Richard Rodger

Running a Java 1.5 SAX XML parser on the UltraSparc T1 is five times slower than running it on my Athlon 2800+.

That's about right. Five times slower. Now when I got delivery of my new T2000, I thought, “Hey, this baby's gonna rock!”. Then weird stuff happened.

First, I ran the performance tests for my Java CSV Component. Twice as fast. Okaaay. Well, fair enough, it is faster. Would have expected more though.

Then, I ran the performance tests for my Java XML Component. Five times slower. Eek!

Alright, I thought, slow disks! Nope. Doing the tests in memory led to the same performance. And the disks were slow, in case you're wondering.

OK. Maybe my code is to blame. Let's do a pure and simple SAX parse, with no actions. Still five times slower! So now I was getting depressed. My lovely new T2000. Broken! Or worse: Crap!

So I went out for lunch.

And then it hit me. Duh! 32 hardware threads over 4 CPUs (and that's just the minimum config). And I was only exercising one of them! My performance test only runs in one thread. You see, the T2000 is designed for throughput. Serving lots of web requests all at the same time. That sort of thing.
[Update: I wrote the original post from memory — the are only 16 hardware threads. But I'll leave the rest of this post the way it is, so you can have a few laughs. The final conclusion still holds as you can get higher spec T2000s with more CPUs].

So I run a really brain-dead test. I opened up a whole load of consoles and started the test on all of them, all at the same time.

Pay Dirt!

Not a wince. Not a whine. Cool as a breeze. Same speed on all consoles! The T2000 just laughed at me. A few samples from mpstat, and I was a happy camper again. You see, each thread may be a bit on the slow side, but you do have 32 of 'em. So if each one is five times slower, you need five threads to do the same work. But: five into 32 total threads gives you six or so. Which means:

Six Times Faster! Baby!

So what you get with the T2000 is a big old scalability lever, and a fairly small performance lever. Same old story really. There's always a trade-off.

Now maybe I was doing some wrong. I don't know. You tell me. I've been out of the game for a while when it comes to Solaris.

Here are the performance tests I was running, in case you're interested: XML Manager, CSV Manager.

So I don't know what this means for my report on the T2000. The single thread numbers suck, but the concurrency rocks. Guess I'll need some new charts!

I've just put up a new release of CSV Manager, my CSV file parsing Java component. This release adds the ability to handle comments in CSV files.

I should explain how the Ricebridge versioning system works. It's pretty standard in fact. The first number is the major version. All releases with the same major version are forwards and backwards compatible with each other. And when you buy into a major version, you get all releases in that version. I don't believe in making customers constantly pay for upgrades. A major version bump occurs when there is a really big change in functionality.

The second number is the minor version. This indicates a release series with a stable feature set. No major structural changes occur within a minor version. The minor version is bumped whenever there is a reasonable increase in functionality. Compatibility with earlier minor versions is retained.

The third number is the build number within the minor release. This is incremented each time there is a bug-fix release. It is also sometimes used for small features, such as, oh…, comment support, say. All bugs, by the way, are documented. You know exactly what you're getting when you buy a Ricebridge component.

The only thing I haven't got quite sorted just yet is a proper change log for the components. But hey, it's on the to-do list!

So now you know how I work. Just to repeat, if you buy the 1.1 version of any of our products, you'll get all the versions up to 1.9 (or 1.whatever we get to), as free upgrades.

Technorati Tags: java csv file component release Del.icio.us Tags: java csv file component release

Literally. This thing is so loud I now have a headache. No more benchmarking for me today.

I've decided to introduce free trials for my two main products. As a micro-ISV, these types of decisions are always a bit tricky. I've done pretty well up to now sticking to online demos. By their nature, the products (a csv parser and an xml parser), are not that suited to downloadable demos. They are API-based components, after all, so there's nothing to see, as such.

But I've decided the time is right to experiment. You don't get anywhere in business if you stick to the same old strategy regardless. Based on visitor comments and usage of the online demos, I think that the lack of a downloadable trial for the products is hurting my sales. By how much is hard to say. But we'll soon find out!

