Tech Jargon - Decoded

What is Telemetry? Telemetry is the automatic collection and transmission of data from a remote or "hidden" source to a central system for monitoring and analysis. How it works: • Capture: Software agents or hardware sensors record specific data points (like voltage, errors, or speed). • Packing: This raw data is bundled into small packets. • Transmission: The packets are sent over a network (wired or wireless) to a receiving station. • Display: A central server decodes the data and presents it on a dashboard. Problem it solves: It eliminates the "visibility gap." Instead of guessing if a remote system is working or manually checking it, you get real-time updates on its health and performance. Use Case Scenario: A system is running in a location you cannot access. It sends its internal temperature and "heartbeat" signal to your screen. If it gets too hot, you know immediately and can fix it from your desk.

What is Observability? It is the ability to understand a system's internal state just by looking at the external data it produces. It allows you to ask questions about your system’s health without knowing the problems in advance. How it works: It collects and correlates three main types of data: • Metrics: Values like CPU usage or memory levels measured over time. • Logs: Timestamps and text records of specific events or errors. • Traces: The path of a single request as it moves through different services to show where delays occur. Problem it solves: Monitoring only tells you "if" a system is down. Observability solves the problem of "why" it is down, helping you find root causes in complex setups where the source of a failure is hidden. Simple Scenario: If a specific user's request fails, you can use traces to see exactly which service blocked it and check the logs of that service to see the specific error message generated at that moment.

What is CI/CD? CI/CD is an automated workflow that bridges the gap between writing code and making it live for users. The Problem it Solves: • Eliminates manual errors during code transfers. • Fixes "integration hell" where code from different developers doesn't work together. • Removes the need for slow, manual testing before every release. How it Works: • Continuous Integration (CI): Every time you push code to a repository, a server automatically triggers a "build" and runs tests to ensure the new code hasn't broken anything. • Continuous Deployment (CD): Once the code passes all automated tests, the system automatically pushes it to the production server. Simple Scenario: A developer updates a feature and pushes the code. The pipeline automatically checks for bugs and updates the live website immediately without any manual file uploading.

What is DevOps? It is a practice that merges software development (Dev) and IT operations (Ops) to shorten the systems development life cycle. It focuses on automation and frequent communication between teams. How it works: • Continuous Integration: Developers merge code changes into a central repository where builds and tests are run automatically. • Continuous Deployment: Once code passes tests, it is automatically pushed to production environments. • Infrastructure as Code: Servers and networks are managed using configuration files instead of manual setup. Problem it solves: • Stops the "it works on my machine" conflict between developers and admins. • Prevents slow, manual release cycles that delay new features. • Reduces bugs by catching errors early through automated testing. Use Case: A developer updates a login feature. The system detects the change, builds the app, runs security tests, and deploys it to a live server in minutes without any manual intervention.

What is Event-Driven Architecture? It is a design pattern where software components communicate by producing and reacting to events. An event is a message representing a specific change in state. How it works: • Producer: A service that performs a task and emits a message (event) to a broker. • Event Broker: A middleman that receives events and keeps them in a queue. • Consumer: A service that "subscribes" to the broker. It stays idle until an event arrives, then processes it immediately. Problems it solves: • Tight Coupling: Services don't need to know who receives the data. • Sync Bottlenecks: The sender doesn't stop and wait for the receiver to finish its job. • System Failures: If a consumer is down, the broker holds the event until it comes back online. Simple Scenario: An Auth Service emits a "UserCreated" event. The Email Service and Profile Service both detect this event and run their internal logic independently.

What is Fault Tolerance? It is the design that allows a system to continue operating without interruption, even if one or more of its components fail. How it works: • Redundancy: Keeping duplicate hardware or software parts ready as backups. • Monitoring: The system constantly checks the health of its components. • Failover: When a fault is detected, the workload is instantly shifted to a backup component. • Isolation: It ensures that a failure in one part doesn't spread to the whole system. Problem it solves: • Prevents total system crashes during hardware or software errors. • Eliminates the "Single Point of Failure." • Ensures high availability and prevents data loss. Simple Scenario: In a dual-disk system, if Disk A fails, the system immediately switches to Disk B. The operation continues smoothly without the user noticing any service break.

What is High Availability (HA)? High Availability is a system design that ensures your application remains operational and accessible even if some of its components fail. It aims for "zero downtime" by removing single points of failure. How it works: • Redundancy: You run multiple identical copies of your servers or databases simultaneously. • Health Monitoring: A controller constantly checks if each copy is working correctly. • Failover: If one server crashes, the system automatically detects it and redirects all traffic to the healthy servers. Problem it solves: It solves the issue of unexpected downtime. Without HA, a single hardware crash or software bug can take your entire service offline for hours. Simple Scenario: Imagine a setup with two servers and a Load Balancer. If Server A breaks, the Load Balancer stops sending traffic there and routes everything to Server B. The users stay connected and never notice a problem.

