Centralised Observability: A Strategic Approach to Safeguarding Business-Critical SLAs

By Vincent Hsu, IBM Fellow, Vice President & CTO for IBM Storage Infrastructure and Sandeep Patil, IBM STSM, Master Inventor

In today’s digital world, the Service Level Agreements (SLAs) of applications play a critical
role for any business. These SLAs, directly impact customer satisfaction, operational
efficiency, and, ultimately, the bottom line. However, meeting SLAs for business applications directly depends on the underlying infrastructure, be it in hardware or software-defined form factors. With the digitalisation of businesses, the reliability and performance of applications have become crucial for success. SLAs that ensure uptime, availability, and data integrity must be upheld and this responsibility also trickles down to the infrastructure layer. Among these infrastructure components, storage, which hosts the invaluable business asset – the data, plays a key role.

To ensure adherence to committed business SLAs, it is imperative to establish a centralised
observability framework that includes the entire infrastructure stack, with a particular focus
on the data repositories. This approach becomes even more critical in today’s hybrid cloud environment, where workloads can seamlessly transition between on-premises and cloud-based environments. Observability of the infrastructure across this diverse fleet is a
challenging but essential task.

Understanding Centralised Observability of Data Repositories
Centralised observability of data fabric across hybrid cloud deployments involves real-time
monitoring, data collection, co-relations, analysis, and visualisation of various metrics and
events related to storage systems, which is a key data repository component of the overall
data fabric. These metrics encompass aspects like I/O performance, cache hit ratio, latency,
throughput, capacity utilisation, availability, and the health of storage systems and their
fabric. By gathering and analysing this data, IT teams can gain insights into the health and
performance of their storage infrastructure, thus enabling the SLA adherence of the data-
dependent business applications.

Storage Observability and Application SLAs
The relationship between storage observability and application SLAs is symbiotic. Here’s how they are intertwined:

 Performance Impact: Inefficient storage can impact application performance, leading to SLA violations. Observing and analysing storage metrics helps identify performance bottlenecks and optimise configurations to meet application demands.

 Data Availability: High availability and data redundancy are crucial for ensuring that applications can access their data when needed. Storage observability helps prevent data loss and maintain application availability.

 Capacity Planning: Predictive analysis of storage metrics allows organisations to plan for future capacity needs and better plan their OPEX or CAPEX for applications.

 Data Security: Observing data security metrics is vital for protecting sensitive data. Unauthorised access to storage or data breaches can have severe consequences for both applications and SLAs.

 Fault Tolerance: Storage observability and insights can help detect and mitigate hardware failures or software errors promptly, reducing the risk of application downtime.

Trends in Storage Infrastructure Observability
Under the umbrella of AIOps, machine learning models are increasingly being used to analyse vast amounts of storage data in real time. These technologies can identify patterns, and anomalies, and predict potential issues before they impact SLAs. Further, observability tools are becoming central to the DevOps pipeline. This enables quicker response times to
identify issues in the overall stack and improve collaboration between development and
operations teams.

With the adoption of hybrid multi-cloud deployments and the availability of software-defined storage, storage observability solutions are evolving to provide visibility across multiple cloud providers and on-premises infrastructure, ensuring consistent monitoring and management.

Another promising trend in the field of infrastructure observability (applicable to storage observability) is the integration of Generative AI. This technology is poised to play a pivotal role, empowering administrators and Site Reliability Engineering (SRE) teams to enhance observability, monitoring, and problem determination through conversational interfaces to meet the eventual goal of ensuring business SLAs.

In conclusion, the trends in the digital transformation space along with the influence of AI, indicate a growing need for intelligent, coherent, and multi-cloud-compatible observability technology to stay ahead in the dynamic world of IT infrastructure and asset management.