Right now your looking at it from a completely x86 view. Look at it from teh point of view from a hardware based system that's been doing it for years. A lot of these problems have been solved already.

Let's say you have a lot of servers at big corp. Each runs a specialized application, and each application is required to be isolated from the rest. A good VM system like zVM can help you a ton. You get a hardware platform that has tons of mature disaster recovery solutions, and a hypervisor that can dynamically allocate resorces between different VMs to the point where you don't even see it.

I mention zVM a lot because I know a lot of folks that are involved with large scale rollouts of it, in production, with great results.

The downside is that you need people who know what they're doing, and the hardware is expensive as hell.

isn't vmware esx just a specialized linux distribution that is built just to run vm's?

Everything has some overhead, sure. However, when you have hardware level virtualization (where the logic is in the firmware like the IBM mainframe systems), and not in some software hypervisor, the overhead is very minimal. On mainframes they where running LPARs with native performance over 20 years ago.

if submitting 20 small queries instead of 1 join is faster, that app must really suck. Doing network fetches isn't free, most simple joins are fine if your tables are indexed correctly (assuming your DB doesn't suck). I do joins on tables with > 100 million rows in oracle and they are very fast.

Not any more. the last of that code is gone.

as a user of oracle on AIX, I must agree.

If you want to see the best virtualization, don't look at power. Look at the zVM systems. I know a couple of very large sites that run it and it's very impressive.

Sparcs get spanked in most workloads. Stuff like price/performance matters, performace/watt doesn't matter at all when you have real work to do, unless your work involves sitting idle.

I use the IBM JVMs under AIX in a high volume production environment. It works pretty good except the headless AWT is broken. Otherwise, it does pretty well.

Webturd runs on the IBM JVM. IBM has it's own JVM that runs on mainframes, AIX, linux, and windows.

Oracle fucks up everything they touch.

Sons of bitches, what a crappy day.

I'd agree that a lot of J2EE applications and containers are complete shit (i'm a java developer and I like it, dammit). If you want to write good and scalable java code, a lot of times it's better to go with a lightweight system. an example is the Pojo message beans you can use with spring, or using embedded Jetty instead of a big slow ass websphere to serve up web services. Leaks in java apps crack me up because it's usually somebody who doesn't understand the feature they're using.

I use an open source MQ at work and it's never the bottleneck in high message rates (> 100 msgs/second easily)

well, sometimes you want more then one message consumer per process.

block all js from here and it loads a lot faster. A lot of web sites send the no cache headers for all JS so you have to wait for their slow ass init routines to finish.