Best OAS R3 Architecture

We have 80,000 to 100,000 users, where we expect 30% to 40% will be using the application concurrently.
What is the best proposed clustering architecture (active/active vs. active/passive) Oracle recommends for OAS J2EE Web container, Please provide some benchmarks and cases that are similar to our case.
Thanks.

Check the High Availability Guide:
http://download.oracle.com/docs/cd/B25221_04/core.1013/b15977/toc.htm
--olaf

Similar Messages

EFP and Intranet Portal - Best Practices for architecture

Hi All,
   We are planning to create a portal for our partners (b2b scenario). This portal will provide anonymous and user specific access. We also have a intranet portal with ESS and MSS. I have some doubts around implementing EFP and Intranet portal.
   1) What is the best practice solution architecture around portal instances? Is it suggested to implement both intranet and internal portals on same portal instance? Do you implement them as 2 seperate instances (and 2 seperate boxes) one for intranet and another for internet?
   2) There are couple of functions shared between inranet and internet portals. Has any one attempted   a FPN connection between a internet portal and intranet portal?
   3) What is the best practices around an external portal connecting to ECC directly?
Any suggestions are greatly appreciated.

Hi Pallayya,
         We are implementing a similar kind of thing for our client.I can explain that so you find some glimpse out of it.
We have decided twp ways.
1) We are having one instance of the portal that will be accessed both in Intranet & Internet.We are using two different URL for intranet & internet.We have a reverse proxy in the picture also,coz it will increase the security of the portal in the Internet.The reverse proxy will be available in the DMZ Zone.We have integrated ECC & BW with the portal and some web dynpro application(For intranet only).
Now the user comes from the Internet will hit the reverse proxy first->then it will go to EP Server->in EP it will be decided which kind of request client is asking ->
If it is a ECC request->Portal Server sends back the request to Reverse Proxy->RP will send it to web dispatcher->Then it will go to ECC.
If it is a BW request ->Portal server sends the request to BW Server.
2) There will be two different URLS for the portal in Intranet & Internet
     For that we used two reverse proxy
regards
Indranil

Best (Thrifty) physical architecture for medium-size environment?

What is the absolute best physical architecture you would come up with for this medium size environment?

Thrifty? The SharePoint CALs alone will run you at least $1.5 million unless you already own these or have a good EA. You said "basic collaboration" so I'm assuming you don't also need Enterprise CALs though note some of the features you listed
could require enterprise licensing depending on how you're using them.
The information you haven't provided isn't enough to make an appropriate recommendation. While you have 10,000 users it's not clear how many concurrent users you expect, or what sort of workloads these users will have.
Are the users all in one location? Spread out geographically? What types of network connections? Why types of networking devices (such as a load balancer for your second option)
Is this virtual or physical? What are the hardware specs of the servers?
How many documents do you expect to have? How large will they be? What sort of storage is backing this farm? How many site collections, sites, lists, libraries, items do you expect? Workflows, third party solutions, customizations, etc. The list goes on.
Neither of these options provide high availability. Is this a concern for you? I ask because in my experience a SharePoint farm used by over 10,000 people needs to be highly available. If you need HA then you'll need more servers.
Honestly this forum isn't the best place to get the information you're looking for (i.e. an architectural design). Instead I recommend asking specific questions about specific problems you have with your design.
My question(s) then for you are: 1. Why are these the two designs you have? (in another way: What are the decisions and requirements you have that led you to these options) and 2. What is the challenge you're having in picking one of them?
Jason Warren
@jaspnwarren
jasonwarren.ca
habaneroconsulting.com/Insights

What is the best laptop for architectural design?

I mostly need if for revit. I currently have a lenovo thinkpad T61. It has been great but is getting old, and I think it is about to crap out. I would like to stay with lenovo but im up for any suggestions.

