| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.server -> Re: ORACLE ON NT SERVER vs. UNIX
I actually wonder if internal European Commission reports should be =
posted on
Usenet (I work for them too). Still, there are a few remarks to be made =
here. One
is (of course) that the comparison of Sun and PC equipment is open to =
criticism, as
the Sun used is about 3 years old, has half the RAM (SPARC is RISC and =
memory
hungry) and the PCs are new and have double or 4x the memory. Solaris =
2.4, as used
in this comparison is 3 releases old (2.4, 2.5, 2.5.1, now 2.6), and =
the latest
version is considerably faster and certainly more reliable and =
manageable. I am not
totally happy with this analysis as you might notice. As it is now in =
the public
domain I have comments on a few of the system-related points...
The guy who instigated this thread should not automatically run to NT =
as a
solution. It is something you have to look at quite hard. Remember, as =
a platform,
it has problems - OK< so have UNIXen, but they are generally lesser =
ones. And you
can always run UNIX on a PC platform if you mant a more open hardware =
solution. Sun
(as an example) posted their highest SPECweb96 results on a 4-way NCR =
XENON box.
iolo_at_my-dejanews.com wrote:
> In article <35BFA6A0.45694C4F_at_netexplorer.com>,
> Paul Nguyen <pnguyen_at_netexplorer.com> wrote:
> > Hello everyone,
> >
> > can anyone tell me if Oracle database run faster on Unix than it =
does on
> > NT server ?
> >
> > Please help me out.
> >
> >
> Hi,
>
> Oracle on Unix or NT - A few points to take into consideration
>
> 1) The Unix architecture with several independent processes =
communicating
> through shared memory has been converted into one process with =
several
> threads. No shared memory is required since all the threads share the =
same
> virtual address space. The advantage of this implementation is that =
the
> creation of thread requires less memory and it is done faster than =
the
> creation of a process in Windows NT. The architecture does not =
support the
> MultiThreaded Server option yet. Therefore, each connection to the =
database
> originates the creation of a new thread in the server (shadow =
thread). The
> number of threads associated with the Oracle7 Server process varies =
depending
> on options selected (background threads) and user connections. The =
maximum
> number of shadow threads, and thus the maximum number of user =
connections, is
> 500.
Threading is not necessarily faster in all situations of course. For =
UNIX
architectures, shared memory can be faster simply becuase of OS =
implementation -
especially with SMP boxes. It's a dark area, as in the end bottom you =
have to enter
kernel space for everything. There are certainly undocumented thread =
limitations in
NT. As for memory, remember that process spaces are shared quite =
efficiently in
UNIXes. The overhead in memory can be lowerer than you think, but quite =
hard to
analyse precisely.
> 3) Oracle and Windows NT have both been certified at C2 level. =
However, the
> implementation of Oracle on Windows presents some aspects that could =
be
> improved to achieve a higher security. Oracle recommends the creation =
of an
> special account for database administration and to prevent any =
physical
> access to the database files by means of Access Control Lists (ACLs) =
that
> allow the access only to the administrator account. However, the =
Oracle tools
> that create database files (Instance Manager and Enterprise Manager) =
don't
> set up their ACLs. Thus, the newly created files inherit the values =
from the
> ACLs specified at the parent level (which may be inadequate). =
Therefore, the
> DBA is left the responsibility of properly setting the access rights.
ONLY NT 3.51 HAS BEEN CERTIFIED AS C2. And that is in a standalone =
configuration.
NT4 was never certified. So I wonder how the combination of NT4 and =
Oracle can
actually add up to a C2 secure solution. Plus there are horrible =
networked
Denial-of-Service attacks for NT - much worse than for UNIXen as some =
apply to the
IP implementation. It is difficult to upgrade multiple servers with all =
the
security patches on a frequent basis, especially as many NT OS services =
require a
reboot.
> 5) FAT/NTFS ? Response time of Oracle using different disk =
configurations and
> the tests carried out show that there is very little or no difference =
at all
> from the performance viewpoint. However, there is from the security
> viewpoint.
Yup. NTFS has security problems. Nasty ones...
> Therefore, NTFS is recommended for all disk partitions when
> setting up Oracle on Windows NT because of security reasons. The use =
of Raw
> Files is not recommended because of the complex administration. In =
special
> situations, tests should be carried out to prove whether Raw files =
improve
> performance.
In UNIX systems yes, raw often works best.
