Oracle Server Performance
RDBMS Version: 9.2.0
Operating System and Version: Red Hat Linux relase 3
Database performance issue
we are using oracle 9i in LINUX platform, since last 4 days database performance is drastically decreased it is when number of active sesseions increases then performance is slow even queries and report are taking time. There was no major changes made in the database except some rebuilding of indexes.
Based on our technical support we have increased our number of processes to 350 and sessions_cache_cursor to 150 but there is no improvement. We have 60+ users and sessions parameter is 390. Ram is 1 GB. As this is our production server we are having a very hard time
Can you please help me in this regard
Thanks in advance
Have looked at any of the application SQL? Find some SQL that the users are executing through the application or reports that have poor response time and use SQL Trace and TKPROF to see were the SQL is spending its time.
Try the following:
alter session set timed_statistics=true;
alter session set max_dump_file_size=unlimited;
alter session set tracefile_identifier='SLOW_SQL';
alter session set events '10046 trace name context forever, level 12';
<insert sql with poor response time>
disconnect
Use the TKPROF utility on the file found in USER_DUMP_DEST that contains the string SLOW_SQL.
For information on how to interrupt the TKPROF output, see the following link.
http://download-east.oracle.com/docs/cd/B10501_01/server.920/a96533/sqltrace.htm
Similar Messages
-
ORACLE SERVER AND UNIX TP MONITOR-2
제품 : ORACLE SERVER
작성날짜 : 1995-01-24
Subject: Oracle Server and UNIX Transaction Processing Monitors-2
Page(3/4)
This file contains commonly asked questions about Oracle7 Server and UNIX
Transaction Processing Monitors (TPMs). The topics covered in this article are
o Oracle Parallel Server and TP Monitors
o Oracle and DCE-based TP Monitors
o Other commonly asked questions
The questions answered in part 3 provide additional detail to the information
provided in part 1.
Oracle Parallel Server and TP Monitors
======================================
How does Oracle Parallel Server (OPS) work with TP Monitors?
If you are using Oracle-managed transactions, there are no special
considerations. But if you are using TPM-managed transactions, and
thus need to use the XA interface, then Oracle requires release 7.1.3
or later and a special version of the Distributed Lock Manager, called
the session-based lock manager. This version of the DLM is not yet
available for all platforms. To understand this restriction, let's take
a look at one of the technical details of XA.
The XA specification requires that the Resource Manager be able to
move a transaction from one process to another, and even to be
able to commit in a separate process. In Oracle, transactions are
attached to sessions, so that means that we also have to be able to
move sessions. Therefore, the session/transaction can't have any state
which is tied to a particular process. The first generation distributed
lock managers were all built to use the process id as the lock owner,
which doesn't work for locks which need to move with the transaction.
Oracle and DCE-based TP Monitors
================================
How does Oracle interface to the Encina TP monitor? To CICS/6000? I've
heard that they require OSF DCE facilities in order to run?
Oracle interfaces to Encina and CICS/6000 just as it does to any other
TP Monitor. The TP Monitor issues XA commands to control transactions, and
Oracle executes the commands. Encina and CICS/6000 do use DCE features for
their own operation. However, this use is transparent to the Oracle Server.
What DCE facilities can Oracle products take advantage of when working with
a DCE-based TP Monitor?
The two most commonly mentioned DCE features which might be useful
to Oracle users are multi-threading and security. We look at these in
the subsequent questions in this section.
Encina documentation suggests that a Resource Manager such as Oracle can
be either single-threaded or multi-threaded? Which way is Oracle XA
implemented?
The Oracle XA implementation is single-threaded, as is any Oracle client.
Within a single process, at most one thread can access Oracle at a time.
Does that mean that only a single Encina application can access an instance
of Oracle transactionally at any given moment?
No. Oracle XA is only single-threaded within a single application server
process. Multiple applications can access Oracle simultaneously using XA
by using different application processes. Encina allows
(1) serial reuse of a single server by different clients. There are
two options for this. The server can use long term reservation
but be defined to be in shared or concurrent access mode, which
allows the server to be used by another client as soon as an RPC
completes. Alternatively, the server can use default reservation
and exclusive mode, which allows the server to be used by another
client as soon as the current transaction ends.
(2) concurrent execution by multiple servers, even if they are accessing
the same Oracle database. These may be executing the same or different
procedures.
These two features should let you get as much concurrency as you need.
Why isn't the Oracle XA library multi-threaded?
The XA specification specifically states that its use of the phrase
"thread of control" means a process. If an RM were to multi-thread its
XA, it would be in violation of the specification. This restriction
was put place in because at the time the specification was written,
there were numerous thread packages: if the TM used one, the application
another, and perhaps the RM yet a third, there's no way it could work.
As threads standards settle down, the later versions of XA will probably
relax this restriction.
Will Oracle change if the XA specification changes?
Very likely. The exact time frame will of course depend on the priority of
all work items at that time.
Does Oracle use DCE security via the TP Monitors?
The integrity of the connection between a DCE TP Monitor client and DCE
TP Monitor server is protected by the DCE security functionality.
Theoretically, the TP Monitor could make the DCE-protected client security
information available to Oracle. Unfortunately, there's no standard way
for a TP Monitor to pass security information information to a Resource
Manager such as Oracle. Oracle is leading an effort to extend the X/Open
model to allow use of the security information provided by the Monitor.
In the meantime, the basic DCE security features such as encryption are
useful within TP Monitors.
Effective use of DCE security would normally also mean that the security of
the TP Monitor client be passed through the TP Monitor, through the Oracle
client (application server), to the Oracle Server, and possibly on
to other Oracle Servers through database links. The ability to transfer
security information to other processes, called delegation, is missing
in DCE version 1.0. DCE version 1.1, expected to emerge in late 1994,
has some delegation features. Oracle is examining these features to see
how they might be used.
Are there any special considerations for CICS/6000?
There are two:
(1) It is inefficient to run without XA. CICS/6000 is designed to
use XA. It uses XA so that the CICS server can log on to Oracle
when it starts, after which it makes that Oracle connection available
to any transaction it executes. If you don't use XA, the CICS server
does not itself log on to Oracle so each transaction has to log on
and log off - a very expensive mode of operation. Also, it is very
un-cics-like in that the application does the log{on,off} and also
commits - in a mainframe CICS database program CICS would implicitly
do these operations. Oracle does not recommend this mode because of the
performance penalty.
(2) CICS servers are generic and dynamically load application modules.
In order for these modules to access the Oracle connection made by
CICS, the applications must be built with a shared object version of
the Oracle libraries. This is an installation option on platforms which
support CICS/6000 and other products using its architecture such as
CICS 9000.
Other commonly asked questions
==============================
What other Resource Managers can be included in an Oracle XA transaction?
Several other relational database vendors have an XA implementation
available or in progress. There is an XA C-ISAM product from
Gresham Telecomputing. There are also Resource Managers contained
within some of the TP Monitors which can be coordinated in the same
transaction. For example, CICS/6000 has VSAM files and other data
stores, Encina has its RQS queuing system, and Tuxedo has its /Q queuing
system.
What is Recoverable Queuing Service (RQS) and how does it interoperate with
Oracle7 and Encina? What about /Q?
Recoverable Queuing Service is a feature provided by Encina which allows
transactional, distributed queuing (enqueue/dequeue). Tuxedo has a similar
product called /Q. Because these products are themselves coordinated by the
TM component of the TP Monitor, their queue operations are atomically
coordinated with with operations on XA Resource Managers such as Oracle7
Server. That is, they can atomically put something on one of their queues
and commit an Oracle transaction, then at some later time dequeue an
entry atomically with doing some other Oracle transaction. The queue
system guarantees that the message will not be lost or transmitted twice.
Can I mix TP Monitor applications with standard Oracle7 Server applications?
Yes, you can have existing Oracle applications connected to the database
with alongside TPM applications against the same database. The TPM does
not manage the whole database, just those transactions which are started
by the TPM. The Oracle Server will properly handle concurrency control
between the transactions managed by itself and those managed by the TPM.
Is Oracle planning to change its tools to be more suitable for TP Monitors?