Hello,
It might be best to look at the requirements for the architectural design software that you want to run, and then look at which systems match those specifications.
Regards,
Aryeh Goretsky
I am a volunteer and neither a Lenovo nor a Microsoft employee. • Dexter is a good dog • Dexter je dobrý pes
S230u (3347-4HU) • X220 (4286-CTO) • W510 (4318-CTO) • W530 (2441-4R3) • X100e (3508-CTO) • X120e (0596-CTO) • T61p (6459-CTO) • T43p (2678-H7U) • T42 (2378-R4U) • T23 (2648-LU7)
Deutsche Community Comunidad en Español Русскоязычное Сообщество

What is the best way of architecturing SWCV in PI

Hi,
I have 4 different ways of architecturing my interfaces in SWCV,
Approach 1: Single Software Component version (SWCV) for all Interfaces
Approach 2: Single SWCV per interface scenario
Approach 3: Separate SWCVs for sender and receiver
Approach 4: Separate SWCVs for sender, receiver and mapping
Can any body tell me from their designing experience which is best recommended and used with examples.
Also provide some advantages of it in terms of version management,maintainance,reusability and transport.
Thnx
Chirag

Chirag,
if you think about the logics - the design shoud meet the company architecture. It is no problem to design all interfaces in 1 SWCV or do separate SWCV for each scenario if u have 10 interfaces. But think about this, if you'll have 500. So I don't think that approach 1 and 2 are useful.
So go for 3 or 4. I for example always use 3. Yes, the truth is, that mapping is done by XI, but why to create SWCV for only data types and then SWCV for mapping? This was not the way how SAP engineers thought about it I think. And where to put the mapping in approach 3? I place for example mapping always in the target system's SWCV. Approach 4 is also possible and used, but as I said - why to have mappings and other objects separatly? But this dependes only on your decision.
Peter
p.s. If you want to have every interaface separatly, you can create them in different namespaces
Edited by: Peter Jarunek on Feb 12, 2008 10:55 AM

Best practices, software architecture question

I'd like to offer a challenge to everyone in good software architectural design principles. Here is a mocked up version of some code that I've been working with, grossly over-simplified for this example. Take a look at the following classes and see if you can determine what the parentComponent will be for the JOptionPane that is displayed in the HardDriveController's constructor:
public abstract class Controller {
    private JFrame parentFrame = null;
    public void setParentFrame(JFrame parent) {
        this.parentFrame = parent;
    public JFrame getParentFrame() {
        return parentFrame;
public class MyGUI extends JFrame {
    private CPUController cpuController = null;
    public MyGUI() {
        setSize(300, 300);
        setVisible(true);
        cpuController = new CPUController();
        cpuController.setParentFrame(this);
    public static void main(String [] args) {
        new MyGUI();
public class CPUController extends Controller {
    private RAMController ramController = null;
    public CPUController() {
        ramController = new RAMController();
public class RAMController extends Controller {
    private BusController busController = null;
    public RAMController() {
        busController = new BusController();
public class BusController extends Controller {
    private HardDriveController hardDriveController = null;
    public BusController() {
        hardDriveController = new HardDriveController();
public class HardDriveController extends Controller {
    public HardDriveController() {
        JOptionPane.showMessageDialog(
                getParentFrame(), "Hey! We're inside the HardDriveController!");
}Did you figure it out? If you said the parentComponent will be null, then you are correct. The logic behind the code had intended to set the parentComponent to refer to MyGUI, but of course, the code does not behave as expected. Now this seems simple enough, but the legacy code I'm working with is full of many examples like this. Notice how MyGUI creates an instance of CPUController which creates an instance of RAMController, which creates an instance of BusController, etc...In some cases, I see a "Controller" class that is 8 or 9 "levels" deep with respect to one controller creating another, creating another, etc.
Let's say that I want the JOptionPane in the above example to have its parentComponent refer to the instance of MyGUI. The following code is one means of passing the reference down to the HardDriveController, while still using the getParentFrame() method in HardDriveController:
public class MyGUI extends JFrame {
    private CPUController cpuController = null;
    public MyGUI() {
        setSize(300, 300);
        setVisible(true);
        cpuController = new CPUController(this);
    public static void main(String [] args) {
        new MyGUI();
public class CPUController extends Controller {
    private RAMController ramController = null;
    public CPUController(JFrame frame) {
        setParentFrame(frame);
        ramController = new RAMController(frame);
public class RAMController extends Controller {
    private BusController busController = null;
    public RAMController(JFrame frame) {
        setParentFrame(frame);
        busController = new BusController(frame);
public class BusController extends Controller {
    private HardDriveController hardDriveController = null;
    public BusController(JFrame frame) {
        setParentFrame(frame);
        hardDriveController = new HardDriveController(frame);
public class HardDriveController extends Controller {
    public HardDriveController(JFrame frame) {
        setParentFrame(frame);
        JOptionPane.showMessageDialog(
                getParentFrame(), "Hey! We're inside the HardDriveController!");
}Note that the reason I'm calling the Controller class's setParentFrame() method in each extended class's constructor is that I want to ensure that any call to getParentFrame() in these classes will return a reference to MyGUI, and not null.
After this long description, I want to ask all you Java gurus the following question: What is the best means of offering the JOptionPane a reference to MyGUI? My solution passes a JFrame reference all the way down from the CPUController to the HardDriveController. Is this an "acceptable" solution, or is there a more elegant means of getting a reference to MyGUI to the JOptionPane (for example, modifying the MyGUI class to follow the singleton pattern)? Thanks for your input.

