[an error occurred while processing this directive]
Back to newsletter 078 contents
Capacity management is about optimizing the overall balance of the resources available currently and in the future for your system - including considering the hardware, the software, and the monetary and power budgets. Performance tuning fits into capacity management as a tool to more effectively utilize the resources available, and to minimize the resources that are needed. And those of you who need to consider the wider capacity management picture are saying pretty much the same thing - capacity requirements are increasing linearly overall (even including all the effort to keep down those requirements); and the datacenters keep getting full all too quickly.
That linear increase can only be reduced by making the software work more effectively on the available resources - and that's the responsibility of the performance tuning community. But ultimately information technology is such a win that there is still going to be a need to expand those full datacenters for quite a while yet. And on that front, it would be ideal if there was a way to incrementally expand the datacenter instead of guessing what we need in the future, and building as big a datacenter as we can afford. And with a simple idea, Sun have just done that. It's called Project Blackbox (at the moment, the name will change soon). I recently had the opportunity of a preview, and I concluded that here is one of those deceptively simple ideas that is going to make a big difference. To put part of a datacenter in a standardized shipping container is such a simple and powerful idea - and note that it doesn't need to be stocked with Sun kit, you can put any hardware in there that fits on the racks, just like any datacenter. In one swell foop, Sun have a solution to incremental datacenter expansion; to modular datacenter improvements; to enhanced disaster recovery support; to datacenter location changes; to datacenter support in remote locations.
The biggest effect is going to be on Sun of course. I had previously said that Sun can't exist as a hardware seller, and that they need work out how to make money from Java. I was wrong. Here they have re-invented themselves again, and I expect them to make good money out of this idea. But looking further into the future, is this the beginning of the end of Sun? The mix of businesses is already a real melange and if, as I suspect, the shipping container datacenter takes off as a significant reveue earner, then breakup would look to be inevitable as the parts would be worth more separately than together: the various sections of Sun would be worth significantly more with a variety of companies (e.g. Dell or HP with the containers, Oracle or IBM with Java, Red Hat to integrate Solaris into Linux, and the servers going to the inevitable Chinese company looking to move into higher end server hardware). Java itself would be unaffected by a breakup of Sun, which is a measure of how far it has come. Interesting, don't you think?
Now on with the rest of our newsletter. Of course we have our usual news, article links, tools, and extracted tips. Over at fasterj.com, we look at Su Doku solving efficiency; Javva the Hutt continues to tell us all about how he got his new job; and, of course we have extracted all the tips from those articles. And yet another treat for this month, we have a tool report from eoLogic on their eoSense analysis tool.
Java performance tuning related news.
Java performance tuning related tools.
Back to newsletter 078 contents