With Oracle Procedure Builder 1.5, to be available with CDE2,
Oracle will provide a foreign function interface that allows you to
dynamically set up PL/SQL calls that access C functions. In other
words, you can access C routines in Windows DLLs from within your
PL/SQL procedures. This will allow PL/SQL under Windows easy access to
TP Monitor APIs.
Does Oracle7 Server itself use XA-compliant TPMs as the interface to
foreign RMs?
No, for this purpose Oracle Server uses the SQL*Connect products or the new
Transparent and Procedural Gateway products.
Does Oracle7 Server use XA to coordinate Oracle7-only distributed
transactions?
No, it uses an internal mechanism.
Can database links be used with XA?
If an Oracle7 database is running under XA, it can access other Oracle7
databases through database links, with some restrictions. First, the
access to the other database must use SQL*Net V2 and be running MTS.
Second, it must currently be to another Oracle7 database. Assuming those
restrictions, the Oracle 7 database can do distributed update to another
Oracle 7 database by using a database link, whether it is started by an
Oracle application or a TP Monitor application. The TPM will see Oracle
as only a single RM, but Oracle7 will propagate all the transaction
commands to the other database, including the two-phase commit. If
the transaction is started by a TP Monitor application and is using XA,
it can also update non-Oracle resources managed by the TPM. If it
is started from an Oracle application, it can only include resources
managed by Oracle.
Here's a sample configuration:
| TPM | | TPM |
| client | | client |
| |
| |
| TPM |
| |
| |
| Oracle | Forms, Forms, | Oracle | | non-XA | | XA |
| client | Plus, Plus, | client | | TPM | | TPM |
--------- Pro, Pro, --------- | server | | server |
| Financials, Financials, | |(note 1)| ----------
| etc. etc. | ---------- |
| | | |
| SQL | SQL | SQL | XA
| commit | commit | commit | commit
| | | |
| Oracle | | Oracle | | Oracle | | Oracle |
| server | | server | | server | | server |
| | | |
| | | |
| | | |
| Database 1 | | Database 2 |
| | | |
| A | A
| | dblink to database 1 | |
| ------------------------------------ |
| |
dblink to database 2
Note 1: Oracle will work having both XA and non-XA servers but some TPMs
may have restrictions on this.
Are multiple direct connections possible from a Pro* program?
Using XA, you can not only specify multiple direct connections to Oracle7
databases, you can also update them both in the SAME transaction. The
way to do this is to use a precompiler feature called a named database.
When you use a named database, you qualify the SQL statement with the
database name. For example, you write EXEC SQL AT dbname UPDATE emp ....
We have a complementary feature in the xa open string to let the user
associate the name with a particular RM instance, called the DB clause.
You will also want to use the SqlNet clause in the open string so you
can give the two different SIDs. This clause does not require the use of
the SQL*Net product, it is just a naming convention. For more information,
see Oracle7 Server for UNIX Administrator's Reference Guide.
Some TP Monitors may not support having multiple Resource Mangers in the
same server; check with the TPM vendor.
Is there any collateral available for XA or TP Monitors?
Oracle At Work 52684.0692
Oracle7 Server for UNIX Administrator's #A10324-1
Reference Guide
Guide to Oracle's Products and Services #A10560
Oracle7 Server and CICS/6000 #A14200
Where can I get more information on the DTP model?
X/Open's address is
X/Open company Ltd (Publications)
P O Box 109
Penn
High Wycombe
Bucks HP10 8NP
Tel: +44 (0)494 813844
Fax: +44 (0)494 814989
Request
G307 Distributed Transaction Processing: Reference Model Version 2
X/Open Guide G307 ISBN 1-859120-19-9 28cm.44p.pbk.220g.11/93
Page(4/4)
This file contains commonly asked questions about Oracle Server and UNIX
Transaction Processing Monitors (TPMs). The topics covered in this article are
o Performance with Oracle Server and TP monitors
o Performance using Oracle's XA Library
The questions answered in part 4 provide additional detail to the information
provided in part 1.
Performance with Oracle Server and TP Monitors
==============================================
I have heard that Transaction Processing Monitors (TPMs) will increase
Oracle Server performance. Is this true?
Several hardware and TPM vendors have made the claim that TPMs
will increase RDBMS performance. This claim is based on TPC-A
benchmarks. The key point to understand about TPC-A is that it
requires, for every transaction-per-second, ten times that many
users to be connected. For example, to get 600 TPS, you need 6000
users. The next question will answer in more detail how the the
three-tier architecture addresses this requirement, but first let's
look more generally at what TP Monitors can and can't do to improve
performance.
TP Monitors can provide better performance:
(1) When there are more than several hundred users connected.
This is because of the TP Monitor's role in the three-tier
architecture, described in the next question. In this
architecture, terminal handling is offloaded to one or more
separate machines, freeing up those cycles to do database work.
Note that this does NOT mean that Oracle itself runs faster,
just that we've given it more CPU cycles to use.
(2) When, because of the high potential concurrency of requests,
significant resource contention exists. Use of a TP Monitor can
limit the degree of concurrency and thus reduce contention.
TP Monitors can not provide better performance:
(1) For existing applications. The applications must be designed
to fit the TP Monitor architecture.
(2) For applications which are highly interactive in their use of
the database. These applications put many messages
through the transport system, and the TP Monitor is not as
efficient as SQL*Net for point-to-point communication.
(3) For CPU intensive single-query decision support. When executing
a single large command, Oracle query facilities work efficiently,
especially with the use of Oracle Parallel Query, available in 7.1.
How does the three-tier solution help TPC-A, or other situations with
thousands of on-line users?
The TPC-A test calls for a large number of users to produce a given
result. In the high-end results we produced in June, 1992, for example,
6150 terminals were simulated to produce 618 TPC-A transactions.
Thus, terminal concentration accounts for a large portion of the total
processing time used.
First, let's look at how the Multi Threaded Server would work for
this benchmark. In this case, there are many client processes,
but only a few server processes, which handle client requests on a
first-come first serve basis. When they are done with a request,
they take another client's request.
ORACLE7 CLIENT/SERVER ARCHITECTURE WITH MULTI THREADED SERVER
| Client | | Server |
| __________ |______________|_____ _____________ _____________ |
| | Client | | SQL*Net | |_|Dispatcher | | | |
| | Process| | | ____| Process |___| | |
| |________| | | | __|___________| | | |
|____________| | | | | | | | |
| | | | | | Oracle7 | |
______________ | | | __|__|____ | Server | |
| Client | | | | __|_|_____ | | | |
| __________ | | | | | Shared | |____| | |
| | Client | | SQL*Net | | | | Server |_|____| | |
| | Process|_|______________|__| | | Process|_| | | |
| |________| | | | |________| |___________| |
|____________| | | |
| | |
______________ | | |
| Client | | | |
| __________ | | | |
| | Client | | SQL*Net | | |
| | Process|_|______________|____| |
| |________| | | |
|____________| | |
|_______________________________________|
Client processes = N Dispatcher processes >= 1
Shared server processes >= 1
If there are 500 clients in this environment, there will be one or more
dispatcher processes, dynamically tunable, and one or more shared
server processes, dynamically tunable, on the server. The reduction
in the total number of processes handled by the server system
results in more processing time available for RDBMS activity. Thus
higher RDBMS transaction throughput can be obtained on the
server system.
But the problem for the TPC-A, and for certain large customer
configurations, is not the only ability of the Oracle Server to
process transactions, but also the ability of the operating
system to handle huge numbers of incoming connections.
There is one incoming connection for each client. Most UNIX
operating systems have a limit on how many such connections they can
handle. Even if a particular operating system allows a large number of
connections, each takes some amount of overhead to manage.
In order to service all 6150 terminals, we selected a 3-tier hardware
environment where the middle tier, using a TPM, acted as a terminal
concentrator. The high-end TPC-A architecture looked like the following.
The Application Servers, which contain the Pro*C statements used to
perform the transaction also run on the terminal concentrator machine
in order to offload as much work from the database serve as possible.
