Migrating to Oracle on Windows

#11

"OS religious wars are totally besides the point"
I agree.

My testing shows Windows operations are faster than linux in exactly
same hardware.
Also, I almost never use a GUI in windows unless it is the better way
to go for a particular task.
Get to know PowerShell. It's very nice.

#12

On 21 mai, 15:28, Mladen Gogala <mgogala.SPAM_ME.... (AT) verizon (DOT) net>
wrote:

Quote:

On Sun, 20 May 2007 12:10:51 -0700, hasta_l3 wrote:
Hmm.. Not sure what you mean here, Mladen. Memory-mapped files can be
used to share memory between processes.

Moreover, memory mapped files are not shared in real time.
For the change to be visible by other processes, it has first
to be flushed to the disk. This type of sharing is completely
inadequate.

I think that is incorrect, Mladen

" With one important exception, file views derived from a single file
mapping object
are coherent or identical at a specific time. If multiple processes
have handles of
the same file mapping object, they see a coherent view of the data
when they
map a view of the file." (http://msdn2.microsoft.com/en-us/library/
aa366537.aspx)

Recall that mapped files are backed up in virtual memory. You need to
flush if
you want to update the underlying file. But people using mapped files
for shared
memory usually dont even have such a file (you can map directly the
paging file)

This said, I would really *love* to be shown wrong...

Quote:

What is needed are standard IPC primitives, not supported by windows.

Which primitive do you feel is missing ?

Quote:

That is why Oracle has multi-threaded architecture on Windows,
as opposed to the independent processes everywhere else.

Well, I suspect that oracle chose the standard multi-threaded
architecture for efficiency reasons alone.

For example, some reports on the improvements of bulk collect
lead me to suspect that the PL/SQL -> SQL switch is faster
under windows than under unix.

--- Raoul

#13

On Mon, 21 May 2007 10:21:40 -0700, hasta_l3 wrote:

Quote:

" With one important exception, file views derived from a single file
mapping object
are coherent or identical at a specific time. If multiple processes have
handles of
the same file mapping object, they see a coherent view of the data when
they
map a view of the file." (http://msdn2.microsoft.com/en-us/library/
aa366537.aspx)

What is a "coherent view"? Does it reflect any changes to the mapped
area immediately? My information tells me that it is not the case.
The "consistent view" means that all users will see anything written
to the file immediately, which is useless.

Quote:

Recall that mapped files are backed up in virtual memory. You need to
flush if
you want to update the underlying file. But people using mapped files
for shared

Unless you flush, other users will not see your changes. This is not mmap
because Windows doesn't support shared memory.

Quote:

memory usually dont even have such a file (you can map directly the
paging file)

This said, I would really *love* to be shown wrong...

What is needed are standard IPC primitives, not supported by windows.

Which primitive do you feel is missing ?

Standard IPC primitivies. Look into manual pages under IPC:
shmat, shmget,shmop,shmctl, semget,semop and all the rest:

Quote:

ATTRIBUTE TYPE | ATTRIBUTE VALUE |
_____________________________|____________________ _________|
Availability | SUNWipc (32-bit) |
_____________________________|____________________ _________|
| SUNWipcx (64-bit) |
_____________________________|____________________ _________|

SEE ALSO
ipcrm(1), savecore(1M), msgget(2), msgrcv(2), msgsnd(2),
semget(2), semop(2), shmctl(2), shmget(2), shmop(2), attri-
butes(5), environ( 5)

The manual page is ipcs on Solaris:

$ uname -a
SunOS lpo-oracle-01 5.8 Generic_117350-39 sun4u sparc SUNW,Ultra-80
$

--
http://www.mladen-gogala.com

#14

On 21 mai, 20:08, Mladen Gogala <mgogala.SPAM... (AT) not-at-verizon (DOT) net>
wrote:

Quote:

On Mon, 21 May 2007 10:21:40 -0700, hasta_l3 wrote:
" With one important exception, file views derived from a single file
mapping object
are coherent or identical at a specific time. If multiple processes have
handles of
the same file mapping object, they see a coherent view of the data when
they
map a view of the file." (http://msdn2.microsoft.com/en-us/library/
aa366537.aspx)

What is a "coherent view"? Does it reflect any changes to the mapped
area immediately?

Yes. Assume process X writes value V at address A. If it reads back
the content of address A, it will get back value V. By the coherency
guarantee above, process B reading address B mapped to the same
virtual memory cell will also read value V.

You may want to browse
http://msdn2.microsoft.com/en-us/library/ms810613.aspx
notably the first sentences of the section on shared memory

Quote:

My information tells me that it is not the case.

Well, I would really like to browse a reference
(life teached me to be a die-hard skeptic :-)

Quote:

Which primitive do you feel is missing ?

Standard IPC primitivies. Look into manual pages under IPC:
shmat, shmget,shmop,shmctl, semget,semop and all the rest:

Unfortunatly, I dont know unix...

Windows synchronization objects include events, semaphores
and mutexes, which can be used across processes.

Communication include shared memory and queues

--- Raoul

#15

On Mon, 21 May 2007 12:55:30 -0700, hasta_l3 wrote:

Quote:

Unfortunatly, I dont know unix...

Windows synchronization objects include events, semaphores and mutexes,
which can be used across processes.

Not the way POSIX defines them. POSIX IPC primitives very well
defined standard which specifies arguments, behaviors and return
values. Microsoft chose not to implement that. Actually, Microsoft has
a POSIX layer and the corresponding section is here:

http://tinyurl.com/36yh34

Also, this is an additional layer, not native callse, and this software
has a very bad rap. Portability is simply not Microsoft's MO.

