Adaptive Tactics in Software Operations: How People Manage Complexity by Asking for Help

Sammanfattning: Society relies on the continued operation of software systems, from healthcare to social media platforms. These systems can be engineered to be highly reliable, but complexity assures surprise. People have the ability to manage this complexity, filling gaps in design by recognizing, diagnosing and correcting anomalies. Rarely can surprises be managed by individuals alone. When engaged in anomaly response, people ask for help, recruiting others into the response. This study explores recruitment – what criteria influence people to recruit additional resources, who do they recruit, why do they recruit them, and what methods do they use for recruitment? A case study approach was used, focusing on instances of recruitment across four cases of anomaly response in software operations. Data were collected in the form of chat transcripts, and records from monitoring and alerting tools and code repositories. These data were used to inform semi-structured interviews and cued recall with participants to elicit perspectives related to decision-making around specific instances of recruitment. In attempting to maintain operations, individuals sought help for a variety of reasons: diagnosis, repair, coordination, cross-checking, information gathering and approval. Recruitment occurred across a range of roles, and requests for help occurred throughout the response. People also employed various strategies in their attempts to recruit others, such as communicating urgency by sharing a “scary” graph and leveraging the organizational structure to identify who might be available to help. In all cases, restoring system operations required recruiting additional people into the response, and this recruitment was enabled by the accumulated experiences and relationships of the recruiters.