Emmanuel Egede

Database passwords, API keys and tokens don't belong in application code. Separate the secret from the code and control which workload can retrieve it. #AWS #CloudSecurity

Admin in development doesn't automatically mean admin in production. Cloud access should follow responsibility. Environment separation can help enforce that boundary. #AWS #IAM

Dev, test and production don't always need to live in one AWS account. Account separation can create stronger boundaries around workloads with different risks. #AWS #CloudSecurity

EC2 has almost no CPU usage for months. Terminate it? Not yet. Check ownership, dependencies, traffic and purpose first. Clean up with evidence, not assumptions. #AWS #EC2

AWS tags aren't just labels. They can tell me: What is this resource? Who owns it? Which environment? How is it managed? Cloud infrastructure should be understandable after deployment too. #AWS #CloudOperations

Snapshot. AMI. Terraform. Documentation. They don't all solve the same problem. Good recovery planning means knowing what must be restored, what must be recreated and what the workload depends on. #AWS #CloudOperations

An EC2 instance is working perfectly. Good. Now ask: Could I rebuild it if I lost it today? Running infrastructure and recoverable infrastructure aren't always the same thing. #AWS #CloudInfrastructure

Before using 0.0.0.0/0, ask: Who actually needs access? A rule should match the requirement, not simply make the error disappear. #AWS #CloudSecurity

AWS traffic is blocked? Don't start opening ports everywhere. Follow the path. Source → destination → port → Security Group → NACL. Find where the expected traffic stops. #AWS #CloudNetworking

If an EC2 workload needs AWS access, I don't immediately ask where to store the access key. I ask: Can I use an IAM role instead? Temporary credentials. Required permissions. No hard-coded keys. #AWS #CloudSecurity

EC2 is healthy. Application is running. Security Group looks right. Still no connectivity? Follow the route. Sometimes the server isn't broken. The traffic simply has no valid path. #AWS #CloudNetworking

Two private subnets across two AZs. Both use a NAT Gateway in AZ-A. It works. But what happens to AZ-B's outbound connectivity if that dependency becomes unavailable? Follow the entire path. #AWS #VPC

Cross-AZ NAT can work. But I still ask: What's the traffic path? What's the resilience requirement? What's the cost? “Working” and “well-designed” aren't always the same thing. #AWS #CloudNetworking

EC2 instances in two AZs don't automatically make the whole architecture highly available. Follow the dependencies too. If both private subnets depend on one NAT path, that's worth examining. #AWS #CloudInfrastructure

S3 says AccessDenied. Don't immediately add broad permissions. Check the identity, requested action, IAM policy and bucket policy first. Fix the cause, not just the error. #AWS #IAM

Before making an S3 bucket public, ask: Does it actually need to be public? Start restricted. Open only what the requirement calls for. #AWS #S3 #CloudSecurity

High availability isn't just “add another server.” Ask: What happens if one instance fails? What about an Availability Zone? Can traffic move? Can capacity recover? Design for failure, not appearance. #AWS #CloudInfrastructure

An EC2 instance can be running perfectly and still have no internet access. Before blaming the server, check the network. Route tables. NAT Gateway. Security Groups. Network ACLs. Cloud troubleshooting starts with understanding the traffic path. #AWS #CloudNetworking #EmmanueleTech

Private EC2 needs access to S3. No public IP. No general internet access wanted. What would you consider? S3 VPC endpoint. Private doesn't mean disconnected. #AWS #CloudSecurity

Before adding a NAT Gateway, I ask: Does this traffic actually need internet access? For supported AWS services, a VPC endpoint may provide a more direct private path. Understand the traffic first. #AWS #VPC

People aren't the only identities accessing systems. Applications and services need permissions too. A service account shouldn't have administrator access just because it's convenient. Give it only what it needs. #ITAdministration #Security

Windows shows 169.254.x.x. That's a clue. The device didn't receive the IPv4 configuration it expected from DHCP. Now I'd investigate why before changing unrelated settings. #DHCP #Troubleshooting

User is connected but can't reach anything. Before blaming the router, check: IP address Subnet Gateway DNS DHCP The client's configuration can tell you where to start. #Networking #ITAdministration

You find an undocumented server that's been running for 3 years. Nobody knows what it does. Would you shut it down? I wouldn't not until I understand its services, traffic and dependencies. #ITOperations #Infrastructure

An old resource looks unused. Don't immediately delete it. First ask: Who owns it? What depends on it? When was it last used? Clean up with evidence, not assumptions. #Infrastructure #AssetManagement

