ZFS solves 1990s era problems - nobody has filesystems that large on local disk. We use clustered filesystems for that. On SAN, GFS works nicely.
DTrace is good, but in practice you don't often run into problems where kernel-level tracing is required. The recent example of 'look at MySQL running SQL queries by watching printfs' could be done with strace (or for that matter MySQL in the first place).
Everything else Solaris sucks as usual, except for the app situation, which sucks even more now that x64 is the flavor of the month and there are so few apps for Solaris x64.
I agree that that recent example of printing the mysql queries is kind of dumb. I can definitely see a use for dtrace for gathering metrics and mining data that it isn't worth the application provider building in, or digging into software that you can't get the source for (traditionally much less of an issue on Linux than on Solaris). But listing the queries that are being sent through your mysql instance is a feature that not only should exist as part of the base mysql feature set, it already does.
This example is in the same class as using Fibonacci as a good example of recursion.
[+] [-] nailer|17 years ago|reply
ZFS solves 1990s era problems - nobody has filesystems that large on local disk. We use clustered filesystems for that. On SAN, GFS works nicely.
DTrace is good, but in practice you don't often run into problems where kernel-level tracing is required. The recent example of 'look at MySQL running SQL queries by watching printfs' could be done with strace (or for that matter MySQL in the first place).
Everything else Solaris sucks as usual, except for the app situation, which sucks even more now that x64 is the flavor of the month and there are so few apps for Solaris x64.
[+] [-] thwarted|17 years ago|reply
This example is in the same class as using Fibonacci as a good example of recursion.