Designing an Internet
Why the Internet was designed to be the way it is, and how it could be different, now and in the future. How do you design an internet? The architecture of the current Internet is the product of basic design decisions made early in its history. What would an internet look like if it were designed, today, from the ground up? In this book, MIT computer scientist David Clark explains how the Internet is actually put together, what requirements it was designed to meet, and why different design decisions would create different internets. He does not take today's Internet as a given but tries to learn from it, and from alternative proposals for what an internet might be, in order to draw some general conclusions about network architecture. Clark discusses the history of the Internet, and how a range of potentially conflicting requirements—including longevity, security, availability, economic viability, management, and meeting the needs of society—shaped its character. He addresses both the technical aspects of the Internet and its broader social and economic contexts. He describes basic design approaches and explains, in terms accessible to nonspecialists, how networks are designed to carry out their functions. (An appendix offers a more technical discussion of network functions for readers who want the details.) He considers a range of alternative proposals for how to design an internet, examines in detail the key requirements a successful design must meet, and then imagines how to design a future internet from scratch. It's not that we should expect anyone to do this; but, perhaps, by conceiving a better future, we can push toward it.
Series
Series
2 released booksInformation Policy
Information Policy is a 2-book series first released in 2018 with contributions by David D. Clark and Sasha Costanza-Chock.
Reviews
Popular Reviews
Reviews with the most likes.
This book takes a two-star hit because it's in dire need of editing. The author clearly is an expert in the field and communicates a lot of interesting ideas.
My favorite chapter is the one on security.
However, whenever Clark gives concrete examples of lessons learned or reasons for decisions the Internet's architects made, the book shines. This is how the book has been marketed, so that's definitely good that it meets its promises.
But for a book so dense with technical discussion, it is lacking in visual aids (which hurts understanding) and it needs better headers and signposting. A textbook on network architecture would have plenty of visual aids, and (in my opinion) this is a textbook advertised as a book for a passionate and knowledgeable lay person.
No matter how layered the headers get, they always are fully left-justified and in exactly the same font size and weight.
Sometimes you get a space before and after a header. Sometimes you get one just before. Sometimes the header is Not In Title Caps but just in Normal sentence capitalization. Sometimes you're left wondering if there's a typo or if the upcoming section is nested below the previous.
Other times, the headers are not useful or confusing. For example, in the section on Network Management and Control, there is a section entitled “What Are Management and Control?” Two short sections later, another is titled “Management and Control.” This is a sign that the information could have been consolidated or removed.
At one point in the book, I ended up making a note that the author seems to keep repeating himself. “Didn't I just read this?”
I was very excited to read this book because of the positive press it had gotten and the interesting subject matter that was promised. I was disappointed by editorial choices (or oversights) that made reading this a lot harder than it should have been.
If there were a second edition where the above issues were fixed, I wouldn't hesitate to give it a 4/5, if not more.
As it stands, do NOT get this book expecting it to be one you can pick up and read. This book requires silence, and is so dense that it should be read more like a companion text during a policy course. It's not meant to be read in full at once, but in sections, ideally with discussion with other people. It requires too much stamina.
Or maybe I'm just stupid. That's certainly a possibility.
