RAS – Reliability, Availability, and Serviceability

Overview of the RAS system to monitor and maintain L300 and L300N systems

This section provides an overview of the RAS system to monitor and maintain ClusterStor™ L300 and L300N systems.

Reliability, Availability, and Serviceability (RAS) is a set of attributes that reflect the robustness of a hardware and/or software system.

  • Reliability is a function of time that expresses the probability at a given future time that a system will still be working, given that it is working at the present time
  • Availability is the measure of how often the system is available for use (such as a system’s up-time percentage). Availability and reliability may sound like the same concept, but they are different, as a system can have great availability but offer no reliability
  • Serviceability is a broad definition that describes the ease of system service or repair

Reliability and Availability (RA)

Reliability and availability features keep a system working when failures occur. Examples of RA include:
  • Hardware redundancy (power, servers, networks)
  • Storage stacks (SCSI, RAID, local file system)
  • HA stack
  • Clustered file system

Serviceability

Serviceability consists of on-product capabilities and off-product tools, processes and staff. Serviceability supports the needs of the product (its RA model, for example); HPC and archive products will behave differently. There are cost constraints associated with serviceability. Most important, the customer should not be impacted; there should be no loss of availability for maintenance operations. Serviceability focuses on Level 1 service personnel and eventually, end users.

In L300 and L300N products, serviceability includes the following capabilities:
  • Inventory discovery (part/serial numbers, firmware versions, location)
  • Monitoring, diagnostics, fault isolation of
    • data collection from functional codes (RAID, HA, FS)
    • discovery and data collection from hardware (SES, IPMI, SNMP)
  • Reporting service manager policy engine (product-specific)
  • Reporting service notifications (user email, SNMP)
  • Reporting live event telemetry stream (interesting events)
  • Guided repair assistance for service
  • Supports disks, PSUs/PCMs and ESM controller server modules
Figure: RAS Service Notification Scenario
RAS for L300 and L300N provides an "expert" system:
  • Architected specifically for L300 and L300N
  • With topological awareness that enables correct issue prioritization
  • Designed to reduce false positives and repetitive alerts
  • With event-driven data end-to-end across the system, allowing real-time updates

The L300 and L300N RAS service console provides a real-time GUI with guided repair assistance.

Remote support includes email notifications, user alert emails, and telemetry data. RAS-based SUs are designed to be applied with no downtime/interruption.