Don’t Use UUIDs/GUIDs in Databases. Use this Instead

Summary

In this video, Nick introduces the new UUID version 7 in .NET 9 preview 7, which is chronologically sorted and can be overridden with a time offset. However, he highlights the ULID (Universally Unique Lexicographically Sortable Identifier) as a more efficient alternative to traditional GUIDs, addressing fragmentation issues in distributed systems. Nick compares performance between ULID, GUID, and the upcoming UUID version 7, emphasizing ULID's compatibility, efficiency, and widespread implementation across various programming languages. He also mentions a new course on vertical slice architecture available on DotNetTrain.

Takeaways

• 🌟 Nick introduces a new type of GUID being added in .NET 9 called UUID version 7, which is chronologically sorted and takes time into account.
• 🔍 UUID version 7 can be created using the command 'Guid.Create(GuidVersion.V7)', which generates a GUID with the same initial bytes representing time in milliseconds.
• 📅 The time component of the new GUID can be overridden by providing a time offset, which is useful for specific time zone adjustments.
• 🆕 The script compares UUID version 7 with ULID (Universally Unique Lexicographically Sortable Identifier), which is an alternative that is widely used and efficient.
• 🔑 ULID offers a more character-efficient way of encoding 128 bits of randomness compared to UUID version 7, making it more space-efficient.
• 🚫 ULID version 1 and 2 are impractical due to the requirement of a unique stable MAC address, which is not commonly available.
• 🔄 ULID versions 3 and 5 are less common due to their reliance on Unix time and randomness, which can cause data structure fragmentation.
• 📈 ULID version 4, similar to GUID, provides randomness but lacks additional information, which can also lead to fragmentation.
• 🔗 ULID is designed to be unique per millisecond without needing to check for existence in a database, assuming non-existence due to the astronomically low chance of collision.
• 💻 ULID can be encoded as a 26-character string, which is more efficient than the 36-character UUID, and is case-insensitive and URL-safe.
• 📊 The script includes a performance benchmark comparing ULID with the traditional GUID and the new UUID version 7, showing ULID to be faster than both.
• 📚 Nick mentions a new course on dome train called 'From Zero to Hero: Vertical Slice Architecture' by Kevin Docks, which covers a popular architectural pattern.

Q & A

What is the main topic of Nick's video?

The main topic of Nick's video is the introduction of a new type of Globally Unique Identifier (GUID), version 7, in .NET 9 preview 7, and its comparison with ULID (Universally Unique Lexicographically Sortable Identifier).

What is the unique feature of GUID version 7?

GUID version 7 is unique because it is chronologically sorted, taking time into account, and all GUIDs start with the same few bytes indicating time in milliseconds.

What is ULID and why is it used?

ULID stands for Universally Unique Lexicographically Sortable Identifier. It is used in distributed computing to generate unique IDs that are lexicographically sortable and can be encoded more efficiently than GUIDs.

Why might ULID be considered suboptimal for some use cases?

ULID might be suboptimal because it is not the most character efficient way of encoding 128 bits of randomness, and it can cause fragmentation in data structures, leading to performance issues.

What are the criticisms of using ULID in distributed systems?

The main criticism of ULID is that it can cause fragmentation in data structures, which can lead to performance issues, especially when used as primary keys in databases.

What is the difference between ULID and GUID version 4?

ULID is designed to be more character efficient and sortable, while GUID version 4 provides randomness but lacks additional information and can also cause fragmentation in data structures.

How does ULID ensure uniqueness when generated?

ULID ensures uniqueness by generating a unique 128-bit identifier per millisecond, and the chance of collision is astronomically low, so it is assumed that each ID does not exist unless proven otherwise.

What is the performance comparison between ULID, GUID version 4, and the new GUID version 7?

According to the video, ULID performs faster than both GUID version 4 and the new GUID version 7. GUID version 7 is reportedly twice as slow as the original GUID generation method.

What is the advantage of using ULID over the new GUID version 7?

The advantage of using ULID over the new GUID version 7 is its better performance and the fact that it is already widely used and supported across different programming languages.

How can one implement ULID in a .NET application?

One can implement ULID in a .NET application by installing the ULID NuGet package and using it to generate ULIDs, which can then be converted to GUIDs if needed.

What is the 'From Zero to Hero: Vertical Slice Architecture' course mentioned in the video?

The 'From Zero to Hero: Vertical Slice Architecture' is a new course on .NET train that teaches everything about vertical slice architecture, a popular architectural pattern, with practical code examples and diagrams.