You can't properly manage what you don't know you have. Servers, devices, domains, licenses and cloud resources all need ownership and purpose. Good inventory is part of good IT operations. #ITAdministration #ITOperations

Database fails at 4 PM. Last usable backup: midnight. The backup works, but 16 hours of changes could be lost. The backup succeeded. But was the recovery strategy good enough? #DisasterRecovery #ITAdministration

How often should you back up a system? That depends on a better question: How much data can the business afford to lose? Let the requirement drive the technology. #Infrastructure #ITOperations

A green “backup successful” message doesn't prove you can recover. The real test is whether the data can actually be restored when you need it. Backup is preparation. Recovery is proof. #ITOperations #Backup

App was working yesterday. Today it's unreachable. Servers are healthy, but someone changed the network configuration last night with no documentation. Before changing anything else, I'd find out exactly what changed. What would you check first? #Networking #Troubleshooting

When something breaks, one of my first questions is: What changed? Knowing what changed, who changed it, when and why can save hours of troubleshooting. #ITOperations #ChangeManagement

User can log in and view a file but gets Access Denied when editing it. Would you reset the password? I wouldn't. Authentication already worked. I'd investigate authorization. #Troubleshooting #ITAdministration

A user's access may come from a group, role, policy or inherited permission. Don't only check the account. Check where the effective permission actually comes from. #AccessManagement #ITOperations

Authentication asks: Who are you? Authorization asks: What are you allowed to do? A successful login doesn't automatically mean the user has permission to perform an action. #ITAdministration #CloudSecurity

Server is running. Application is running. Public IP hasn't changed. But the domain doesn't resolve to it. Where do you investigate first? I'd start with DNS. #CloudInfrastructure #Troubleshooting

A website being down doesn't automatically mean the server is down. Check the path: domain → DNS → IP → server → application. Troubleshoot the layer that's actually failing. #DNS #Networking

An employee changes departments but still has access to their old department’s systems. Nothing is broken. But is that still a security issue? I think it is. #ITSecurity #EmmanueleTech

Access shouldn’t be permanent just because it was appropriate when it was granted. Roles change. Responsibilities change. Permissions should be reviewed too. #AccessManagement #EmmanueleTech

Creating a user account is easy. The bigger responsibility is managing that access as the person's role changes and eventually removing it when it is no longer needed. #ITAdministration #EmmanueleTech

An application slows down every weekday around 9 AM, then recovers 30 minutes later. I wouldn’t immediately resize the server. That timing is a clue. I’d investigate what regularly happens during that window. #Troubleshooting #EmmanueleTech

A metric is more useful when you know what normal looks like. CPU at 70 percent means more when you know whether the workload normally runs at 20 percent or 65 percent. Baselines matter. #CloudOperations #EmmanueleTech

When an application struggles, adding more CPU or memory might help. But I first want to know why it is struggling. More resources can sometimes hide the real problem. #CloudInfrastructure #EmmanueleTech

Good infrastructure shouldn’t depend on one person remembering everything. Another engineer should be able to understand the environment, the major decisions and how it is managed. #CloudInfrastructure #EmmanueleTech

Users suddenly can’t access an application. I wouldn’t start making random changes. Confirm the impact, check logs and metrics, review recent changes and communicate while investigating. Calm troubleshooting matters. #IncidentResponse #EmmanueleTech

Incident response isn’t only technical. While you investigate the problem, people also need to know what is happening. You may not have the answer yet, but you can communicate what you know. #ITOperations #EmmanueleTech

Five people need access to the same system. Would you make all five administrators? I wouldn’t. I’d first understand what each person actually needs to do and assign access around those responsibilities. #CloudSecurity #EmmanueleTech

I don’t think everyone needs administrator access just because it makes things easier. Access should match responsibility. Give people what they need to do their work without giving away unnecessary control. #AccessManagement #EmmanueleTech

I prefer individual accounts over shared credentials. When everyone has their own identity, it becomes easier to understand who has access, who made a change and whose access should be removed. #CloudSecurity #EmmanueleTech

You inherit an AWS environment you didn’t build. Everything works. Would you start changing things? I wouldn’t. First I’d understand the architecture and trace how traffic, routing and access work. Understand before you optimize. #CloudInfrastructure #EmmanueleTech

I like asking why before how. How do I configure it? Important. Why does the architecture need it? Just as important. Knowing how helps you build. Understanding why helps you design. #CloudEngineering #EmmanueleTech