What is Scalability? It is the ability of a system to handle an increasing amount of work by adding hardware resources without crashing or changing the code. How it works: • Vertical Scaling: You increase the power of a single server by adding more RAM, CPU, or Storage. • Horizontal Scaling: You add more servers to the network to share the workload. • A load balancer is used to distribute incoming requests across these multiple servers. Problem it solves: • Prevents "Server Down" errors during high traffic. • Stops slow response times and system lag. • Ensures the system stays available even if one machine fails. Simple Scenario: If a server is built to handle 500 users but 2,000 users try to connect, the system automatically triggers more server instances to run in parallel to process the extra data.

What is Load Balancing? It is a technique to distribute incoming network traffic across multiple backend servers to ensure no single server carries too much load. How it works: • A Load Balancer sits between the user and the server fleet. • When a request arrives, the balancer picks an available server using an algorithm (like Round Robin). • It forwards the request to that server and manages the response flow. Problem it solves: • Stops servers from crashing during high traffic peaks. • Eliminates downtime; if one server fails, the balancer moves traffic to healthy ones. Scenario: Your site gets 1,200 hits at once. Instead of one server handling all and slowing down, the balancer sends 400 hits to Server A, 400 to Server B, and 400 to Server C.

What is Load Balancing? It is a mechanism to distribute incoming network traffic across a group of backend servers to ensure no single server is overwhelmed. How it works: • The balancer acts as a single point of entry for all requests. • It receives a request and selects a server from the pool based on an algorithm (like Round Robin). • It checks the health of each server; if one is unresponsive, it stops sending traffic there. • It forwards the data to the chosen server and sends the response back to the user. Problem Solved: Prevents system crashes caused by high traffic and eliminates a single point of failure. Scenario: Instead of 10,000 requests hitting one server and crashing it, the balancer sends 5,000 to Server A and 5,000 to Server B to keep the system stable.

What are Microservices? Microservices is an architectural style where a single application is built as a set of small, independent services. Each service runs its own process and focuses on a single business task. How they work: • Each service is self-contained and usually has its own dedicated database. • Services talk to each other using lightweight protocols like HTTP/REST or message queues. • They can be written in different languages and deployed on different servers. Problems solved: • Single Point of Failure: If one service crashes, the rest of the system stays online. • Scaling Issues: You can scale only the specific service that is under heavy load. • Slow Deployment: You can update one part of the app without redeploying everything. Simple Scenario: In a system, the "Payment" service and "Inventory" service are separate. If the Payment service needs an update, you can take it offline and fix it while users can still browse the Inventory without issues.

What are Smart Contracts? They are self-executing bits of code stored on a blockchain that automatically run when set conditions are met. How it works: • Code is written with "If/Then" logic. • It is uploaded to a blockchain network. • Once deployed, it lives at a specific address. • When a user sends a transaction to that address, the code triggers. • Network nodes execute the logic and update the database permanently. Problems solved: • Middlemen: Removes the need for lawyers or banks to oversee deals. • Tampering: The code cannot be changed or stopped once it is live. • Human Error: Execution is math-based, not manual. Simple Scenario: A contract holds 100 tokens. The logic says: "If User A sends 1 ETH, release 100 tokens." The moment the ETH hits the contract address, the swap happens instantly. No one can block the transfer.

What is Blockchain? A decentralized digital system that records data in a sequence of linked blocks across a global network of computers. How it works: • Data is bundled into a Block. • Each block gets a unique Hash (a digital ID) and stores the Hash of the block before it. • This creates a mathematical link between them. • Every computer in the network holds an identical copy of this chain. • If someone tries to change a block, its Hash breaks, and the network rejects the change. Problem it solves: It removes the need for a central authority or middleman. It ensures data cannot be deleted or edited once it is written. Scenario: When a transaction occurs, the network validates it. Once recorded in a block, it is permanent. If a user tries to alter their balance, the other computers compare their copies, see the mismatch, and block the fraud automatically.

What is Edge Computing? Edge computing is a method of processing data physically closer to where it is created, rather than sending everything to a distant central server or the cloud. How it works: • Local devices collect raw data from sensors. • This data is processed immediately on a nearby "edge node" (a local computer or router). • Critical actions are taken on-site in real-time. • Only a small summary of the data is sent to the main cloud for storage. Problems solved: • High Latency: Eliminates the delay caused by data traveling long distances. • Bandwidth Pressure: Prevents network clogging because massive amounts of raw data stay local. Simple Scenario: Imagine a pressure valve. If it reaches a dangerous limit, a local edge device detects it and shuts it off instantly. It doesn't wait for a signal from a far-away server to tell it to stop.

