This is one of those posts that isn’t that long, but since I have some aversion to Twitter threads I didn’t want to turn it into some “1/n” thing.
I’m going to start by talking about network requirements, even though that’s not really what this post is about, because network requirements are interesting and topical. The “new normal” has created some considerations/constraints on resources that most organizations don’t usually have to contend with, like “how do we deliver this service reliably to people who are hundreds of miles away and to people in our office at the same time?” At my place of employment there has been some discussion about network requirements and bandwidth needs for schools implementing hybrid learning; a school running teleconference equipment with teachers live-streaming their classes has greater network requirements than a school that doesn’t.
The typical solution to this problem is simple: overprovisioning, or buying more bandwidth than you think you would ever possibly need. For an organization of means, this is a fine solution, although inelegant if you subscribe to the systems engineering principle of “free capacity is wasted.” However, bandwidth is relatively cheap, and future use cases are hard to predict.
The typical thought process for capacity planning is something like “if I have 10 users who need 100mbps each, I need a 100mbps connection.” That’s wrong for a number of reasons, but most obviously, it assumes each capacity-consuming entity is using 10mbps all the time, which they almost certainly are not, even in 2021. However, there are other factors in play, and networks are bottlenecked all the way down to the endpoint.
This is actually interesting stuff; if you want to get into it, you have to dig a bit. I wanted to freshen up my memory on it, so I decided to do some of that.
I was fairly disappointed in what I found, which is an increasingly common feeling I’ve been having using Google. Google is becoming less and less useful, partially because I’d consider myself mid-career at this point, and I know what I don’t know. But Google as a tool is increasingly becoming a victim of the things that make it a viable product in the sense of “we have this thing and it needs to generate revenue.”
Anyway, after reviewing multiple sources, I found out that a pioneer of doing “network math” is William Stallings, an MIT-educated author who has written many books about computer science and systems engineering, and he writes about calculating network load in his book Local and Metropolitan Area Networks, which appears to have been last updated in 2000.
Bandwidth was not cheap in 2000.
My source for this information ended up being a Cisco Press text – Top Down Network Design – I accessed via my O’Reilly subscription. The O’Reilly subscription is an excellent and comprehensive technical resource. A Library of Alexandria, but for neckbeards. It is also $499 per year if you are an individual – I’m fortunate enough to receive a subscription to the service through my employer. Some libraries may offer it as well.
Google would have you believe that network capacity planning is a problem that can only be solved if you are one of many vendors leveraging inbound marketing and optimized SEO to provide your viewer a surface-level understanding, with maybe a smattering of technical information, of the content they are trying to understand. If you do a Google search for “network capacity planning” (without quotes) – of the first 10 results, 9 are vendors. On the second page, I get this result:
It looks like this would be pretty good, but (like many SEO-optimized inbound posts) it’s not really a “guide to bandwidth capacity planning.” More like “a list of bad things that will happen if you don’t pay us to do this work for you.” (God help you if you need to do work on your house and you search for something.)
What this marketing vehicle provides can be pretty informative, and is often engaging, but is not really a path to understanding, more of a path to purchase. To be clear:
- I understand this is a valuable business practice, and some of this content is truly useful. Compared to the advertising of the early web, inbound marketing is much less sinister. It’s also easier. I used to ask vendors for whitepapers all the time. The information here is no different.
- Most people are pretty good at knowing when they’re being sold something, especially when given time and space to read and reflect on it.
- Almost all vendors set up the next step of the engagement for you, which is usually getting ahold of someone inside the sales organization of that vendor.
So what’s the problem? Getting paid is good. I guess.
The early internet was created and used for supporting defense programs and research institutions. A place where knowledge existed for the sake of knowing it. Like many things, capitalism has turned this model on its head. Rather than looking for information, many have leveraged the internet to placate an algorithm they don’t really have any meaningful control over, regardless of the quality of their content or its factual accuracy. The marketplace of ideas has become…a marketplace. The impressions seem to matter, the SEO, the algorithm, but little else. Whatever draws your eyeballs the fastest, not the best. This is dismaying, and has made Google less useful – which of course it would when you consider Google’s incentives as it acts as a mediator in this exchange.
Then, if you actually want to get to quality information, you have to pay for it, behind a paywall or an article limit or a subscription, like the one I used to eventually find the information I was originally looking for. I have some sympathy for publishers here as much as I have disdain, because quality content takes time, effort, and resources to create and distribute, and that effort should be rewarded. At the same time, these controls over information enable questionable content (in accuracy and intent) to thrive, for free. I don’t know what the solution to this is, because this problem seems intractable, but I imagine the solution will include free or mostly free access to the internet at some point.
Maybe this is all one big nostalgia trip for a bygone era, where we sat wide-eyed at our desks in college and embraced the quirky and unknowable “Web 2.0.” Before the machines of capitalization and monetization repaved the information superhighway into a toll road, adorned with billboards of the highest bidder. Garish, wide, a sort of subliminal horror. As I look in the rearview mirror, I wonder if this is the best we could’ve done.
Maybe I’ll try Bing.