Why Google Relies on Software Engineers to Run Its Services: Inside SRE

Using Software to Replace Humans

Google describes its approach with the term Site Reliability Engineering (SRE). The core idea is to let software engineers, rather than dedicated operations staff, run services by building tools that automate operational tasks such as stability and performance maintenance.

Ben Treynor Sloss, a Google employee, wrote that the team prefers writing software to replace manual work.

The practice has become known as DevOps, linking developers and system administrators. Tools like Chef and Puppet evolved from this model, although Google kept its SRE approach relatively quiet for a decade.

Google now promotes SRE openly, publishing a book titled “Site Reliability Engineering” with the first chapter based on Sloss’s original paper.

Hegel’s Dialectic Theory

Sloss founded the SRE project at Google. Todd Underwood, an SRE director, notes that hiring software engineers for operations was natural because they could anticipate problems without wanting to handle them manually.

Adam Jacob, CTO of Chef, agrees that large companies need this shift, stating that connecting development and operations is inevitable.

Balancing Development and Operations

Google does not demand 100 % uptime; instead it uses an “error budget” to allow reasonable downtime, enabling teams to make changes and debug without fear.

Google limits SREs to spending no more than 50 % of their time on traditional operations work; if the balance tips, engineers are moved back to development tasks.

Hiring standards require 50‑60 % of SRE candidates to pass the same rigorous Google engineering interview, with the remainder possessing 85‑99 % of a typical engineer’s skill set plus deep knowledge of UNIX and networking.

SRE’s Ambition

The SRE philosophy draws inspiration from Margaret Hamilton’s work on the Apollo program, emphasizing proactive error detection and handling.

Underwood envisions a future where no one has to do manual operations, as automation handles reliability.