От: Pierre C
Тема: Re: SSD + RAID
Дата: ,
Msg-id: op.u8xl7mjbeorkce@localhost
(см: обсуждение, исходный текст)
Ответ на: Re: SSD + RAID  (Bruce Momjian)
Список: pgsql-performance

Скрыть дерево обсуждения

SSD + RAID  (Laszlo Nagy, )
 Re: SSD + RAID  (Karl Denninger, )
  Re: SSD + RAID  (Laszlo Nagy, )
   Re: SSD + RAID  (Axel Rau, )
    Re: SSD + RAID  (Richard Neill, )
     Re: SSD + RAID  (Greg Smith, )
      Re: SSD + RAID  (Merlin Moncure, )
  Re: SSD + RAID  (Marcos Ortiz Valmaseda, )
 Re: SSD + RAID  (Scott Marlowe, )
  Re: SSD + RAID  (Merlin Moncure, )
   Re: SSD + RAID  (Scott Carey, )
    Re: SSD + RAID  (Merlin Moncure, )
 Re: SSD + RAID  (Heikki Linnakangas, )
  Re: SSD + RAID  (Merlin Moncure, )
   Re: SSD + RAID  (Heikki Linnakangas, )
    Re: SSD + RAID  (Merlin Moncure, )
     Re: SSD + RAID  (Heikki Linnakangas, )
      Re: SSD + RAID  (Merlin Moncure, )
  Re: SSD + RAID  (Laszlo Nagy, )
   Re: SSD + RAID  (Robert Haas, )
    Re: SSD + RAID  (Laszlo Nagy, )
 Re: SSD + RAID  (Greg Smith, )
  Re: SSD + RAID  (Karl Denninger, )
   Re: SSD + RAID  (Greg Smith, )
    Re: SSD + RAID  (Karl Denninger, )
     Re: SSD + RAID  (Scott Carey, )
      Re: SSD + RAID  (Tom Lane, )
       Re: SSD + RAID  (Jeff Janes, )
      Re: SSD + RAID  (Greg Smith, )
       Re: SSD + RAID  (Karl Denninger, )
    Re: SSD + RAID  (Brad Nicholson, )
     Re: SSD + RAID  (Dave Crooke, )
     Re: SSD + RAID  (Greg Smith, )
      Re: SSD + RAID  (Merlin Moncure, )
      Re: SSD + RAID  (Merlin Moncure, )
       Re: SSD + RAID  (Brad Nicholson, )
        Re: SSD + RAID  (Scott Marlowe, )
       Re: SSD + RAID  (Peter Eisentraut, )
       Re: SSD + RAID  (Greg Smith, )
        Re: SSD + RAID  (Merlin Moncure, )
         Re: SSD + RAID  (Greg Smith, )
          Re: SSD + RAID  (, )
           Re: SSD + RAID  (Scott Carey, )
            Re: SSD + RAID  (Merlin Moncure, )
             Re: SSD + RAID  (Scott Marlowe, )
              Re: SSD + RAID  (Greg Smith, )
               Re: SSD + RAID  (Merlin Moncure, )
                Re: SSD + RAID  (Scott Marlowe, )
        Re: SSD + RAID  (Mark Mielke, )
        Re: SSD + RAID  (Scott Carey, )
         Re: SSD + RAID  (Greg Smith, )
        Re: SSD + RAID  (Bruce Momjian, )
         Re: SSD + RAID  (Ron Mayer, )
          Re: SSD + RAID  (Bruce Momjian, )
           Re: SSD + RAID  (Greg Smith, )
            Re: SSD + RAID  (Bruce Momjian, )
             Re: SSD + RAID  (Ron Mayer, )
              Re: SSD + RAID  (Bruce Momjian, )
           Re: SSD + RAID  (Ron Mayer, )
  Re: SSD + RAID  (Merlin Moncure, )
  Re: SSD + RAID  (Matthew Wakeling, )
   Re: SSD + RAID  (Bruce Momjian, )
    Re: SSD + RAID  (Dan Langille, )
     Re: SSD + RAID  (Bruce Momjian, )
      Re: SSD + RAID  (Scott Carey, )
       Re: SSD + RAID  (Bruce Momjian, )
        Re: SSD + RAID  (Ron Mayer, )
         Re: SSD + RAID  (Greg Smith, )
          Re: SSD + RAID  (Arjen van der Meijden, )
           Re: SSD + RAID  (Greg Smith, )
            Re: SSD + RAID  (Mark Mielke, )
             Re: SSD + RAID  (Greg Smith, )
              Re: SSD + RAID  (Scott Marlowe, )
               Re: SSD + RAID  (Scott Marlowe, )
                Re: SSD + RAID  ("Pierre C", )
                 Re: SSD + RAID  (Nikolas Everett, )
                 Re: SSD + RAID  (Scott Carey, )
          Re: SSD + RAID  (Bruce Momjian, )
           Re: SSD + RAID  (Ron Mayer, )
            Re: SSD + RAID  (Greg Smith, )
            Re: SSD + RAID  (, )
             Re: SSD + RAID  (, )
              Re: SSD + RAID  (Aidan Van Dyk, )
               Re: SSD + RAID  (, )
                Re: SSD + RAID  (Mark Mielke, )
                Re: SSD + RAID  (Dave Crooke, )
          Re: SSD + RAID  (Bruce Momjian, )
           Re: SSD + RAID  (Greg Smith, )
            Re: SSD + RAID  (Bruce Momjian, )
             Re: SSD + RAID  (Ron Mayer, )
              Re: SSD + RAID  (Greg Smith, )
               Re: SSD + RAID  (Bruce Momjian, )
              Re: SSD + RAID  (Bruce Momjian, )
               Re: SSD + RAID  (Greg Smith, )
                Re: SSD + RAID  (Ron Mayer, )
               Re: SSD + RAID  ("Pierre C", )
         Re: SSD + RAID  (Bruce Momjian, )
 Re: SSD + RAID  ("Fernando Hevia", )
  Re: SSD + RAID  (Greg Smith, )
 Re: SSD + RAID  ("Kenny Gorman", )
  Re: SSD + RAID  (Kenny Gorman, )
 Re: SSD + RAID  (Lists, )
  Re: SSD + RAID  (Ivan Voras, )
   Re: SSD + RAID  (Laszlo Nagy, )
    Re: SSD + RAID  (Craig Ringer, )
     Re: SSD + RAID  (Laszlo Nagy, )
      Re: SSD + RAID  (Craig Ringer, )
       Re: SSD + RAID  (Laszlo Nagy, )
     Re: SSD + RAID  (Craig James, )
      Re: SSD + RAID  (Heikki Linnakangas, )
     Re: SSD + RAID  (Scott Carey, )
      Re: SSD + RAID  (Craig Ringer, )
       Re: SSD + RAID  (Anton Rommerskirchen, )
        Re: SSD + RAID  (Brad Nicholson, )
      Re: SSD + RAID  (Greg Smith, )
       Re: SSD + RAID  (Matthew Wakeling, )
       Re: SSD + RAID  (Scott Carey, )