They send the compiled SQL over SQL*Net to the Oracle7 Server processes.
ORACLE7 TPS-A CLIENT/SERVER ARCHITECTURE
| Client | | Terminal | | Server |
| ________ | | Concentrator | | |
| | Client | |TPM | | | |
| | Process|_|_____|__ _____ | | |
| |________| |Comm | | | | | | |
|____________| | | | | | | |
| |__| | | | |
____________ | | TPM | | | |
| Client | | ___| | _______ | | ________ _______ |
| ________ | | | | |_| |__|_______|__| Oracle | | | |
| | Client | |TPM | | | | |Appl. | |SQL*Net| | Server |__| | |
| | Process|_|_____|_| |_____| |Server | | | | Process| | | |
| |________| |Comm | |_______| | | |________| | | |
|____________| | | | | | |
|_______________________| | | | |
| | | |
____________ _______________________ | |Oracle7| |
| Client | | Terminal | | |Server | |
| ________ | | Concentrator | | | | |
| | Client | |TPM | | | | | |
| | Process|_|_____|__ _____ | | __________ | | |
| |________| |Comm | | | | _______ |SQL*Net| | Oracle | | | |
|____________| | | | |_| |__|_______|__| Server |__| | |
| |__| | |Appl. | | | | Process| | | |
____________ | | TPM | |Server | | | |________| |_______| |
| Client | | ___| | |_______| | | |
| ________ | | | | | | | |
| | Client | |TPM | | | | | | |
| | Process|_|_____|_| |_____| | | |
| |________| |Comm | | | |
|____________| | | | |
|_______________________| |________________________|
Clients = 6150 Terminal concentrators = 17
TP Monitor instances = 17
Application server processes Oracle Server processes
= 17*8 = 17*8
The TPM is the software component of the terminal concentrator. In this role
it offloads terminal handling from the the machine running Oracle Server.
Since more than one terminal concentrator can be configured, whereas the
database in this case had to run on a single machine, concentrator machines
can be added until the performance of the back-end machine was optimized.
This three-tier solution resulted in the outstanding transaction throughput
announced with Oracle7 Server. Even with Oracle Parallel Server, it may pay
to offload the terminal handling so that the cluster can be exclusively used
for database operations.
Can you summarize the performance discussion for me?
Depending on the number of users required, different architectures may be
used in a client/server environment to maximize performance:
1) For a small number of users, the traditional Oracle two-task
architecture can be used. In this case, there is a one-to-one
correspondence between client processes and server processes. It's
simple, straightforward, and efficient.
2) For a large number of users, Multi Threaded Server might be a better
approach. Although some tuning may be required, Multi Threaded Server
can handle a relatively large number of users for each machine size
compared to the traditional Oracle approach. Using this approach,
customers will be able to handle many hundreds of users on many
platforms. Furthermore, current Oracle applications can move to this
environment without change.
3) For a very large number of users, where transactions are simple and
terminal input concentration is the overriding performance issue, a
3-tier architecture incorporating a TPM may be useful. In this case,
terminal concentration is handled by the TPM in the middle tier. As
you might expect, it is a more complex environment requiring more
system management. For existing Oracle customers, significant Oracle
application modifications will be required.
Oracle provides all of these choices.
Performance using Oracle's XA Library
=====================================
Are there any performance implications to using the XA library (in other
words, to using TPM-managed transactions)?
(1) The XA library imposes some performance penalty. You should use
TPM-managed transactions only if you actually need them. Even if you
are getting the one-phase commit optimization, the code path is
longer because we need to map back and forth between external
formats and internal ones. Also, prior to 7.1, XA requires you
to release all cursors at the end of a transaction, which results
in extra parsing. Even with shared cursors, there is time spent
looking up the one you need and re-validating it. This has been
improved for 7.1.
(2) If you need to use two-phase commit, this will incur additional cost
since extra I/Os are required. If you do need 2PC, you need to account
for that when sizing the application.
(3) Although some TPMs allow parallel execution of services (such as Tuxedo's
"tpacall"), this will not normally enhance performance unless different
resource managers are being used. In fact, Oracle Server must serialize
accesses to the same transaction by the same Oracle instance, and the
block/resume code will in fact degrade performance in that case compared
to running the services sequentially.hello,
the role is the same on all plattforms. the reports server takes requests for running reports, spawns an engine that executes the request. in addition to that, the server also provides scheduling services and security features for the reports environment.
regards,
the oracle reports team -
Oracle database performance after server reboot
hi masters,
this is not some kind of question, but a discussion. some statements come from our client that after weekly reboot of system, the oracle database performance is low for some time and increase after some time(say 2 days).
i think it is but obvious, because at reboot oracle flushes all cache, and temporary space, so it need to re parse the sql statements and perform some disc I/O's so it might need some time and hence performance will degrade.
but at the same time some people claim that after reboot their database performance is better than their normal performance for some days. it seems controversial that's why i am posting it here.
what might be the reason behind this?? prior can have a valid reason of hard parsing, but what with second case??
any clarification is highly appreciated...
thank you
regadrs
VDVikrant,
You should wait for some time buddy, its weekend ;-) .
this is not some kind of question, but a discussion. some statements come from our client that after weekly reboot of system, the oracle database performance is low for some time and increase after some time(say 2 days).i think it is but obvious, because at reboot oracle flushes all cache, and temporary space, so it need to re parse the sql statements and perform some disc I/O's so it might need some time and hence performance will degrade.
>
I would start from saying that checking the performance when the system just started, is a wrong approach. There would be lots of IOs , parsing, calculations(related to memory allocations) happening so there would be a delay/bad performance at that time. Very simple example can be parsing, another can be memory allocation. Oracle doesn't allocate the entire memory in the instance startup that is allocated to the memory areas but allocates just the bare minimum that is needed to start the instance and than after the startup, it keeps on allocating the memory. So surely enough, with the startup and after a while of it, there would be a different performance than that after the instance hsa already been started and the workload informations have started coming up.
Its correct that Oracle would deallocate all the caches with the reboot as the instance is on the memory(physical) and with the reboot , that would be flushed including the SGA which is allocated over it. Temporary tablespace is now not freed with the reboot. I guess its a rather illogical thing to do but that's what is there now. Oracle keeps the segment allocated even after the reboot is issued, hence the reason for larger temporary tablespaces.
>
but at the same time some people claim that after reboot their database performance is better than their normal performance for some days. it seems controversial that's why i am posting it here.
what might be the reason behind this?? prior can have a valid reason of hard parsing, but what with second case??
>
This should not come as a surprise once we understand what might be happening with this process. Assume a situation where you have undersized caches. For example, shared pool . which is very heavily used for database , if this is going to be undersized and you are not using automatic memory management, you won't be enjoying the dynamic management of this parameter. Now, if you do lots of parsing , thanks to your wrongly written queries, you would eventually end up filling up shared pool to its max thus leaving no space for incoming new hard parsed cursors. Here , if you can't manage to add more memory to add to it, the only solution left would be to flush the shared pool( as good as rebooting the db because this would do the same) and than make space for the new cursors. The performance is going to be better becausethe cursors would not be getting flushed out immediately and will be kept in the shared pool as long as its not filled up again.Once you have reached to limit of it, again there would be performance benefit. So there are always odds added to the statements like this that I rebuilt my index , I got better, I rebooted my server, my querie are much faster now. Most of the time when these kind of statements are given, they are based on what we have seen, without understading what actually might have happened. So I would siggest to hear the statement but not take them as a rule of thumb to follow.
Hope this all makes some sense for you and would help somewhat.
Aman.... -
Weblogic server performance is very slow and memory consumption is 99%
I am facing one critical issue with the weblogic server..
The server performance is very slow and one of the process is consuming more that 99% of the memory. Bouncing the server is not resolving the issue.
Can see the memory usage below...
Tasks: 134 total, 2 running, 132 sleeping, 0 stopped, 0 zombie
Cpu(s):100.0%us, 0.0%sy, 0.0%ni, 0.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 7990656k total, 7947652k used, 43004k free, 9164k buffers
Swap: 16386260k total, 4691704k used, 11694556k free, 56352k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
10263 oracle 24 0 10.9g 6.5g 14m S 99.2 85.3 34:31.52 java
7409 oracle 16 0 12764 768 508 S 0.3 0.0 0:16.45 top
Can some body help me on this.
