when cloud computing makes financial sense. The consensus emerging from the tech community suggests that while cloud services excel for variable or unpredictable workloads, they may be unnecessarily expensive for steady-state, predictable operations - a perspective that challenges the current industry default of cloud-first thinking.
Looking Forward
37signals plans to further optimize their infrastructure by replacing their remaining AWS S3 storage (currently housing 10PB of data) with a dual datacenter Pure Storage configuration offering 18PB capacity. This move is expected to cost approximately the same as one year of S3 expenses, potentially leading to even greater long-term savings.
Industry Implications
While this success story is compelling, it's important to note that cloud repatriation projects remain relatively rare, with analysts estimating that only a single-digit percentage of companies are actively moving workloads back on-premise. The decision ultimately depends on each organization's specific circumstances, including existing infrastructure, workload patterns, and technical expertise.
This case study serves as a reminder that while cloud computing has revolutionized the tech industry, it shouldn't be treated as a one-size-fits-all solution. For companies with stable, predictable workloads and the technical capability to manage their own infrastructure, the traditional on-premise approach might still offer significant financial advantages.
The tech community is discussing the implications of cloud computing and on-premise infrastructure, highlighting trends and shifts in industry perspectives |