It's really not that hard to imagine. There is an embedded library version of MySQL called libmysqld which is made for this very purpose. Of course as you point out it's quite overkill for something that would work perfectly well with Berkeley DBs.
_SQL Server Express_ is just a free (as in beer) limited version of _SQL Server_. It's still a full-blown traditional server like MySQL, Oracle and other that require services and administration tools to be installed.
Mirosoft has 2 equivalents to SQLite: The Jet database engine is pre-installed in all versions of Windows and allows creation of .mdb databases (those used by older versions of MSAccess). You don't need to include any dll file to use it.
The other is _SQL Server Compact_ which, like SQLite, is an embedded engine that you can bundle with your application by including a library.
All of these embedded databases are able to do multi-user writes to some extent. Jet is actually quite good if careful with locking (emphasis on careful). SQLIte is a de-facto standard because it's simple, performant, cross-platform and flexible.
While it can replace full-blown databases in some cases, it's far from being always true. There are still many cases where using something like SQL Server Express may make more sense, for instance if you want to offer a path to your customer for drop-in replacement of the database based on the growth of their needs over time. Not saying it's not a costly lock-in, but it's an easy one to sell.
I also don't find the admin side much of a faff. Provide your users a backup/restore option within your app and ensure you install your own instance.
Probably because it runs in the same address space as each server process and there is no message passing (TCP/unix socket) overhead.
That is a good opportunity to ask questions and learn about the person, their knowledge and their use case.
Yeah sometimes the answer is they don't know how stuff works underneath and are just following a blog post or a trend or heard someone talk about it at meetup.
Programmers love programming tools / frameworks / languages. There are so many new shiny things out there. New ones pop up every month. New language, new build system, new db, new UI library, new message queue etc. Interstingly over the years, lots of those have had great performance improvements. Maybe they took advantage of new kernel API or some new algorithm and so on. That feeds the hype -- put new shiny things in and performance doubles.
So people have been trained to expect all the new things to just magically be better. But like said, you have to probe and find out the reason they propose the change, only then you can only see if it is justified or not (one of the resonable responses could be "I've actually measured with a production load and it does look better"),
Say you have an MVC web app, could you have asynchronous calls to a thread performing the actual DB writes? From your app's perspective, concurrent writes would be placed into a queue that performs them serially.
Are there any functional issues to this approach? Thanks in advance for your thoughts or info.
Otherwise you'll end up serving stale data that might be several minutes out of date.
Anyway, I'm not saying "never use MySQL for mail users" (though, I think the percentage of deployments where it makes sense is closer to none than it is to one), I'm just trying to make the point that MySQL is, in some folks minds, a magical solution to performance problems. Often, it not only introduces needless complexity, it won't even improve performance. It's a classic example of "when all you have is a hammer, everything starts to look like a nail". MySQL is a very fine hammer. It just isn't the right tool for every job.
The article we're talking about is another case where a little knowledge is a dangerous thing. A desktop app serving one user with a tiny data set (as I understand it, we're talking about the metadata for a person's music collection) is exactly the right workload for SQLite. I'd be shocked if a naive port to MySQL were faster (though they acknowledge that there's room for query optimization), and not at all surprised if it were slower. And, I know it'll require more memory and disk space for the same working set.
Shouldn't user groups give a bit more accountability there?
Amarok music player on KDE linux desktop environment (at least on version 3.5) kept the music collection metadata in database. You could choose sqlite, mysql or postgress during installation if I remember correctly. It worked ok, because of the package system.
I guess if you used separately installed databases there would be some conflicts, though.
I've seen it so many times over the years regarding users and email configurations. I can't count the number of times I've dropped into someone's badly behaving mail configuration and found they had MySQL hosting the users, and explained it was for "performance" reasons. Somehow they didn't grasp that /etc/passwd fits entirely in memory, and the map files Postfix uses for various lookups and stuff are already a database (just one specifically designed for the task at hand) and also fit entirely in memory. Putting that MySQL layer in there is disastrous if performance matters; it is orders of magnitude slower for any case I've seen...still plenty fast for most cases, but it's ridiculous that this idea gets cargo-culted around that if you store your mail users in MySQL your mail server will be faster.
A little knowledge is a dangerous thing, is what I'm trying to say, and people who know MySQL is "fast" may not know enough to know that it's not the right tool for the job in a lot of cases...and is probably slower for many use cases. I'm pretty confident this is one of those cases. SQLite is wicked fast on small data sets, and being smaller means more of it will fit in memory; I can't think of any way MySQL could be a more performant choice for this workload.
Also, I don't even want to try to imagine shipping an installable desktop application for non-technical users that relies on MySQL!