Thanks in advance.
-PrasadUse the weblogic forum-
WebLogic Server - Upgrade / Install / Environment / Migration -
ORACLE SERVER AND UNIX TP MONITOR-1
제품 : ORACLE SERVER
작성날짜 : 2002-05-17
====================================================================
Subject: Oracle Server and UNIX Transaction Processing Monitors - 1
=====================================================================
PURPOSE
This file contains commonly asked questions about Oracle Server and UNIX
Transaction Processing Monitors (TPMs). The topics covered in this article are
o What is a Transaction Processing Monitor (TPM)?
o What is the X/Open Distributed Transaction Processing Model?
o How does the Oracle Server works with TPMs?
o How should I position TPMs with my customer?
o What Oracle products must a customer purchase?
o Where can my customer purchase a TPM?
o Availability and packaging
Explanation & Example
What is a Transaction Processing Monitor?
=========================================
Under UNIX, a Transaction Processing Monitor (TPM) is a tool that coordinates
the flow of transaction requests between front-end client processes that issue
requests and back-end servers that process them. A TPM is used as
the "glue" to coordinate transactions that require the services of several
different types of back-end processes, such as application servers and
resource managers, possibly distributed over a network.
In a typical TPM environment, front-end client processes perform screen
handling and ask for services from back-end server processes via calls to the
TPM. The TPM then routes the requests to the appropriate back-end server
process or server processes, wherever they are located on the network. Through
configuration information, the TPM knows what services are available and where
they are located. Generally, the back-end server processes are specialized so
that each one handles one type of requested service. The TPM provides
location transparency as well and can send messages through the network
utilizing lower-level transport services such as TCP/IP or OSF DCE.
The back-end servers process the requests as necessary and
return the results back to the TP monitor. The TP monitor then routes
these results back to the original front-end client process.
A TPM is instrumental in the implementation of truly distributed processing.
Front-end clients and back-end processes have no knowledge of each
other. They operate as separate entities, and it is this concept that provides
flexibility in application development. Front-end and back-end processes are
developed in the UNIX client-server style, with each side optimized for its
particular task. Server functionality can be deployed in stages, which makes
it easy to add functionality as needed later in the product cycle. It also
makes it easy to distribute both the front-end and back-end processes
throughout the network on the most appropriate hardware for the job. In
addition, multiple back-end server processes of the same type might be
activated to handle increasing numbers of users.
What is the X/Open Distributed Transaction Processing Model?
============================================================
The X/Open Transaction Processing working group has been working
for several years to establish a standard architecture to implement
distributed transaction processing on open systems. In late 1991,
X/Open published the initial Distributed Transaction Processing (DTP)
model specification and defined the first of several interfaces that
exist between the components of the model. Subsequently, other publications
and a revised model specification have been published.
An important function of the TPM in the X/Open DTP model is the
synchronization of any commits and rollbacks that are required to complete
a distributed transaction request. The Transaction Manager (TM) portion
of the TPM is the entity responsible for ordering when distributed commits
and rollbacks will take place. Thus, if a distributed application program
is written to take advantage of the TM portion of the TPM, then it,
and not the DBMS, becomes responsible for enabling the two-phase commit
process. Article 2 has more detail on this model.
How does the Oracle Server work with TPMs?
==========================================
When a TPM is used without invoking an X/Open TM component to manage the
transactions, Oracle Server needs no special functionality. The transaction
will be managed by Oracle itself. However, when the TPM X/Open TM component
is used to manage the transaction, the Oracle Server, that is the Oracle DBMS,
acts as a Resource Manager--a type of back-end process. In the case of
TPM-managed transactions, the TM needs a way to tell the RMs about the stages
of the transaction. This is done by a standard, X/Open defined interface
called XA. Article 2 of of this document gives more information about both
the X/Open model and Oracle7's use of XA.
Because the XA interface provides a standard interface between the TM and the
resource manager, it follows that the TM can communicate with any XA-compliant
resource manager (e.g., RDBMS), and, conversely, that a resource manager can
communicate with any XA-compliant TM. Thus, the Oracle Server, beginning with
Oracle7, works with any XA-compliant TM.
How should I position TPMs with my customer?
============================================
There's been a great deal of confusion about the need for TPM technology. Some
software suppliers, most notably IBM, will assert that a TPM like CICS is a
necessary requirement for high volume OLTP. Other vendors will assert that
there is seldom a need for such technology. And yet others promote TPMs as
providers of higher transaction throughput.
From Oracle's standpoint, customers might choose TPM technology under any of
the following conditions:
1. For heterogeneous database access, especially for 2PC capability
This means that a TPM can be used to coordinate 2PC between Oracle
DBMS and any other XA-compliant database, such as Informix. This
does NOT provide SQL heterogeneity - SQL calls to Oracle DBMS may be
different than SQL calls to Informix. The TPM handles the routing,
communication, and two-phase commit portion of the transaction, but
does not translate one type of SQL call into another.
2. For transaction monitoring and workload control
The leading TPMs supply tools to actively manage the flow of
transactions between clients and servers and to load balance the work
load across all available processors on a network, not just on a
single multi-processor system. Some TPMs also have the ability to
dynamically bring up additional back-end services during peak work
hours.
3. For more flexible application development and installation
One of the key features of the DTP model is application modularity.
Modularity, that is, the decomposition of a large program into small,
easily defined, coded and maintained "mini-programs" makes it easy to
add new functionality as needed. Modularity also makes it much easier
to distribute the front-end and back-end processes and the resource
managers across hardware throughout a network.
4. For isolating the client from details of the data model
By using the service oriented programming model, the client program
is unaware of the data model. The service can be recoded to use a
different one with no change to the client. To get this advantage,
the application developer must explicitly code the server and client
to fit the service model.
5. For connection of thousands of users
TP Monitors, because of their three-tier architecture, can be used
to connect users to an intermediate machine or machines, removing
the overhead of handling terminal connections from the machine
actually running the database. See Article 4 for more information.
There are also several cases where TPM technology is not the right answer.
These include:
1. If the customer is simply looking for a performance improvement
The customer may have heard a theory that "higher performance
is possible for large scale applications only if they use a
TP monitor". First, no performance gain can be achieved for
existing applications; in fact, they won't even run under a TP
Monitor without recoding. Second, performance improvements have
only been documented for large numbers of users, and "large"
means many hundreds or thousands. Without a TP Monitor,
Oracle Server can handle several hundred users with its normal
two-task architecture and several times that using the Multi
Threaded Server. For more on performance, see Article 4.
2. If the customer has made large investment in his existing Oracle
applications
TP monitor applications must be designed from the ground up to take
advantage of TP monitor technology. Current Oracle customers will find
it difficult to "retrofit" a TP monitor to their existing applications.
The Multi Threaded Server, on the other hand, allows the use of
existing Oracle applications without change.
3. If the customer is committed to the Oracle tool set
Currently, none of Oracle's front-end tools (Oracle Forms, etc.) is
designed to work with TP monitors. It is possible to invoke a
TP Monitor by using user exits. However, the fact that the TP
Monitor model hides the data model from the client means that only
the screen display parts of Forms can be used, not the automatic
mapping from screen blocks to tables.
4. If the customer does not have a staff of experienced software engineers
This is still very young technology for UNIX. There is not a lot of
knowledge in the industry on how to build TP monitor applications or
what techniques are most useful and which are not. Furthermore,
integrating products from different vendors, even with the support
of standard interfaces, is more complex than deploying an integrated
all-Oracle solution. Because TP monitor technology is fairly
complex, we recommend that you let the TP monitor supplier promote
the virtues of their technology and differentiate themselves from
their competitors.
What Oracle products must a customer purchase?
