Troubleshooting The Cloud
Troubleshooting the Cloud.
Troubleshooting in a cloud environment utilizes the same concepts as physical or on-premises problem-solving, but many additional variables exist. Downtime may involve network connections, CSP outages, or security-specific access control misconfigurations. However, the approach of identifying the problem, understanding its cause, and implementing a solution still stands.
Troubleshooting Methodology.
The following list represents the basic steps in a troubleshooting methodology. Let's examine these steps in more detail.
Identify the Problem
Discovered by users, logs, or alerts. Ask: "If it worked yesterday, what changed?" Consider infrastructure, IT, and facility/environmental changes.
Determine Scope
Gather info, replicate the issue, and back up data if at risk. Is it one user or many? One location or several? This isolates the fault domain.
Establish Theory
Question the obvious and keep it simple. Don't assume complexity just because the cloud is complex. Identify common elements across services.
Test the Theory
Verify the likely cause. You may solve simple problems here. If confirmed, move to the plan. If not, establish a new theory.
Plan of Action
Recognize that interruptions and data loss must be avoided. Define steps ahead of time and notify end-users with expected downtime.
Implement or Escalate
Follow the plan. Make only one change at a time and test. Reverse if ineffective. Escalate to the CSP or MSP if you lack the knowledge.
Verify Functionality
Test the service or web application to ensure it has returned to established service levels.
Preventive Measures
Reconfigure to avoid repeats. Implement RAID, backups, scalability, or high availability.
Root Cause Analysis
Evaluate why the problem occurred to change processes and avoid future instances.
Document Findings
Maintain documentation throughout the entire process—not just at the end. Log symptoms, research, and outcomes.
Some service desk management software requires the use of tickets. This software may require that troubleshooting documentation be entered before closing the ticket.
Policies, Procedures, & Impacts.
Carefully consider corporate policies governing deployment, networking, and security troubleshooting. Standard operating procedures (SOPs) govern how to accomplish tasks and should be strictly followed. For example, adhere to IAM policies to avoid granting too much or too little access.
Cloud policies manage case escalation. Escalating to a CSP's technical support may incur a cost, so administrators must have a complete view of the environment before escalating upward.
Service-Level Agreements (SLAs) might enforce penalties on your organization for outages. Conversely, if an outage falls within the CSP's responsibility, your organization may be awarded reduced service fees or compensation.