> I always assumed SCSI disks had a write-through cache and therefore
> didn't need a drive cache flush comment.

Maximum performance can only be reached with a writeback cache so the
drive can reorder and cluster writes, according to the realtime position
of the heads and platter rotation.

The problem is not the write cache itself, it is that, for your data to be
safe, the "flush cache" or "barrier" command must get all the way through
the application / filesystem to the hardware, going through a nondescript
number of software/firmware/hardware layers, all of which may :

- not specify if they honor or ignore flush/barrier commands, and which
ones
- not specify if they will reordre writes ignoring barriers/flushes or not
- have been written by people who are not aware of such issues
- have been written by companies who are perfectly aware of such issues
but chose to ignore them to look good in benchmarks
- have some incompatibilities that result in broken behaviour
- have bugs

As far as I'm concerned, a configuration that doesn't properly respect the
commands needed for data integrity is broken.

The sad truth is that given a software/hardware IO stack, there's no way
to be sure, and testing isn't easy, if at all possible to do. Some cache
flushes might be ignored under some circumstances.

For this to change, you don't need a hardware change, but a mentality
change.

Flash filesystem developers use flash simulators which measure wear
leveling, etc.

We'd need a virtual box with a simulated virtual harddrive which is able
to check this.

What a mess.



В списке pgsql-performance по дате сообщения:

От: "Pierre C"
Дата:
Сообщение: Re: SSD + RAID
От: Robert Haas
Дата:
Сообщение: Re: No hash join across partitioned tables?