1. 24
  1.  

  2. 20

    I don’t think “naysayers” are saying that the author is wrong that many websites are faster over HTTPS. I think the “naysayers” are saying that the author is being misleading. If the author wasn’t looking for the controversial, click-baity title, this would have been something along the lines of “HTTP/2 is faster than HTTP/1.1” or “Browsers haven’t bothered to implement HTTP/2 over cleartext.”

    1. 3

      Honestly, I think your proposed titles would be more misleading. They’re technically correct, but contrary to a certain cartoon, that’s the worst kind of correct.

      There are a lot of people out there who still, in 2016, believe serving their sites over HTTPS would make them slower than serving them over HTTP to all people in all circumstances.

      In the physical universe we currently occupy, to any modern browser, HTTPS is simply going to serve faster. It doesn’t matter why. It doesn’t matter that if we froze the universe at HTTP/1.1, HTTP would win. The universe isn’t frozen at HTTP/1.1. HTTPS is faster in the real world. Demanding that we continue to qualify that is misguided and misleading. The message needs to get out and the people it needs to out to aren’t going to clue in that statements like “browsers haven’t bothered to implement HTTP/2 over cleartext” means “HTTPS is faster than HTTP in 2016” in technically correct pedant-speak.

      1. 1

        As you mentioned, this article targets folks who own servers, not clients, because most users don’t make a conscious decision between https and http. For folks who own servers, just turning on TLS won’t be enough to provide a speed increase. In order to get the speed increase, their servers must also switch to support HTTP/2. This is because it’s HTTP/2 that’s providing the speed increase, not TLS.

        I think there are two ways of looking at this. The first is, “When you switch your servers to use TLS, you can now start using HTTP/2, which will make your server faster.” The second is, “When you switch your servers to use HTTP/2, you must use TLS, or your clients won’t be able to use it.” I think that shortening the first to “TLS makes your website fast” with the implicit, “actually it doesn’t but now you can use HTTP/2” is more misleading than “HTTP/2 makes your website fast” with the implicit “and browsers only support HTTP/2 over TLS, so you will also have to use TLS”.

        If you want to convince someone to use TLS, and want to use HTTP/2 to do it, you can feel free. It’s still convincing to say, “HTTP/2 is faster than HTTP/1.1” and “browsers only support HTTP/2 over TLS”.

    2. 13

      HTTP/2 is faster than HTTP/1.1, but that is not the point. The point is that you should stop sending lots of crap, i.e. CSS, fonts, JS, silly huge filler images to the user’s browser that are not really needed to get to the content.

      1. [Comment removed by author]

        1. 5

          Which makes Troy’s article equally useless for me. I too want my site to go fast. Turning on https isn’t actually going to do much to accomplish that.

          1. 9

            Well, in a way, HTTP/2 does. Individual requests are less expensive because of multiplexing, so you’ll increase performance about as much as you would if you implemented the various HTTP/1 performance hacks (like image spriting).

          2. 3

            For those clickbait sites it is just a more efficient way to invade on your privacy and extract money from you by selling your data ;) HTTP/2 solves that problem, not a problem a site that actually cares for its users has.

            That said, without verifying if they already do, I wish Google Maps would use HTTP/2. It is dog slow on my computer just scrolling through the map… but maybe that is just the excessive JS.

            1. 3

              but maybe that is just the excessive JS.

              Porque no los dos

              1. 4

                My guess is that something like google maps probably needs that JS to work, but as I said, that’s just a guess.

            2. 2

              And so apparently all the tech folks who should know better are going to bend the technology to help people who, by definition, don’t understand the technology. Good work, everyone. >:(

              1. 5

                No, this isn’t true. With HTTP/1, we had to do a lot of hacks to speed up sites, since requests needed to be made one at a time. HTTP/2 will make it easier to increase your site’s performance.

                What @fkooman said is also true though, it’s not going to magically make your javascript crunch down to 1 byte or anything. I do take exception to his point on CSS, I have yet to see a site where the CSS was causing severe loading issues. JS almost always outweighs the CSS in terms of filesize. Also the real page load bottlenecks are usually images, not JS or CSS. Most sites you see that load slowly have megabytes of images, and kilobytes of JS and CSS.

                When I have time tonight I’ll update this comment with the real averages that have been obtained from various web analysis projects.

                1. 5

                  I’ll cheerfully point out that, on most of the sites I’ve seen, the sheer quantity of shit getting loaded–and the shit getting loaded the slowest–is in the form of ads, tracking gizmos, testing and surveillance codes, and whatnot, scattered over dozens of domains.

                  The requests don’t have to be made one at a time; if memory serves, browser will make concurrent connections to a point, and connections can use a keep-alive header to prevent expensive setup and teardown.

                  Sure, images are large–but even imageboards like 4chan do a better job than shitty buzzfeed clones on mobile that pull in eighty bajillion other resources.

                  1. 2

                    I guess what I’m saying is there is no such thing as a magical device that makes your site load super fast. HTTP/2 improves one web performance bottleneck, but developers still need to do their part.

                    I’ll cheerfully point out that, on most of the sites I’ve seen, the sheer quantity of shit getting loaded–and the shit getting loaded the slowest–is in the form of ads, tracking gizmos, testing and surveillance codes, and whatnot, scattered over dozens of domains.

                    I agree. Then again, do the owners of these sites actually have the end user in mind in terms of performance? I doubt it. HTTP/2 will not help in that case.

                    However, in the case of many other sites who do care about performance (or at least aren’t actively trying to add every last gadget they can think of to their pages), HTTP/2 will be a huge boon.

                    Let’s face it, you’re not going to fix sites like Wired, Buzzfeed, etc. with a technical improvement. Changing those will either require market force, a huge change in web practices, or other external mechanism. This doesn’t make the technical improvements bad.

          3. 9

            This is like saying “walking is faster with rubber under your feet”. And then revealing that by “rubber under your feet”, you were referring to the tires on your car.

            And then wondering why people find this misleading.