[Bioclusters] Any issues porting applications to OS X?

[Tim Cutts]

On 5 Mar 2004, at 15:53, Michael Chute wrote:

> For my two cents I would have to disagree with the Xeon approach.  A 
> cluster of Xserves is probley going to give you more speed and storage 
> for the buck than a Xeon machine.  We have  a small cluster of Xserves 
> running osX server and we find that it is very fast, and the new g5 
> slusters are even faster.

In my hands, 2.8 GHz Pentium IV Xeon matches a 1.7 GHz POWER4+ (the 
G5's big brother) at almost every genomics code I've thrown at it, so 
I'll believe you when I see some numbers, and not from Apple's website  
:-)

>  Another alternative that has been done in the past is to actually run 
> Linux on Xserves.  I don't know the details of this but I do know that 
> this has been done.

I had thought of mentioning that, but there's almost no commercial 
support for Linux on PPC by independent software vendors.  It seems a 
little pointless to me - if you're going to run Linux, you might as 
well run it on the best supported platform, which is still x86.

> If you look at the bioteam software as well there are over 200 
> bioinformatics tools included with the package and they all have a gui 
> interface which is very nice for the average user.

Sounds like this product is ideally suited to small to medium sized lab 
environments with relatively modest compute requirements.

> As far as management features you can't beat osx server.

I'd like to hear more, because I don't believe it.  Can you power cycle 
a crashed node remotely?  What sort of remote console do you have?  Can 
you do everything you need to through a command line as well as a GUI?  
I know GUIs are friendly, but when your cluster gets large you get 
tired of clicking buttons *really* quickly.  Your requirements are 
probably different from mine, though.

>   Everything is so easy to do you don't need a bunch of IT people to 
> do it for you.  I am a microbiologist and I admin our cluster.

As I said, I guess this depends on scale.  I'm part of a team of four 
that run a cluster of over 1200 machines.   They're physically located 
in a building 500 metres from my office - I really need to be able to 
do almost anything to them remotely, short of having to physically 
remove them to change parts, and without having to use a mouse.

>   I think the "most tools for bioinformatics under linux" is not 
> exactly true.  I don't think you will find much trouble finding an osx 
> port for most of the tools.  FYI there is going to be a webcast about 
> the Xserve cluster for use in bioinformatics next thursday, you might 
> want to watch, you might get some of your questions answered.

Sounds worth a listen...

> http://education.apple.com/webcast/workgroupcluster/

Tim

-- 
Dr Tim Cutts
Informatics Systems Group
Wellcome Trust Sanger Institute
Hinxton, Cambridge, CB10 1SA, UK