c0t0d0s0.org

Göttlich. Ein Bericht im Deutschlandfunk über den Befall weiter Kreise des Managertums mit der Erkrankung BSE und den Folgen.

Andrew Leventhal and Brendan Gregg gave an interesting presentation how DTrace can help you to find nasty performance problems in your software (and how dtrace helped Twitter to find performance problems).

Kris hat mal wieder einen seiner gewohnt guten Texte auf die Menschheit losgelassen: Wie gestaltet man einen anständigen mysql-Server?. Ich will da eigentlich nichts gross zu sagen, sondern den Text ein wenig mit der Sun Brille betrachten.

The Register writes in Investors stunned by Sun's apparent health:

And the company deserves credit for making the most of these moves and beating analyst expectations on the costs front. The company now seems able to post profits with consistency even during so-so quarters, which is what you want in a turnaround story.

Jonathan comments the results of FY07 in his blog. But more interesting, there was a photo in the press kit for the results:

Jonathan holds one of the reasons in his hand, why we will see a very successful FY08. It´s a wafer full of Niagara II dices. And as Jonathan states at the end of his block entry: You ain't seen nothin' yet.

Ashlee Vance of Register is one of the few really insightful reporter in IT journalism. For example, he is the first in mainstream IT media who got the point behind Project Indiana.

Many pundits have said that Sun hopes to make Solaris more "Linux-like" with Project Indiana, although we struggle to see how that's accurate. Sun is really just tweaking the Solaris release cycle in a way it should have done once the company committed to revitalizing Solaris x86 and to upping developer interest in the OS.

We gave the financial community 8.5 percent operating margin in the fourth quarter of the last! This is great ... no ... it´s awesome! From the press release:

Net income for the fourth quarter of fiscal 2007 on a GAAP basis was $329 million, or $0.09 per share on a diluted basis. For the full fiscal year, net income was $473 million, or $0.13 per share, on a diluted basis, as compared with a net loss of $864 million, or ($0.25) per share, for fiscal 2006.

A Microsoft researcher writes in a paper about Sun Fire X4500 with Microsoft SQL:

This is the fastest Intel/AMD system we have ever benchmarked.

... Schuhe aufzubewahren:

The support for logical domains made it into the 2.6.23 linux kernel. Wow, this was really fast ...

Many people think, that custom chips for appliances are faster for their respective task than general purpose chips. For such areas like graphical processing this is true. GPUs are so powerful, many people think about their usage in computing. But i made an interesting observation in the SPECjbb2005 list: The company Azul Systems developed a special purpose processor for Java. The top of the line 7280 system consists out of 16 processors and provides 872972 SPECjbb2005 bops. Thus the performance per processor is 54561 SPECjbb2006 bops. A single T2000 has an perfomance of 96523 SPECjbb2005 bops. Thus the general purpose processor UltraSPARC T1 (okay, almost general purpose) has twice the performance of the custom build Vega2 processor on a per core basis.

Another comparision is interesting to: Assume a workload that can be scaled horizontally . You need an aggregated performance of at least 850.000 SPECjbb2005 bops. You could use 1 Azul 7280 or 9 T6300 blades. You would assume that the Azul solution is more efficient. But the the 7280 takes 14 rack units and is rated with a typical wattage of 3250 Watts (as specified on their website). 9 T6300 would have a size of 10 rack units and round about 2700 watts. With Intel C2D the performance calculation would be similar, although not in the same power envelope.

The Azul system has an different advantage, it provides a single image to application, thus the java application would be able to use the full 756 Gigabytes of memory. But at the end, it´s an interesting evidence that custom build hardware isn´t necessarily faster or more efficient than general purpose hardware.

I´ve wrote some articles about the separated ZIL and the usage of flash disks in the last few days. I proposed the usage of FlashSSD to store the separated log. Put perhaps it´s much easier to use hybrid harddisk than i thought: Hybrid harddisk have commands to pin certain blocks into a flash based NVcache. Thus, it should be feasible to create a partition in the size of the NVcache and use it as the seperate ZIL device. To leverage the NVcache, it should be sufficient to pin down the ZIL partition to the NVcache.

One of the must-read blogs about the developments in the x64-arena is Scientias blog. He published a really insightful article about the AMD Analysts Day. Really worth a read.