Proflux, One thing I failed to mention is that this is a very old codebase and the likelihood of the object graph being reworked is slim to none. I agree with you 100% that the dependency doesn't feel right. But I have to live with it. I mentioned the idea of using a singleton pattern. Here's a mock up of how that would look:
public class FrameRetriever {
    private static FrameRetriever instance = null;
    private HashTable frameHash = new HashTable();
    private FrameRetriever() {
    public static getInstance() {
        if(instance == null) {
            instance = new FrameRetriever();
        return instance;
    public void addFrame(JFrame frame)
        frameHash.put(frame.getTitle(), frame);
    public JFrame getFrame(String frameTitle)
        return (JFrame) frameHash.get(frameTitle);
public class MyGUI extends JFrame {
    private CPUController cpuController = null;
    public MyGUI() {
        setTitle("Test GUI");
        setSize(300, 300);
        setVisible(true);
        FrameRetriever.getInstance().addFrame(this);
        cpuController = new CPUController();
    public static void main(String [] args) {
        new MyGUI();
public class HardDriveController extends Controller {
    public HardDriveController() {
        JFrame parentFrame = FrameRetriever.getInstance().getFrame("Test GUI");
        JOptionPane.showMessageDialog(
                parentFrame, "Hey! We're inside the HardDriveController!");
}My only concern with this solution is that it couples the HardDriveController to MyGUI because we need to know the frame's title in order to get the frame instance. But at the same time, I won't have to add constructor arguments to any "in between" classes for the sole purpose of passing a reference down the object graph. Any thoughts?

Best Practices & Strategy Planning for SAP BI Architecture

What are best practices and strategy planning that SAP BI Architect should know?
What are the challenges are involved with this role ?
What are the other information that this Architect should know to deliver the robust BI solution?
Is there any white papers on the best practices on Architecture & Data Modeling, please ?
Thanks,
Venkat.

Hi
As per the Best Practice first load the master and next transaction data .
Please find the link for best practices
http://www.sap.com/services/pdf/BWP_SAP_Best_Practices_for_Business_Intelligence.pdf.
Regarding the architecture it depend upon the size of volumen and how much frequency your load and hard ware sizing
based on this we can provide best solution
If you any issues please let me know
Regards
Madan

