Span and High-Performance Memory

In traditional C#, slicing a string or an array created a brand new copy of that data in memory. For high-performance parsers (like JSON or CSV readers), this literally wasted Gigabytes of RAM. Span<T> allows us to point to a "Window" of existing memory without copying a single byte.

1. What is a Span?

A Span<T> is a **ref struct**—a type-safe, stack-only pointer. It can point to an Array, an unmanaged memory block, or even data on the Stack.

string bigContent = "ID:54321;STATUS:ACTIVE";

// ❌ OLD WAY: Creates a new string "54321" on the heap
string id = bigContent.Substring(3, 5); 

// ✅ MODERN WAY: 'idSpan' is just a pointer to the middle of the original string!
// ZERO memory allocation.
ReadOnlySpan<char> idSpan = bigContent.AsSpan(3, 5);

2. Ref Struct Constraints

Because Spans are stack-only for performance, they have strict rules. You cannot use them inside an async method, you cannot store them as fields in a class, and they cannot be used in a List<Span>.

3. Memory<T>: The Async Alternative

If you need "Span-like" behavior but you are working in an async method, you use Memory<T>. It is a heap-based wrapper that can eventually be converted into a Span when it hits a synchronous method.

4. Interview Mastery

Q: "How did Span allow Microsoft to make .NET 8 several hundred percent faster than legacy .NET Framework?"

Architect Answer: "Before Span, the entire .NET runtime was addicted to 'Substrings.' Every time the framework parsed an HTTP header or a JSON key, it created thousands of tiny string objects. This caused massive Garbage Collector (GC) pressure. By rewriting the core libraries (JSON, HTTP, Regex) to use `ReadOnlySpan` and `ReadOnlySpan`, they eliminated millions of heap allocations per second. The CPU spent less time cleaning up garbage and more time executing business logic, leading to .NET consistently topping the TechEmpower performance benchmarks."