==============================================
If your customer is only interested in building Oracle-managed TP Monitor
transactions, the only Oracle products required are the Oracle Server
and the appropriate Oracle precompiler for whatever language the
application is being written in--most likely C or Cobol. If TPM-managed
transactions are required, the Oracle7 Server with the distributed option
is also required. SQL*Net is optional because the TPM takes care of the
network services. Article 2 describes when you would choose to have the TP
Monitor manage the transactions.
Where can my customer purchase a TPM?
=====================================
There are many vendors offering the UNIX TPM products. (Oracle does not
relicense TPMs.) Information on the most well known products is provided
below:
The following support XA:
Product & Vendor FCS Known OS/Platform Ports
"TUXEDO System/T" 1986 UNIX SVR4 & SVR3: Amdahl, AT&T,
UNIX System Laboratories Bull, Compaq, Dell, Fujitsu, ICL,
190 River Road Motorola, Olivetti, Pyramid,Sequent,
Summit, NJ 07901 Sun, Toshiba, Unisys, NCR, Stratus
Other: IBM AIX, HP/UX, DEC Ultrix
"TOP END" 1992 UNIX SVR4: NCR
NCR Corporation
1334 S. Patterson Blvd.
Dayton, OH 45479
"ENCINA" 1992 IBM AIX, HP, Sun (SunOS and Solaris)
Transarc Corporation Other: OS/2, DOS, HP-UX, STRATUS
707 Grant Street (Depends on DCE)
Pittsburgh, PA 15219
"CICS/6000" 1993 AIX: IBM
IBM Corporation (Depends on DCE)
"CICS 9000" 1994 HP-UX
HP
The following do not currently support XA:
Product & Vendor FCS Known OS/Platform Ports
"VIS/TP" unknown unknown
VISystems, Inc.
11910 Greenville Avenue
Dallas, TX 75243
"UniKix" 1990 UNIX: ARIX, AT&T, NCR, Pyramid,
UniKix Sequent, Sun, Unisys
"MicroFocus 1993 SCO Unix, AIX
Transaction System"
Micro Focus
26 West Street
Newbury RG13 1JT
UK
There are also several third parties who are reselling the products listed
above.
In addition, Groupe Bull, Digital, Siemens-Nixdorf, and several other hardware
vendors are planning to redesign their proprietary TPMs to be XA-compliant and
suitable for use on UNIX systems.
Availability and Packaging
==========================
On what platforms is the XA Library available?
Oracle provides the XA interface with Oracle7 Server on all platforms that
support an XA-compliant TPM. Support for XA is included as part of the
Oracle7 Server distributed option and has no extra charge in and of itself.
Which version of XA does Oracle Server support?
Oracle7 Server supports the Common Application Environment (CAE) version of
XA, based on the specification published by X/Open in late 1991. It will
require that the TM also be at that level. This means Tuxedo /T version 4.2,
for example.
Oracle Server supports all required XA functions. There are some optional
features Oracle Server does not support, such as asynchronous operation.
None of those options affect application programming.
Page (2/4)
This file contains commonly asked questions about Oracle Server and UNIX
Transaction Processing Monitors (TPMs). The topics covered in this article are
o Oracle Server Working with UNIX TPMs
o TPM Application Architecture
The questions answered in part 2 provide additional detail to the information
provided in part 1.
Oracle Server Working with UNIX TP Monitors
===========================================
Do I need XA to use Oracle Server with TPMs? If I don't use it, what are
the consequences?
There are a number of real applications running today with Oracle Server and
TPMs but not using XA. To use a TPM with Oracle without using XA, the user
would write an "application server" program which could handle one or more
"services". For example, a server program might handle a service called
"debit_credit". The key requirement is that the entire transaction,
including the "commit work", must be executed within a single service. This
is the restriction which XA will remove, as we'll see later. Each
server process can serially handle requests on behalf of different clients.
Because a server process can handle many client processes, this can
reduce the total number of active processes on the server system,
thereby reducing resource requirements and possibly increasing overall
throughput.
When Oracle is used with a TPM in this mode, we call it an Oracle-managed
transaction since the transaction commit or rollback is done with a SQL
statement.
What is XA? How does XA help Oracle7 work with UNIX TPMs?
XA is an industry standard interface between a Transaction Manager and a
Resource Manager. A Resource Manager (RM) is an agent which
controls a shared, recoverable resource; such a resource can be
returned to a consistent state after a failure. For example, Oracle7 Server
is an RM and uses its redo log and undo segments to be able to do this.
A Transaction Manager (TM) manages a transaction including the
commitment protocol and, when necessary, the recovery after a failure.
Normally, Oracle Server acts as its own TM and manages its own commitment
and recovery. However, using a standards-based TM allows Oracle7 to
cooperate with other heterogeneous RMs in a single transaction.
The commonly used TPMs include a TM component for this purpose. In order to
use the TM capability of the TPM rather than Oracle7's own transaction
management, the application uses a transaction demarcation API (called TX)
provided by the TPM rather than the SQL transaction control statements (e.g.
"commit work"). For each TX call, the TM then instructs all RMs, by the
appropriate XA commands, to follow the two-phase commit protocol. We
call this a TPM-managed transaction.
The following picture shows these interfaces within a monolithic application
program model. This is the model most commonly described in the
DTP literature. We'll see later what the picture looks like when we add
Oracle7 and when we switch to a modularized client-server application
program model.
| |
| |
| Application Program (AP) |
| |
| |
| | | |
Resource Manager API | | | |
(e.g. SQL) -----|--|------------- | TX API
| | v | |
--------|------------- | |
| v | | v
---------------------- | | --------------------
| | | | | |
| Resource | | |<----->| Transaction |
| Managers | |--- | Manager |
| (RMs) | |<-------->| (TM) |
| |--- | |
| |<----------->| |
---------------------- XA --------------------
Interface
The XA interface is an interface between two system components, not
an application program interface; the application program does
not write XA calls nor need to know the details of this interface.
The TM cannot do transaction coordination without the assistance of
the RM; the XA interface is used to get that assistance.
How does the DTP Model support client-server?
The above picture was actually simplified to make it easier to explain
the role of XA. In a true distributed transaction architecture, there
are multiple applications, each with an Application Program, a Resource
Manager, and a Transaction Manager. The applications communicate by
using a Communication Resource Manager. The CRM is generally provided
as a component of the TPM. It includes the transaction information when
it sends messages between applications, so that both applications can
act of behalf of the same transaction. The following picture
illustrates this:
Client Application
| AP |
||| | |
SQL ||| | TX | CRM
||V V | API
-||-- ----- |
| |V | | | V
--|-- |<---| | -----
| V || | | | |
----- |<----| TM |<-->| CRM |
| || | |XA+ | |
| RMs |<-----| | -----
| | XA | | A
----- ----- | Server Application
| -----------------------------
| | AP |
| -----------------------------
| ||| | |
| SQL ||| | TX | CRM
| ||V V | API
| -||-- ----- |
| | |V | | | V
| --|-- |<---| | -----
| | V || | | | |
| ----- |<----| TM |<-->| CRM |
| | || | |XA+ | |
| | RMs |<-----| | -----
| | | XA | | A
| ----- ----- |
| |
| |
-------- |
/ |
/ |
/ |
Most TP Monitor products include both a TM and a CRM, and also provide
additional functions such as task scheduling and workload monitoring.
What is XA+? What does Oracle need to do to comply with it?
XA+ is an interface that lets the X/Open model actually be distributed
because it allows a communication resource manager to tell a TM on the
server that a message from a client just came in for a particular
transaction. Oracle is not currently planning to provide an X/Open
communication resource manager, so we don't have any plans right now
to do XA+. Version 2 of the DTP model paper from X/Open describes it.
The status of the current XA+ specification is "snapshot".
When would I choose an Oracle-managed transaction vs a TPM-managed
transaction?
Oracle Server is very efficient at managing its own transactions. If
the TPM manages the transaction, in general some additional overhead
will be incurred.
The two main reasons a customer might prefer to use a TPM-managed
transaction are as follows:
(1) He may need to update RMs from different vendors. Experience so far
has been that the most common case is wanting to update both Oracle and
a TP Monitor managed resource such as a transactional queuing service
in the same transaction (see Article 3).