Bring CRM to clients and partners, Architecture DMZ best practices

Hi, we need to bring access to CRM from internet to clients and partners.
So that we need to know the best practices to architecture design.
We have many doubts with these aspects:
- We will use SAP portal, SAP Gateways and web dispatchers with a DMZ:
       do you have examples about this kind of architecture?
- The new users will be added in 3 steps: 1000, 10000 and 50000:
       how can regulate the stress at internal system?, is it possible?
- The system can't show any problems to the clients:
       we need 24x7 system, because the clients are big clients.
- At the moment we have 1000 internal users.
thanks

I use the Panel Close? filter event and discard it and use the event to signal to my other loops/modules that my software should shut down. I normally do this either via user events or if I'm using a queued state machine (which I generally do for each of my modules) then I enqueue a 'shutdown' message where each VI will close its references (e.g. hardware/file) and stop the loop.
If it's just a simple VI, I can sometimes be lazy and use local variables to tell a simple loop to exit.
Finally, once all of the modules have finished, use the FP.Close method to close the top level VI and the application should leave memory (once everything else has finished running).
This *seems* to be the most recommended way of doing things but I'm sure others will pipe up with other suggestions!
The main thing is discarding the panel close event and using it to signal the rest of your application to shut down. You can leave your global for 'stopping' the other loops - just write a True to that inside the Panel Close? event but a better method is to use some sort of communications method (queue/event) to tell the rest of your application to shut down.
Certified LabVIEW Architect, Certified TestStand Developer
NI Days (and A&DF): 2010, 2011, 2013, 2014
NI Week: 2012, 2014
Knowledgeable in all things Giant Tetris and WebSockets

XMII and System Architecture

In the best practices documents and other xMII related literature that I have read, the proper architecture of integration for xMII into a corporate enterprise landscape is based upon local/plant xMII servers that are connected to a corporate xMII server.
This literature does not delve into the details as to why this approach is significantly better compared to, for example, a single enterprise xMII server that retrieves data from local/plant databases that have taken care of the collection of the relevant shop floor data.
I would appreciate any feedback that can explain the best practices rationale further or provide any novel approaches to xMII's architecture of integration.

The reason why SAP suggests the kind of architecture that the Best Practices Guide talks about is -
1. xMII talks about Intelligence - Real Time Dashboards for the Plant Users (Operators/Supervisors). This is a very common scenario where xMII adds value . Here it makes sense to have a xMII server @ the plant layer to make a very real time view possible for the user. So its mainly a bandwidth issue.
2. Another very good point the guide talks about is incase of absence of the Corporate network the plants can go on with there business as usual with xMII acting as a buffer. As soon as the corporate network's up the buffer can then be sent accross to the corporate (read SAP ERP). There have been many instances where customers have come up with these kind of requirements , This kind of a scenario becomes very relevant when one is upgrading their ERP.
The Best Practices have evolved based upon the kind of requirements xMII has seen over the years and hence it reflects what is best from an architecture perspective.
Hope this helps.

Best way to handle several events that may occur simultaneosly

Hi all,
I have an application vi in which I need to take care of several user generated events like menu bar selections, mouse clicks on controls (both single & double clicks) and many more.Also, all these events can occur simultaneously.Which would be the best method or architecture to handle all these events with ease?..All suggestions are welcome.
Can I use dynamic event handling for this?..If so,how can I implement it? (I went through several documents & examples for dynamic handling but none of them explained properly in this context).

mnx wrote:
[..] several user generated events like menu bar selections, mouse clicks on controls (both single & double clicks) and many more.Also, all these events can occur simultaneously.[...]
Hm, can you enlighten us how a user will create these events simultaneously?
Or maybe i am a bad user as i always use either mouse OR keyboard to control an application, most often not both the same time (starting different "events" concurrently).
Beside of this discussion, Producer/Consumer is a very good starting point to address this kind of requirement, as already stated above.
Norbert
CEO: What exactly is stopping us from doing this?
Expert: Geometry
Marketing Manager: Just ignore it.

XI Architecture Level Interview FAQ?

