Hey, it’s a second volume and I haven’t gotten bored yet! I had intended to post one yesterday but I was painting.
I would actually disagree with all 5 of the reasons. Slow web apps suck, there are no two ways about it. I don’t care if I’m doing a hyper complex OLAP query if the site is not responsive.
But I totally agree with the title. I have been saying this for years that performance is irrelevant. It is the “perception of performance” that is important. Does your app take 10 seconds to load? It doesn’t really matter as long as the user is able to interact with it quickly. As long as the user’s perception of the apps speed is that it is fast, it does not matter how long it takes to render… from the user’s perspective. It probably matters to Google Page Rank, though.
Anyway, I posted this because there’s a chart that shows various websites and their render times with, what is most important, the First Paint time. The first paint time would be the first time that someone can see or interact with content from the app. THAT is the most important number.
I liked this article not so much because I like hearing stories about what people do when they “break the build” but things that you can do so that your mainline code doesn’t get foobarred in the first place.
Another day, another X vs Relational Databases post. I’m just waiting for a “Why NoSQL, Big Table, Columnar, Distributed, Sharded, Memory Resident databases are faster than SQLite” post. But I posted this article because graph databases are a good solution to looking for connected, not just related, data. Social is “the now” and so connected data often is as important as related data and relation databases just don’t do this well. The performance difference documented in this post makes me want to spend some time actually learning a graph DB.
John Carmack posted this on Twitter. The author is talking about something Bill Gates said about improving education. He said something along the lines of having to measure things in order to figure out what to do. While Pournelle didn’t disagree, but noted that while measurement is important knowing what to measure and what measurements you determined to be important can mean vastly different things. He gave the example of WWII Britain having the problem of their supply vessels being sunk by German subs. So they used a tactic whereby they would maximize the number of subs sunk. The problem was that it was done at the cost of British ships being sunk as well. The problem that Britain had was not that they needed to sink German subs, but that they needed to feed their supply lines. They changed tactics and while the number of sunk subs went down the tonnage of supplies getting through went up.
Focus is hugely important for any organization. Focus on what is important by way of the goals of the organization first.
That’s enough for today.
Ran into this post after publishing the post. But given that it pertains to article #1 I added it hear. Very interesting info on diagnosing poor render times for browsers.