(2) He may want to use the model of having several different services in
a transaction, even to the same database. For example, the
"debit_credit" service could be split into a "debit" service and a
"credit" service. This is a very attractive model, but this type of
modularity does exact a performance penalty (see Article 4).
Can I get a version of XA to run on Oracle Server version 6?
No, the XA functionality uses two underlying mechanisms in the Oracle
Server which are not available in version 6: two-phase commit and
session switching. The upi calls for these functions do not not exist
in version 6.
When would I use XA vs Oracle7 to coordinate all-Oracle distributed
transactions?
Generally speaking, Oracle Server should be used to coordinate all-Oracle
distributed transactions. The main reason for using XA to coordinate
transactions would be that you want to use the TP Monitor service-oriented
architecture. That is, you would like to construct an application built of
services and service requests in order to benefit from the modularity and
workload control such an environment provides.
TP Monitor Application Architecture
===================================
What might a TP Monitor application look like?
Most TPM applications will consist of two more more programs, where
there are front-end client programs which request services and back-end
server programs which provide services. In this case, the TPM supplies an
additional capability which is transactional communication. The client
describes the boundaries of the transaction, through the use of the TX API,
and the TPM relays that transaction information to each requested service.
The overall application structure generally looks like the following in the
client-server model. The "TP Monitor Services" box is not necessarily a
process. It could be one or more processes, or just libraries coordinating
through shared memory. Each client process and server process could be on
a different machine. Normally, the application server processes would be
connected to their Oracle Server processes using the IPC driver; the TPM
would be used to deliver messages between application client processes on
one machine and application server processes on another. However, the
application server processes could also be connected with the standard
Oracle SQL*Net to shadow processes on different machines. This might be
useful if one of the databases was on a machine which did not support TPMs.
|Application| |Application| |Application|
| Client 1 | | Client 2 | | Client 3 |
| | | | | |
\ TPM API | TPM API / TPM API
| |
| TP Monitor Services |
| |
| --------------------- |
| | Transaction Manager | |
---------------|---------------|---------------------
TPM API | | XA | XA | TPM API
| | inter- | inter- |
| | face | face |
| | | |
----------- | | -----------
|Application| | | |Application|
| Server 1 |--- ---| Server 2 |
| (Pro*C) | | | | (Pro*C) |
| SQL | SQL
| | | |
| Resource ----------- ----------- |
| Manager | | | | |
| | Oracle7 | | Oracle7 | |
| | Server | | Server | |
| | Process | | Process | |
| | | | | |
| ----------- ----------- |
| | | |
| ----------------------------------------------------- |
| | | |
| | SGA | |
| | | |
| ----------------------------------------------------- |
| |
Application client programs might be written in C and be linked with
TPM libraries. Alternatively, they could use a screen painter product.
Application server programs would be written in Pro*C or Pro*COBOL and
be linked with TPM libraries, the normal Oracle7 user-side libraries
and libxa.a. The Oracle7 Server process is the regular Oracle7 executable.
More complicated application architectures can also be constructed. Most of
the TPMs allow a server to become a client of another service, so you can
involve additional servers.
Could I use Oracle7's Multi Threaded Server as the SQL*Net connection in the
previous picture?
Yes, but that will not be needed in many cases. For example, both
application server processes in the previous picture could talk to a
single Oracle7 Server process through the Multi Threaded Server in the
previous picture. However, since the TPM architecture typically reduces
the number of server processes, the reduction in processes using Multi
Threaded Server may be less significant than in an architecture without
TPMs. If the application will use database links, however, then MTS will
be required.
How do I write an Oracle TP Monitor application?
The actual API used to talk to the TPM varies between vendors, so you need
to get the documentation from the vendor. However, all have a way to
indicate where a transaction begins and ends and a way to send a request
and receive a response from a client to a server. Some use an RPC model,
some use a pseudo-RPC model, and some use a send/receive model. The TX API
described earlier is a subset of the TPM API as defined by each of
the TPM providers.
The client program and server program might look something like the
following examples. We h (such as Tuxedo's
"tpacall
Reference Ducumment
---------------------hello,
the role is the same on all plattforms. the reports server takes requests for running reports, spawns an engine that executes the request. in addition to that, the server also provides scheduling services and security features for the reports environment.
regards,
the oracle reports team -
Oracle Spatial Performance with 10-20.000 users
Does anyone have any experience when Oracle Spatial is used with say 20.000 concurrent users. I am not interested in MapViewer response time, but lets say there is:
- an app using 800 different tables each having an sdo_geometry column
- the app is configured with different tables visible on different view scales
- let's say an average of 40-50 tables is visible at any given time
- some tables will have only a few records, while other can hold millions.
- there is no client side caching
- clients can zoom in/out pan.
Anwers I am interested in:
- What sort of server would be required
- How can Oracle serve all that data (each Refresh renders the map and retrieves the data over the wire as there is no client side caching).
- What sort of network infrastructure would be required.
- Can clients connect to different servers and hence use load balancing or does Oracle have an automatic mechanism for that?
Thanks in advance,
PatrickPatrick, et al.
There are lots of things one can do to improve performance in mapping environments because of a lot of the visualisation is based on "background" or read-only data. Here are some "tips":
1. Spatially sort read-only data.
This tip makes sure that data that is close to each other in space are next to each other on disk! Dan gave a good suggestion when he referenced Chapter 14, "Reorganize the Table Data to Minimize I/O" pp 580- 582, Pro Oracle Spatial. But just as easily one can create a table as select ... where sdo_filter() where the filtering object is an optimized rectangle across the whole of the dataset. (This is quite quick on 10g and above but much slower on earlier releases.)
When implementing this make sure that the created table is created such that its blocks are next to each other in the tablespace. (Consider tablespace defragmentation beforehand.) Also, if the data is READ ONLY set the PCTFREE to 0 in order to pack the data up into as small a number of blocks as possible.
2. Generalise data
Rendering spatial data can be expensive where the data is geometrically detailed (many vertices) esp where the data is being visualised at smaller scales than it was captured at. So, if your "zoom thresholds" allow 1:10,000 data to be used at 1:100,000 then you are going to have problems. Consider pre-generalising the data (see sdo_util.simplify) before deployment. You can add multiple columns to your base table to hold this data. Be careful with polygon data because generalising polygons that share boundaries will create gaps etc as the data is more generalised. Often it is better to export the data to a GIS which can maintain the boundary relationships when generalising (say via topological relationships).
Oracle's MapViewer has excellent on-the-fly generalisation but here one needs to be careful. Application tier caching (cf Bryan's comments) can help here a lot.
3. Don't draw data that is sub-pixel.
As one zooms out objects become smaller and smaller until they reach a point where the whole object can be drawn within a single pixel. If you have control over your map visualisation application you might want to consider setting the SDO_FILTER parameter "min_resolution" flag dynamically so that its value is the same as the number of meters / pixel (eg min_resolution=10). If this is set Oracle Spatial will only include spatial objects in the returned search set if one side of a geometry's MBR is greater than or equal to this value. Thus any geometries smaller than a pixel will not be returned. Very useful for large scale data being drawn at small scales and for which no selection (eg identify) is required. With Oracle MapViewer this behaviour can be set via the generalized_pixels parameter.
3. SDO_TOLERANCE, Clean Data
If you are querying data other than via MBR (eg find all land parcels that touch each other) then make sure that your sdo_tolerance values are appropriate. I have seen sites where data captured to 1cm had an sdo_tolerance value set to a millionth of a meter!
A corollary to this is make sure that all your data passes validation at the chosen sdo_tolerance value before deploying to visualisation. Run sdo_geom.validate_geometry()/validate_layer()...
4. Rtree Spatial Indexing
At 10g and above lots of great work went in to the RTree indexing. So, make sure you are using RTrees and not QuadTrees. Also, many GIS applications create sub-optimal RTrees by not using the additional parameters available at 10g and above.
4.1 If your table/column sdo_geometry data contains only points, lines or polygons then let the RTree indexer know (via layer_gtype) as it can implement certain optimizations based on this knowledge.
