[rescue] Oracle kills Solaris

[Andrew M Hoerter]

On 9/6/17 19:29, Phil Stracchino wrote:

> On 09/06/17 19:20, Jonathan Patschke wrote:
>> They were thinking, "Barring a hardware failure, We do not want the user
>> to be able to accidentally lose data; in the service of this, nearly any
>> inconvenience is permissible."
> 
> Which is not an area in which MyISAM can claim any laurels...
> 
> 
> To be fair to MyISAM, one of its principal design constraints was that
> it needed to work *acceptably well* on a small server shared with other
> applications, at a time when a _large_ server was one that might have an
> entire 32MB of RAM.

To me at least, reliability and "ACIDity" are preconditions for being a 
traditional RDBMS, not a design consideration that can be traded off in 
favor of something else.  Admittedly MySQL has improved over time, but 
certain qualities in software (like security) can be very difficult to 
retrofit later if you didn't adopt the right philosophy from the very 
beginning.  See also: Linux containers.

Of course, people with different priorities can make whatever tradeoffs 
they want -- but hopefully they're educated about the risks and know 
what they're doing.  I suspect that's rarely true of people who pick a 
database engine based on microbenchmarks.

Also, I'd mention SQLite as an obvious counterexample against needing to 
sacrifice reliability for performance or small footprint.  Ok, that's a 
somewhat unfair comparison when taking into account the relative feature 
sets, but I don't think it's *too* far off the mark.