Comment Re:De Icaza Responds (Score 1) 498
Sorry, but I do this for a living, and nobody who has a clue is using Java for low-latency application development. All the benchmarks and whitepapers are fine and dandy, but in the REAL WORLD, Java simply can't handle (very) heavy loads without falling over.
I don't know why that is, but suspect that it's simply a matter of too many frickin' layers.
The other issue is that Java performance is simply not deterministic because of the GC -- everything is fine for a while, and then you have a 100ms spike when the VM decides to do GC. And yes, I know that there's a lot of work going on to address that, and some of it is promising, but it simply ain't there yet.
I have no idea how much of this is also applicable to .Net, although I suspect at least some of it is. I also suspect that the bigger problem is that the whole project ended up being a gigantic clusterfsck, between Microsoft and Accenture (and who the heck thought Accenture knew anything about designing trading systems anyway?)
Last point: you think that was bad -- the volumes in UK are WAY lower than here in the US, so one can only imagine what would happen if they tried to roll out something like this in a really high-volume environment.
I don't know why that is, but suspect that it's simply a matter of too many frickin' layers.
The other issue is that Java performance is simply not deterministic because of the GC -- everything is fine for a while, and then you have a 100ms spike when the VM decides to do GC. And yes, I know that there's a lot of work going on to address that, and some of it is promising, but it simply ain't there yet.
I have no idea how much of this is also applicable to
Last point: you think that was bad -- the volumes in UK are WAY lower than here in the US, so one can only imagine what would happen if they tried to roll out something like this in a really high-volume environment.