4.2 With 10g you can set the RTree's spatial index data block use via sdo_pct_free. Consider setting this parameter to 0 if the table/column sdo_geometry data is read only.
4.3 If a table/column is in high demand (eg it is the most commonly used table in all visualisations) you can consider loading (a part of) the RTree index into memory. Now, with the RTree indexing, the sdo_non_leaf_tbl=true parameter will split the RTree index into its leaf (contains actual rowid reference) and non-leaf (the tree built on the leaves) components. Most RTrees are built without this so only the MDRT*** secondary tables are built. But if sdo_non_leaf_tbl is set to true you will see the creation of an additional MDNT*** secondary table (for the non_leaf part of the rtree index). Now, if appropriate, the non_leaf table can be loaded into memory via the following:
ALTER TABLE MDNT*** STORAGE(BUFFER_AREA KEEP);
This is NOT a general panacea for all performance problems. One should investigate other options before embarking on this (cf Tom Kyte's books such as Expert Oracle Database Architecture, 9i and 10g Programming Techniques and Solutions.)
4.4 Don't forget to check your spatial index data quality regularly. Because many sites use GIS package GUI tools to create tables, load data and index them, there is a real tendency to not check what they have done or regularly monitor the objects. Check the SDO_RTREE_QUALITY column in USER_SDO_INDEX_METADATA and look for indexes with an SDO_RTREE_QUALITY setting that is > 2. If > 2 consider rebuilding or recreating the index.
5. The rendering engine.
Whatever rendering engine one uses make sure you try and understand fully what it can and cannot do. AutoDesk's MapGuide is an excellent product but I have seen it simply cache table/column data and never dynamically access it. Also, I have been at one site which was running Deegree and MapViewer and MapViewer was so fast in comparison to Deegree that I was called in to find out why. I discovered that Deegree was using SDO_RELATE(... ANYINTERACT ...) for all MBR queries while MapViewer was using SDO_FILTER. Just this difference was causing some queries to perform at < 10% of the speed of MapViewer!!!!
6. Consider "denormalising" data
There is an old adage in databases that is "normalise for edit, denormalise for performance". When we load spatial data we often get it from suppliers in a fairly flat or normalised form. In consort with spatial sorting, consider denormalising the data via aggregations based on a rendering attribute and some sort of spatial unit. For example, if you have 1 million points stored as single points in SDO_GEOMETRY.SDO_POINT which you want to render by a single attribute containing 20 values, consider aggregating the data using this attribute AND some sort of spatial BUCKET or BIN. So, consider using SDO_AGGR_UNION coupled with Spatial Analysis and Mining package functions to GROUP the data BY <<column_name>> and a set of spatial extents.
6. Tablespace use
Finally, talk to your DBA in order to find out how the oracle database's physical and logical storage is organised. Is a SAN being used or SAME arranged disk arrays? Knowing this you can organise your spatial data and indexes using more effective and efficient methods that will ensure greater scalability.
7. Network fetch
If your rendering engine (app server) and database are on separate machines you need to investigate what sort of fetch sizes are being used when returning data from queries to the middle-tier. Fetch sizes for attribute only data rows and rows containing spatial data can be, and normally are, radically different. Accepting the default settings for these sizes could be killing you (as could the sort_area_size of the Oracle session the application server has created on the database). For example I have been informed that MapInfo Pro uses a fixed value of 25 records per fetch when communicating with Oracle. I have done some testing to show that this value can be too small for certain types of spatial data. SQL Developer's GeoRaptor uses 100 which is generally better (but this one can modify this). Most programmers accept defaults for network properties when programming in ADO/ODBC/OLEDB/JDBC: just be careful as to what is being set here. (This is one of the great strengths of ArcSDE: its TCP/IP network transport is well written, tuneable and very efficient.)
8. Physical Format
Finally, while Oracle's excellent MapViewer requires data its spatial data to be in Oracle, other commercial rendering engines do not. So, consider using alternate, physical file formats that are more optimal for your rendering engine. For example, Google Earth Enterprise "compiles" all the source data into an optimal format which the server then serves to Google Earth Enterprise clients. Similarly, a shapefile on local disk to the application server (with spatial indexing) may be faster that storing the data back in Oracle on a database server that is being shared with other business databases (eg Oracle financials). If you don't like this approach and want to use Oracle only consider using a dedicated Oracle XE on the application server for the data that is read only and used in most of your generated maps eg contour or drainage data.
Just some things to think about.
regards
Simon -
Using archivelog and control file from other Oracle server
I am still bothered with my backup process.
I have 2 AIX boxes (same model, say A and B); both have BAAN 5 and Oracle 10g R2 on. Right now my colleague insists to use the export pump (cold backup) from Prod Oracle server (A) to restore the Oracle server on Box B. The Prod server has the archivelog mode turn on. But it will miss any transaction data from import pump till the crash point of Box A. So this is my confusion.
Can I pass the control files and archivelog files form Box A (prod server) to Box B and use them to restore the Box B as the latest Prod server? How?
I tried to convince them to use the RMAN backup? But not successful?
I think the best way is probably to use the Oracle Data guard. However, there is always one concern to my manager and colleague, that such process will cause the data on the restored server (failover, Box B) not recognizable by the BAAN, which define the objects (tables).
ThanksPerforming a logical backup is not useful to restore to the point of failure. The only valid and available option is a hot backup/archivelog mode. Your recovery manager backup perform a controlfile and redologfile backup, so those can be restored at the destination. You must take care of the way you perform the backup, and ensure the paths where your backup is being deposited are visible by the second node. A shared storage with same mount points is suitable in this case. A tape robot configured at both nodes is also a suitable solution.
Recovery manager perfoms a controlfile and spfile restore, too. This rman command perform the action:
SET DBID <DBID of the database,for which you want to restore the controlfile>;
RESTORE CONTROLFILE FROM <name_of_backupiece_which_contains_the_controlfile backup>';
I don't see any problem on the recovery manager side, and technically speaking, on the Oracle side it is perfectly possible to restore your database at a remote location. I don't know what happens on the BAAN side, if you are required to have it configured to be operative on the target node. You could try to clone your database at the node B, configure BAAN and prepare the proceduere in case of failure.
Configuring a dataguard is also a recomended action. So it is to think about Cold Failed Over clusters. I have recently performed a CFC configuration with BAAN. No problem it works smoothly.
~ Madrid. -
Dynamic column name for Oracle Server Procedure
hello,
May be it's not a right place for my question, but I hope that some of you can help me.
I have source table A that has data of the same kind in 62 fields, let's say employee count for 62 departments. I do some processing with these counts(sum(emp_count) of the same departments for different locations) and store these counts in the variables EC1, EC2, EC3 .. EC62.
Then I need to insert them into my table that has very simple structure let's say table B (dept_no Number(3),
emp_count Number(6))
I want to do a loop and use dynamic colunm EMP_COUNT, something like this:
FOR k in 1..62 LOOP
EMP_COUNT := 'EC'| |k;
IF EMP_COUNT <> O THEN
insert into B
VALUES ( K , EMP_COUNT);
END IF;
END LOOP;
I understand that I need to take a Value of my dynamic colunm EMP_COUNT. Is it possible to do in Oracle Server Procedure? Any idea?
Any help will be apreciated.
nullAfter having worked in other environments where you can manipulate data records with postitional parameters, Oracle can be frustrating because it demands the explicit use of field names in DML/Select statements. A way around this is to use DBMS_SQL or Dynamic Sql. I have used DBMS_SQL to create and execute dynamic statements with great success. Since you are creating, compiling, and running a dynamic block every time you execute your procedure, this is not recommended except in the rare instances when there is no better way around it.
Oracle8i introduces native dynamic SQL, an alternative to DBMS_SQL. Using native dynamic SQL, you can place dynamic SQL statements directly into PL/SQL blocks.
In most situations, native dynamic SQL can replace DBMS_SQL. Native dynamic SQL is easier to use and performs better than DBMS_SQL.