What is Edge Computing? It is a distributed computing model where data processing happens at the "edge" of the network, physically close to where the data is created, instead of a distant central cloud. The Working: - Local devices collect raw data. - Instead of sending everything to a far-away server, a local "edge node" (small processor or server) computes the data on-site. - Actions are taken instantly based on this local analysis. - Only the essential results or summaries are sent to the main cloud later. Problems it Solves: - Latency: Removes the delay of data traveling thousands of miles and back. - Bandwidth: Reduces network traffic by not uploading massive amounts of raw, useless data. Simple Scenario: A sensor tracks pressure. Instead of sending every second of data to a remote server, the local chip processes it. If pressure gets too high, it shuts the valve immediately without waiting for instructions from the internet.

What is Serverless computing? It is a cloud model where you write and deploy code without managing any underlying hardware. Even though servers still exist, the cloud provider handles all the server maintenance, scaling, and setup for you. How it works: • You write your logic as small, independent functions. • The code stays idle until a specific "event" (like an API request) occurs. • Upon the event, the provider spins up a temporary container to run your code. • Once the task is finished, the container is destroyed. Problems it solves: • Idle Costs: You don't pay for servers sitting empty; you only pay for the milliseconds your code runs. • Scaling Issues: It automatically handles 1 request or 1,000 requests without manual configuration. • Ops Overhead: No need to worry about OS updates or security patches. Simple Scenario: A user uploads an image. This event triggers a function that automatically shrinks the image size. Once the resized image is saved, the function stops instantly.

What is Cloud Computing? Cloud computing is using remote servers hosted on the internet to store, manage, and process data, rather than using your own local hard drive or a physical server in your room. How it works: - Large providers maintain massive data centers filled with physical hardware. - They use software to "slice" these physical machines into virtual parts. - You connect to these slices via the web. - When you upload a file or run a program, your request travels over the internet to their hardware, which does the work and sends the result back. Problems it solves: - No need to buy expensive physical hardware. - Fixes the issue of running out of storage space. - Removes the effort of manual hardware maintenance and cooling. Use Case: A developer needs a high-power server for just 2 hours to test some code. Instead of buying a machine, they rent a virtual one online, use it, and then delete it.

What is Data Science? It is the process of extracting meaningful insights and hidden patterns from large sets of raw, unorganized data. It turns messy information into clear, actionable logic. How it works: • Collection: Gathering raw data points into one place. • Cleaning: Removing errors, duplicates, and empty values to ensure accuracy. • Analysis: Using math and programming to find how different variables relate to each other. • Modeling: Creating a logic-based map to predict future outcomes based on past data trends. Problem Solved: It solves the issue of "Information Overload." Humans cannot manually process millions of data rows to find hidden links; this process automates discovery and reduces human error. Simple Use Case: A system looks at thousands of past transactions. It notices that whenever Factor A and Factor B happen together, Event C usually follows. It then flags this pattern automatically for the user.

What is MLOps? MLOps (Machine Learning Operations) is a set of practices used to automate the process of moving a machine learning model from a developer's laptop to a live, working system. How it works: • Automation: It uses scripts to handle data cleaning and model training automatically. • CI/CD: It tests the code and deploys the model to servers without manual intervention. • Versioning: It tracks every version of the model and data so you can roll back if something breaks. • Monitoring: It constantly checks the model's performance. If accuracy drops, it triggers a retrain. Problems it solves: • Eliminates slow, manual deployment processes. • Fixes "Model Drift" where old models stop working well on new data. • Ensures the model works the same in production as it did during testing. Simple Scenario: When new data is collected, the system automatically detects it, trains a new version of the model, and replaces the old one on the server without any human clicking "upload."

What is Fine Tuning? It is the process of taking a model that already has general knowledge and training it further on a smaller, specific dataset to make it an expert in one area. How it works: • Start with a model that has existing weights and parameters. • Feed in specialized data through the existing layers. • Run training cycles to slightly adjust those internal parameters. • The model updates its logic to favor the new data patterns while keeping its core foundation. Problem it solves: Training a model from scratch is extremely slow and requires massive data. Fine tuning allows for high accuracy using minimal data and less computing power. Simple Scenario: A system knows how to identify general shapes. To make it identify specific medical tools, you show it only tool data. It uses its "shape logic" to quickly learn the tool details without relearning what a line or curve is.