> 6a)Endurance tests: Will Oracle7 on Windows NT bear the
> workload? The objective of these tests were to find out whether =
Oracle7 on NT
> will support the load that heavy processes impose on it. A heavy =
process uses
> complex non-optimised queries that involve a big amount of data =
processed for
> retrieval, insert, update and delete. The tests consisted in =
observing the
> evolution of the response time as the number of clients working with =
the
> database increases. The tests were carried out in different =
configurations to
> judge the impact of different configurations in the performance =
("Annex IV.
> Endurance Tests Description" contains a detailed explanation of the =
tests
> carried out, the client and the results obtained). During the tests, =
the
> number of users has been raised from 1 to 99 and, while the response =
time has
> increased noticeably, the main conclusion to retain is that Oracle =
continued
> to behave correctly (without giving errors) while the workload =
increased
> provided that it had enough resources (redolog space, etc). The =
degradation
> in response time is not meaningful since the heavy clients resemble =
more
> batch processes than interactive clients. Additionally, increasing =
the number
> of clients increased the amount of data to deal with because of the =
way the
> test scripts were implemented. Therefore, the performance degradation
> observed in these tests does not represent the perfor-mance =
degradation that
> may occur in an application by raising the number of users. =
Additional
> observations that can be made from the collected data are the =
following: ·
> The disk configuration (1 stripped partition over 4 different =
physical disks,
> 2 stripped partitions over 2 different physical disks or 4 different
> partitions each one on a different physical disk) didn't have big =
impact on
> the performance of the test . · As expected, the amount of =
available RAM has
> a great impact in the performance (The configuration with double =
amount of
> RAM offered a response time of approximately half the response time =
of the
> other configurations). A closing remark concerning the =
simplification of
> administrative tasks gained on Windows NT. The test instance required =
a huge
> SGA to be created. In order to do so, modifications have to be made =
to the
> kernel in the Unix environment: first, to allow for more and bigger =
shared
> memory segments and, then, to increase the number of processes and
> semaphores. Once this was set up, the tests could be started only to =
make the
> machine crash because there wasn't enough swap space. On Windows NT, =
only the
> parameters of the init.ora needed to be set up in order to have the =
instance
> running.
On a Sun at least...
1. vi /etc/system (something you do at install) 2. mkswap (however big you want it, and it can be a file) 3. swapon
It's intersting that no attempt was made to evaluate Veritas or ODS =
during this.
Also, it is possible to administer UNIX from a single point. It is hard =
to remote
administer NT systems.
> ...This is specially applicable to environments were
> the migration of Oracle databases to a Windows NT environment may =
reduce the
> number of different systems to support (eg, by getting rid of Unix =
servers).
And replacing them with (according to this) lots of different types of =
boxes (count
the motherboards, disk manufacturers etc). People, remember that PCs =
are not all
built from the same internals...
> The tests were carried out in the following
> configurations to judge the impact of different configuration =
parameters in
> the performance:
>
> ECC1-D1
> Computer Prioris ZX 6200
> Processor 4 x Pentium Pro 200MHz
> RAM 512 MB
> OS Windows NT4
> Disk space 1 logical disk (9GB) corresponding to one physical =
disk
> (mirrored with RAID 5)
>
> MADRID-D1
> Computer Olivetti Strada 7000
> Processor 4 x Pentium Pro 200MHz
> RAM 256 MB
> OS Windows NT4
> Disk space 1 logical disks corresponding to one striped =
partition over four
> physical disks
> 1( 3,5 GB + 3,5 GB + 3,5 GB + 3,5 GB)
>
> MADRID-D2
> Computer Olivetti Strada 7000
> Processor 4 x Pentium Pro 200MHz
> RAM 256 MB
> OS Windows NT4
> Disk space 2 logical disks each one corresponding to one striped =
partition
> over two physical disks
> 1( 4 GB + 4 GB )
> 1( 4 GB + 4 GB )
>
> MADRID-D4
> Computer Olivetti Strada 7000
> Processor 4 x Pentium Pro 200MHz
> RAM 256 MB
> OS Windows NT4
> Disk space 4 logical disks each one corresponding to one =
physical disks
> (4 GB + 4 GB + 9 GB + 9 GB)
>
> Unix
> Computer SunSparc Station
> Processor 1 Sparc
> RAM 128 MB
> OS Solaris 2.4
> Disk space 4 physical disks (3 used for the database)
> 1,2 GB + 3,2 GB
> ~ 550 MB of swap space
Wow, that's an old SPARC and an OLD Solaris....
>
>
>
> --
> Oliver Willandsen
> European Commission
> http://europa.eu.int
> All remarks are my own and do not necessarily
> reflect official European Commission policy
Kind regards,
Neil Corlett
nb my remarks dont reflect Comission Policy either.
--
 |
 |