Example of Dynamic SQL<------------------------
BEGIN
-- bulk fetch the list of office locations
SELECT location BULK COLLECT INTO loc_array
FROM offices;
-- for each location, give a raise to employees with the given 'job'
FOR i IN loc_array.first..loc_array.last LOOP
dml_str := 'UPDATE emp_' | | loc_array(i)
| | ' SET sal = sal * (1+(:raise_percent/100))'
| | ' WHERE job = :job_title';
EXECUTE IMMEDIATE dml_str USING raise_percent, job;
END LOOP;
Look for more information on dynamic SQL by
searching for DBMS_SQL.
Good Luck
null -
Oracle 10G Performance Tuning Documents
Hi all
Can any one tell where can I get the oracle 10G Performance Tuning materials(PDF),Documents.
Thanks in advancehttp://download-uk.oracle.com/docs/cd/B19306_01/server.102/b14211.pdf
-
Hi,
We are using Oracle Express 6.3.2 & Oracle 8.1.5 for a decision
support system that we are building.
We are using a server with 2GB RAM, Dual CPU( 550MHz) along with
more than 100 GB HDD.The OS being used is Windows NT enterprise
edition server 4.0.On which both Oracle Relational database and
Express express server resides.
The building of multi-dimensional database(MDDB) doesn't take
time, we are using around 9 dimensions with total data in MDDB
being in the range of 5- 10 Lakhs records.The building time is
about 3 minutes.
However, the running of an application(XPJ) based on the MDDB
takes lot of time. The report takes about 2 Minutes to appear
and situtation is even worse in what-if analysis kind of
reports!.The MDDB is being accessed from Windows NT Client
workstation 4.0 with 128 MB RAM.
We have read Oracle Express Performance Tuning and Database
Design Guide which is available from both Oracle Technology
Network (technet.oracle.com) and Metalink, however it doesn't
seem to be of much help.The sparsity etc. have been taken care
of by us.
CAN ANYBODY PROVIDE ASSISTANCE ON THIS ISSUE OR IS IT NORMAL
THING TO EXPECT FROM ORACLE EXPRESS????...., .Request someone to
respond to this issue, as it very urgent & bothersome to us.
ThanksDear Friend,
I am trying to configure reports for express data and in this regard I have done the following :
1.Machine A having reports 9i Developer suite Release 2 and Pluggable Express Connection Editor installed on Windows 2000 professional.
2.Machine B having which Express Server 6.3.4 installed on Windows 2000 professional.
3.In Machine A created user OESDBA with Administrator privilige.
In the reports I am able to connect to the Express Server and attach the database and choose the measure.Once all the other steps are done I am able to print the data on the reports.
But if I use a parameter form in which I give the HOST,USER and PASSWORD then substitute the same in the express_server parameter value in the AFTER PARAMETER FORM trigger block then the report gives an error saying that
Express query contains incorrect,missing or damaged information.
How to solve this error ?
Can you please help.
Regds,
Ramakrishnan
[email protected] -
Oracle apps performance tunning
Hey
could u any one tell me where i will get the Oracle Apps Performance
Tunning on Module wise like OM/AP/AR.
how to verify that my oracle Apps[11.5.10] instance running fine on linux based !!
i need check list Oracle Apps Performance !!best would be to start with http://docs.oracle.com/cd/E11882_01/server.112/e16638/toc.htm
Regards
Karan -
Oracle 10g performance is slow
Dear Exports
how we can imporve the Oracle 10g performance........we are upgrading from Oracle 8 to Oracle 10g. Windows platform. and using Oracle developer 6 as front end .
thanks in advanceDo you have statistics gathered on the tables in the 8i database? Can you post the explain plan for the query in both databases?
Since you know what SQL is having poor performance you can use TKPROF and SQL TRACE to see where your query is spending its time.
Try the following:
alter session set timed_statistics=true;
alter session set max_dump_file_size=unlimited;
alter session set tracefile_identifier='BAD_SQL';
alter session set events '10046 trace name context forever, level 12';
<insert sql with poor response time>
disconnect
Use the TKPROF utility on the file found in USER_DUMP_DEST that contains the string BAD_SQL.
For information on how to interrupt the TKPROF output, see the following link.
http://download-east.oracle.com/docs/cd/B19306_01/server.102/b14211/sqltrace.htm -
How to monitor SERVER performance
hello,
how to monitor the performance of the server,whether it is hit ratio is high or not
can u please send the query
regardsDatabase Performance Monitoring is a little more complicate than running a query to get ratio. And different version of Oracle has different tools and utilities to help you.
Strongly recommend you read Oracle® Database Performance Tuning Guide -
Crystal Developer crash, connecting with "Oracle Server" (Unicode DB): "buffer overrun"
Post Author: mdde
CA Forum: Data Connectivity and SQL
Hello everyone!
We're experiencing the following error, connecting with "Crystal Developer XI" or "Crystal Developer 2008" to a unicode-configured Oracle 10g Database using "Oracle Server"-driver and choosing a table:
ERROR MESSAGE: Buffer overrun detected!
Program: ...essObjects Enterprise 12.0\win32_x86\crw32.exe
A buffer overrun has been detected which has corrupted the program's internal state.The Program cannot safely continue execution and must now be terminated.
This error is reproducable by 100%. We're normally using it because of performance, but now we have do work with "Microsoft OLEDB" instead.
STEP BY STEP: Open CR Developer, File, New Report, New Connection (to unicode instance - this is a must), "Oracle Server", (Login without checking "OS auth."), Double click on table, Error.
DETAILS: "Oracle Database 10g Release 10.2.0.1.0 - Production", Oracle-client used by CR Developer: "11.1.0". OS: Windows XP, SP2
We tried CR SP4 and SP for 2008 edition. Didn't help. Nothing on google, nor in this forum, nor in BO knowledge base.
Does anyone also experience this error / know anything about? Thanks in advance!
David
P.S.: Picture:Hi Aaron,
I highly recommend that you only have ONE oracle client installed. Oracle tends to have problems finding the correct dependency versions of their runtime. One way is to change your PATH statement so it only has the one Client listed but I highly suggest you uninstall both and then install just the 11G client.
Also, when CR connects to Oracle we query the DB and load all tables, SP's Views, etc. and when the local resources are filled up we stop. So when connecting be sure to check off the options not being used and add filtering in that UI also for your schema as well.
One last change, close the designer down and then go into regedit and this location:
HKEY_CURRENT_USER\Software\SAP BusinessObjects\Suite XI 4.0\Crystal Reports\DatabaseOptions - DoAutoSmartLinking and set the default value to No if it's not already.
Thanks
Don -
What is the best platform for Oracle Server?
What is the best platform for Oracle Server?
What is your criterion?
Thanks,
Somsak B.I believe that Oracle is developed and coded on a Solaris
platform and then ported to other operating systems. If that is
the case, you can assume that Solaris is the best platform or at
least the safest in terms of performance and risk.
If anyone knows anything to the contrary, pls correct me.
Maybe you are looking for
-
Generate report using CURSORS? - Simple question
Folks, I'm a real newbie still with PL/SQL so please excuse my ignorance I have a simple report I need to generate. The following SQL statement gets me all my "header" records: SELECT OHA.ORDER_NUMBER, HEADER_ID, ATT11, ATT12, ATT16 FROM XXXWD.WD_DUP
-
as I can download apps from app world and that can be used in the work area? I have a BES server if it's any reference. I have applications in personal area I can not see in the work area.
-
Computer monitor affected by studio monitors?
Recently I have been trying to record some audio and I had to change the midi settings. Now whenever I turn on my studio monitors (m-audio av40) the computer monitor glitches or turns off for a moment. I never had this behavior before when turning on
-
Error when including xmlrpc header file in proC program
Hi, a simple "hello world" .pc program fails to compile on Mac 10.4.6 (Tiger) when xmlrpc.h is included, with the following error: System default option values taken from: /Users/Oracle/10g/orahome/precomp/admin/pcscfg.cfg Syntax error at line 4, col
-
Can't get rid of Facebook notifications from somone elses account
I upgraded my imac27 to 10.8.4. My wife logged into her facebook account through Safari on my iMac and now I cannot get rid of her facebook notifications from showing up in notification center. She is not logged in to facebook on my computer and I ha