Last week, Benchmark announced a $3.5 million investment in Engine Yard, which provides commercial support for Ruby on Rails applications. Engine Yard is doing $3 million in business and growing. It's also profitable. It didn't need the investment.

The investment, however, is very telling. When one of the top venture capital firms on the planet puts hard dollars behind a support model, it's significant. It becomes doubly so when the firm (or its investors) in question previously invested in JBoss, MySQL, SpringSource (Interface21), and other support-based open-source companies.

It may mean that Benchmark knows something that the rest of the industry seems determined to ignore: services-based businesses may well be the future of the software industry.

...

One of Mitt Romney's sons used to be my neighbor. As such, it was hard to not contribute to the campaign.

I'm finding, however, that it's even harder to disengage from the campaign. I get four emails per day (sometimes more) from the campaign (usually two of the same message sent to the two email addresses of mine they somehow have on file). I get calls. I can't get away. Dana calls it basic database marketing. I call it annoying.

It's a bit like the traditional proprietary software model, where obnoxious sales people sit in your office haranguing would-be buyers until you purchase the proprietary ball-and-chain to get rid of the salesperson. (Which is exactly what happens when you write the check - the salesperson disappears. Completely.)

Today I found myself pining for an open-source political campaign. It would operate something like this:

…

A couple of days ago, Jacob DeHart and Harper Reed pinged me with an interesting idea for a MySQL storage engine. A "write-only" engine into a Bloom Filter.

What a neat idea!


The bloom filter, and it's varients and improvements, is almost magic. You can insert items. You can tell if you have inserted an item in the past, You can tell if you havn't inserted an item in the past (probably). But you can't get a list of all the items inserted.

And the performance is interesting. The space used is fixed. Insert and lookup time is constant.


CREATE TABLE t (v VARCHAR(255)) MAX_ROWS=100 ENGINE=BLOOM;

Have to specify about how many elements it will hold.

It …

I dashed off a hasty post about speeding up replication slaves, and gave no references or explanation. That's what happens when I write quickly! This post explains what the heck I was talking about.

You should be warned. Amazingly, in 10 years of PHP/MySQL development work I never hit this issue in the manner I did this week. There are several reasons that PHP could not be able to connect to MySQL. The MySQL daemon could be down. It could be an authentication problem. Or, perhas the entire server is offline. The last one there is the one I want to talk about.

In your php.ini you will find a value called mysql.connect_timeout. In the PHP ext/mysql it defaults to 60 seconds. Likewise, the php.ini has the same value. As far as I can tell, this timeout only comes in to play when the server is completely offline. If the server is up, but mysqld is not, the server refuses the connection immediately. I suppose if the server was under high load it could be used as well.

Well, IMO, 60 seconds is way to long to wait on a connection to the database for a web application. We had a server offline and expected the …

I’ve finally decided to work on a number of improvements in Instrumentation within the MySQL Server I’ve wanted for the first year, See What is the optimal thread specific buffer size?. It’s been a while since I’ve compiled from source, and from these issues, the first under Ubuntu 7.10 (a fresh install). Here are some of the problems, and solutions overcome, just for some others that may experience them.

I should have simply read my own notes from years ago in Compiling MySQL, specifically the pre-requisites list, but it sometimes helps to remember why things are so.

In summary, I needed the following:

apt-get install automake libtool g++ ncurses-dev

Thanks Miademora, I meant to say that, forgot

First error, “aclocal: not found”

$ …

So you want to know what join_buffer_size, sort_buffer_size, read_buffer_size and read_rnd_buffer_size values for your application should be? These MySQL thread specific buffers are variables I can never get right because there are insufficient metrics, instrumentation or even abstract details with the present MySQL Versions. These are important because Memory is a resource that you want to maximum towards your database data (the System Global Area), and not towards the Process Global Area in which there is no limitations.

I’ve wanted to know this answer for some time, I’ve asked many people including internal MySQL resources, I’d hoped that when joining MySQL more details would be available, but I’ve never been able to get an answer. I’ve always been meaning to work this out, it’s now 2008 and well the time has now come to do something about it.

The …

Giuseppe Maxia had a stop-over yesterday on his way to the MySQL Staff Meeting. We met for community discussions, and rounded off with Weißbier and dinner in Munich’s famous Hofbräuhaus.

What a fantastic feeling! The Bavarian spirit of joy and happiness has been decoupled from tobacco stench! I’ve always liked the first and hated the second.

So the German non-smoker protection legislation of 1 January 2008 is good news for anyone considering arranging meetings in Germany. You will no longer be “welcomed” at the airport by an offensive smell. Headaches in German …

Paul Tuckfield of YouTube has spoken about how he sped up his slaves by pre-fetching the slave's relay logs. I wrote an implementation of this, tried it on my workload, and it didn't speed them up. (I didn't expect it to; I don't have the right workload). I had a few email exchanges with Paul and some other experts on the topic and we agreed my workload isn't going to benefit from the pre-fetching.

In the meantime, I've got a pretty sophisticated implementation of Paul's idea just sitting around, unused. I haven't released it for the same reasons Paul didn't release his: I'm afraid it might do more harm than good.

However, if you'd like the code, send me an email at [baron at this domain] and I'll share the code with you. In return, I would like you to tell me about your hardware and your workload, and to do at least some rudimentary benchmarks to show whether it works or not on your workload. If I find that this is beneficial for …