--
http://www.mladen-gogala.com

#16

goooooglegroups (AT) yahoo (DOT) com wrote:

Quote:

I would have thought that this would be a common thing to do and thus
would have a solution that was well known among the database community
- so I am just looking for some help\guidance.
Actually switching data servers is not common at all.

One thumb rule that some go by is that any decision on a DBMS for a
given app remains in place for seven years (one may argue the exact
number but long by software standards it is).
After those seven years the app is reasonably dependent on the DBMS and
any change is quite expensive. Hence you see some interesting vendors
popping up like Ants (Sybase -> Oracle)or EDB (Oracle -> Postgress).

The regulars in this group will be the first to tell you that an app
should exploit the DBMS intricacies to the fullest....

Cheers
Serge
--
Serge Rielau
DB2 Solutions Development
IBM Toronto Lab

#17

On May 21, 5:53 pm, Serge Rielau <srie... (AT) ca (DOT) ibm.com> wrote:

Quote:

goooooglegro... (AT) yahoo (DOT) com wrote:
I would have thought that this would be a common thing to do and thus
would have a solution that was well known among the database community
- so I am just looking for some help\guidance.

Actually switching data servers is not common at all.
One thumb rule that some go by is that any decision on a DBMS for a
given app remains in place for seven years (one may argue the exact
number but long by software standards it is).

Talk Like Yoda Day, it is! http://www.yodaspeak.co.uk/index.php

Quote:

After those seven years the app is reasonably dependent on the DBMS and
any change is quite expensive. Hence you see some interesting vendors
popping up like Ants (Sybase -> Oracle)or EDB (Oracle -> Postgress).

The regulars in this group will be the first to tell you that an app
should exploit the DBMS intricacies to the fullest....

I agree that it is not common, but it certainly is a big selling point
among many app vendors.

On the other hand, when it is done, it can be a lot of work. Some
people can get quite lucrative work specializing in such things.
Large organizations are likely to be db-heterogenous and have a
backlog of old apps that could benefit from a migration like the OP
asks about. I got an unsolicited PDP job email yesterday... yeesh.

jg
--
@home.com is bogus.
"Your update subscription (customer number xxxxx-xxxx) expired on
December 31, 1969." - F-Prot

#18

On Tue, 22 May 2007 11:11:09 -0700, joel garry wrote:

Quote:

I got an unsolicited PDP job email yesterday

PDP? Like PDP-11, RSX, KED and alike? Somebody is still using that?
It looks like the old computers never die. They live forever in our
nightmares. Don't get me wrong, I spent significant part of my career
being a VAX/VMS system administrator and, as a student, I was using
FORTRAN and PASCAL compilers on PDP-11. We had some fun with the
DEFINE FILE command and old RZ drives but I am now 46 years old,
quarter of a century after the last use of an RSX box that I can
remember. Anyone who has kept these boxes in a corporate IT after
all these years deserves to be shot, regardless of my nostalgia and
Peter Pan complex.

--
http://www.mladen-gogala.com

#19

On May 22, 8:22 pm, Mladen Gogala <mgogala.SPAM_ME.... (AT) verizon (DOT) net>
wrote:

Quote:

On Tue, 22 May 2007 11:11:09 -0700, joel garry wrote:
I got an unsolicited PDP job email yesterday

PDP? Like PDP-11, RSX, KED and alike? Somebody is still using that?
It looks like the old computers never die. They live forever in our
nightmares. Don't get me wrong, I spent significant part of my career
being a VAX/VMS system administrator and, as a student, I was using
FORTRAN and PASCAL compilers on PDP-11. We had some fun with the
DEFINE FILE command and old RZ drives but I am now 46 years old,
quarter of a century after the last use of an RSX box that I can
remember. Anyone who has kept these boxes in a corporate IT after
all these years deserves to be shot, regardless of my nostalgia and
Peter Pan complex.

--http://www.mladen-gogala.com

I dunno about corporate IT, but a quick google at the pdp groups shows
people are still using them (I assume for data collection or some sort
of control activities).

I have a couple of RSTS machines in my basement (one I bought in 1984,
I think, must have the paperwork around somewhere, saved the other
from a dumpster in the '90s), but I haven't even tried to turn them on
in a decade. I started my paid career on an 11/34. I first saw
Oracle on a VAX around 1983, but at the time, 11/780's still were
better for lots of I/O. Around that time, VAXen started to be cheaper
than the support+electricity for the big PDP's... I think DEC must
have taught Larry something about forcing upgrades...

jg
--
@home.com is bogus.
http://antwrp.gsfc.nasa.gov/htmltest/rjn_dig.html

#20

On May 22, 8:22 pm, Mladen Gogala <mgogala.SPAM_ME.... (AT) verizon (DOT) net>
wrote:

Quote:

On Tue, 22 May 2007 11:11:09 -0700, joel garry wrote:
I got an unsolicited PDP job email yesterday

PDP? Like PDP-11, RSX, KED and alike? Somebody is still using that?
It looks like the old computers never die. They live forever in our
nightmares. Don't get me wrong, I spent significant part of my career
being a VAX/VMS system administrator and, as a student, I was using
FORTRAN and PASCAL compilers on PDP-11. We had some fun with the
DEFINE FILE command and old RZ drives but I am now 46 years old,
quarter of a century after the last use of an RSX box that I can
remember. Anyone who has kept these boxes in a corporate IT after
all these years deserves to be shot, regardless of my nostalgia and
Peter Pan complex.

--http://www.mladen-gogala.com

Of course this made me totally curious, so I had to ask:
http://groups.google.com/group/vmsne...%3Ajgar&rnum=1

jg
--
@home.com is bogus.
Last words of Robert Charles Comer before execution: "Yes, go
Raiders."