As decentralized and complex systems shape the landscape, site reliability engineering (SRE) practices are evolving to meet the challenges posed by this paradigm shift. The recent SRE Report 2024, a comprehensive survey-based exploration conducted by Catchpoint, provides insights into the dynamic nature of SRE practices and the key considerations influencing the reliability landscape.
Catchpoint's sixth SRE Report uncovers findings from a survey of more than 400 IT professionals globally. Specifically, the report delves into the adjustments and adaptations SRE practices are undergoing in response to organizations recognizing the need to extend their monitoring and learning purview beyond directly controlled elements and embrace third-party services and endpoints. This shift reflects a departure from limiting scope to first-party services toward a more federated approach, compelling organizations to reimagine reliability within the context of distributed architectures.
The report highlights several key insights that illuminate the current landscape of SRE practices:
Decentralized Monitoring
64% of organizations believe that SRE practitioners should monitor endpoints outside their direct control, such as third-party services, indicating a growing emphasis on critical visibility beyond organizational boundaries.
Tool Diversity
66% of organizations utilize two to five monitoring tools, aligning with their staff size and unique capabilities. With 81% of organizations having multiple telemetry types feeding their observability frameworks, this underscores the recognition that a varied toolkit enhances the ability to address the complexity of modern architectures.
Structural Evolution
51% and 44% of companies structure their teams by product or service, or by platform or capability, respectively. The use of these structures trends upward with larger company sizes, reflecting the evolving nature of roles and team structures within the SRE domain.
Learning from Incidents (LFI)
LFI emerges as a universal business opportunity, with 52% acknowledging the need for improvement in reviewing major incidents, irrespective of company size.
Artificial intelligence (AI) in SRE
While 53% anticipate AI making work easier in the next two years, mixed views on AI's usefulness are evident across different organizational ranks. Management leans towards AI for potential cost savings, whereas individual contributors express reservations, citing a preference for pride in their work over efficiency.
As the SRE landscape continues to evolve, practitioners anticipate significant challenges in the coming years. Balancing costs, time constraints, aligning ranks, and navigating the complexities of decentralized systems are identified as prominent challenges. Resource constraints, particularly concerns related to cost or budget (44%), underscore the need for organizations to explore monitoring elements outside their direct management, including content delivery networks (CDN) and domain name systems (DNS).
Learning from incidents has also emerged as a focal point for improvement across organizations. The report underscores the need for organizations to dedicate protected time to learn from both major and non-major incidents, emphasizing that each presents a valuable learning opportunity. With 71% of respondents dealing with dozens or even hundreds of non-ticketed incidents monthly, the need for refining blameless feedback loops is critical to fortify the resilience of companies over time.
But one of the biggest takeaways from this year’s report is the nuanced perspectives regarding the role of AI in SRE. Views on AI's usefulness are notably influenced by organizational rank. Management and leadership view AI as a potential avenue for cost savings, considering its application in reducing headcount or accelerating time to market. In contrast, individual contributors exhibit a less positive sentiment, emphasizing the importance of being proud of their work over operational efficiency. This divergence in mindset is expected to drive mixed views on the integration of AI in SRE. However, survey respondents identified GenAI as a promising application, although this perception may be influenced by the overarching hype surrounding AIOps.
It is evident that SRE practices are at a crossroads, adapting to the demands of decentralized systems and the evolving expectations of practitioners across various organizational ranks. The report serves not only as a reflection of current practices but also as a guide for the future, offering insights into potential challenges and opportunities on the horizon. As organizations grapple with the intricacies of decentralized architectures, the SRE landscape continues to evolve, driven by a collective commitment to reliability, learning and the pursuit of excellence.