One reason that people do not offer a downloadable trial is fear of copying. Well, to be frank, this is just not a problem for me. I don't know why people are so afraid of copying. To my mind, those who copy could probably not afford to buy anyway, so you've lost nothing, and gained a user. I already offer my products free to charities and public bodies, so this is an extension of that line of thinking. Sure, some will never buy, but down the line, quite a few will. Maybe that's terribly innocent, but hey, I believe in people!

The other question about demos is how to structure them — time-limited or function-limited. I don't really like function-limited. It makes a bad impression. And it mucks about with basic functionality. So I've gone time-limited. You'll get your standard 30 days. That should be enough time to see if the components work on your CSV or XML file.

So check back here in a month or two and we'll see how this little experiment goes.

Well, you've got hand it to Jonathan Schwarz (Sun CTO), when he says “free server”, he means “free server”.

A while back I decided to apply for this “free server” promotion that Sun Microsystems is running. You get a free server if you can run some benchmarks and if you blog about it. It seemed way too good to be true, but hey, I'll try anything once. So off I went to their online application form, filled out the details, and thought; “there's no way they'll go for a micro-ISV!”

But they did — I got a call from Sun a while back and was told it was on. A Sun reseller, Horizon Open Systems, are looking after the details. I still can't quite believe it! Way to go Sun. I know Sun bashing is a bit of a sport, and they have done some silly things, but I doubt you'll find a opportunity like this with any other vendor.

Here are the benchmarks I'm going to be running: CSV Manager Performance Report, and XML Manager performance report. Let's just say I expect significant speed increases.

So what is the server anyway? It's a Sun Fire T2000. Here's a picture of the box. It sure is purty. It has a 4-core ULTRASparc T1 Processor, 4Gb of memory and two 73Gb SAS disks. And that's just the starter kit. Makes you want to cry if you've been living in x86-land.

There is one tiny fly in the ointment though. Sun sort of forgot to pack an RJ-45 to DB9 adapter, which you need to access the machine controller, so that you can actually boot Solaris. Oops. So at the moment it just a big lump of metal and silicon sitting in my office. The Frustration! Anyway, Horizon are getting me sorted with the cable, so we should be up and running next week. It's gonna be a long weekend.

Congratulations to Roman Nowak, winner of last week's Ricebridge Java Components Competition! A free XML Manager for you sir!

So this week, let's get pay homage to our descendants: grab those HTML anchors! All you have to do is figure out some simple XPath expressions to get at the data in the document, and a free copy of our XML Manager Java component is yours.

Alright people, don't just stand there! … Go!

…

Uh, begin, commence, start moving… theoretically you have been racing for about forty seconds now, and so far Mr. Schaffer is winning because he's nearest to the door.

CSV Manager, our Java CSV file component, has just got a JARS TOP 5% rating!

Previously we had a TOP 25% rating, but now we've been moved up! Here is the mail I got this morning:

Resource Name  : CSV Manager
Developer ID   : RR22
Resource  ID   : RR22-1

-------------------------------------------
Presentation : 400
Function     : 300
Originality  : 250
-------------------------------------------
Total        : 950 (out of a possible 1000.)

JARS - The Java Applet Rating Service 
- http://jars.developer.com/

Not bad. The presentation score is really great. We put a ton of work into our documentation (there's loads of it for a start). We really are pretty obsessive when it comes to good documentation. It's the one thing that lets great components down. Document or die. That's the way it is.

Right. Next job: a TOP 5% Rating for XML Manager. Here we go…

Congratulations to A. Luepke, winner of last weeks Ricebridge Java Components Competition!. A free XML Manager license is on it's way.

And on to this weeks competition. Just get your head around a simple address book XML document. All you have to do is figure out some simple XPath expressions to get at the data in the document.

And thanks to all of last weeks entrants — let's see what you can do with this week's challenge.

Good Luck!