alanwilliamson

Profiling Netbeans within Amazon Ec2

It was only a matter of time before we started seeing real how-to's pop up on how we can start profiling our Java applications running deep within the cloud. When my long term friend and colleague, Kirk Pepperdine, started talking about taking his world class Java performance tuning skills out of the server room and into the clouds I was excited.

He's just getting started, with this handy start-up for getting your live clouded Java applications directly profiled using Netbeans.

Elephant in the room

The big elephant in the clouds, is of course the fact that the hardware to which you are running your suite of applications on is shared, albeit in a highly optimized manner. You are never getting the true power of the CPU if you were on a dedicated machine sitting in your own server rack. But just how much of a penalty are we paying?

Many different, crude, tests exist that let you see just how much processing the likes of Amazon will be allocating you, and you'll be surprised to discover how dramatically it differs over the course of the week.

Keep an eye on what Kirk produces over the next coming months as he begins to take his benchmarking into the clouds.

Comments

Edward, you are absolutely right, performance is many things to many people. For me, the time it takes to number crunch isn't as important to say i/o levels, as i am more akin to using clouds in web applications.

The problem at the moment is that there is no official standardised tests to run against cloud instances to see if there is any type of guidance.

At the moment, we can't really say if a Flexiscale instance is faster than a GoGrid instance or an Amazon instance. So anyone that is willing to look into these and come up with a range of benchmarks then they have my vote.

left by Alan Williamson — Saturday, 18 October 2008 1:11 PM — web site

Alan,

Performance is a many sided question.

Are you talking about how fast you can start up a new server? The cost for each page served to the user? How fast you can compute Pie?

I look forward to any numbers that are relevant to common Cloud Applications. But I already know that the servers are weak simple "blades" or "pizza boxes". This is not new to me.

As for the variable performance of the servers it would be good to understand why this is true and in general how I need to plan for this variability.

Edward M. Goldberg http://blog.edwardmgoldberg.com

left by Edward M. Goldberg — Thursday, 16 October 2008 11:55 PM — web site

Leave Comment

please note, all comments will be moderated for spam and abuse before being publicly posted.


 

Recent Cloud posts

Recent JAVA posts

Latest CFML posts


 
Site Links
Recommended Sites/Blogs

Follow javachampion on Twitter