Hi Xperts,
I am looking new job in SAP XI, Could any one send me XI Architecture Level Interview <b>Questions</b> and <b>Answers</b>?
to -> [email protected]
thanx in advance
best wishes
Gopi

XI Architecture questions will be totaly based on the components of xi. u may refer to this doc and have it cleared:
https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/docs/library/uuid/a3d3c390-0201-0010-c490-bd85917138c8
https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/docs/library/uuid/34a1e590-0201-0010-2c82-9b6229cf4a41
https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/docs/library/uuid/40fde890-0201-0010-fa86-ee9a507bc62e
any questions on the components can be opened as a separate thread with this one closed if anwsered.

ADF Architecture Training in EMEA

Some training running in EMEA delivered by Frank Nimphius on the best practices and architecture patterns around developing ADF Applications.
https://blogs.oracle.com/grantronald/entry/adf_architecture_masterclass_in_emea
Places are limited.

Hi Bjarte,
There is a new BPA class called "Oracle BPA Suite: Model and Implement Business Processes" being rolled out right now. Unfortunately the only class planned in Europe will be run in German, but you could contact your local Oracle office and ask what the upcomming plans for Norway (or Scandinavia) are.
You can find more information on the training at Oracle University's web [http://education.oracle.com/pls/web_prod-plq-dad/show_desc.redirect?dc=D70464GC10&p_org_id=34&lang=D&source_call=]
Best,
Mathias

Clustering architecture

Hi,
I am working for one of the Bank and we are going to implement infra for SIT,UAT,DR and PROD Env and we are still in designing phase.
We have provisioning of two physical boxes,Please suggest the best suitable clustering architecture for our banking app with which can achieve HA as well as Load Balancing features of WLS cluster.
Regards,
Jaya Rathod.

you can go for a vertical clustering environment.. you can create a cluster and the members of the cluster would reside on different machines, So in case of failure of any of the machines..the request could be directed to another server running on different machine..
please respond if it looks good to you..

System Architecture Question

I'm trying to figure out the current system architecture of a potiential client.
All I know is that they're running a JSP web application on Apache Tomcat against a MySQL database.
1. If they're running Apache Tomcat does this mean they are not using EJB?
2. Is it true they could be using an application server as well as Apache Tomcat?
3. What is the best possible system architecture if running only Apache Tomcat against an MySQL database?
I know these questions are vague, but I'm trying to understand the best possible JSP design without using EJB.
4. Will there still be a layered approach?
Thank you,
Todd

1. If they're running Apache Tomcat does this mean
they are not using EJB?According to your information, it seems that they are not using EJBs. Tomcat does not handle that part of the J2EE specification.
2. Is it true they could be using an application
server as well as Apache Tomcat?A "full" J2EE application server would furnish the same functionality.
3. What is the best possible system architecture if
running only Apache Tomcat against an MySQL database?JSP, servlets, and POJC (Plain Ol' Java Classes). Possibly a Singleton or two to act as database "handlers" for connection pooling, etc.
I know these questions are vague, but I'm trying to
understand the best possible JSP design without using
EJB.Don't use JSPs for anything except display logic (View). Use Servlets for processing (Controller) and POJC for data access (Model).
4. Will there still be a layered approach?In my opinion, kind of - you're still operating in the Servlet container, which has it's own rules regarding class behavior.

Passing data to subpanel

I have an architecture question for the labview 8.2
community...
I have two indicators that I'll be passing data to and I would like to create a
single picture (png, jpg, ect) of both indicators. The key aspects is that the
data to these indicators will be changing and it's not until the end user
clicks "snap picture" on the front panel when the picture is taken.
I'm leaning towards creating a subpanel on the main vi then passing the data by
"call by reference" to a subvi. When the GUI event happens in the
event structure a "Get Image" of the subpanel is taken.
I have two questions.
1) I haven't gotten data to update properly in the subvi. The "Get
Image" always snaps a picture of the default vi and without the
values. What's the best way to pass the data in?
2) Is their a better architecture to use for this goal? I'm open to
other ideas it their is a more efficient and streamlined approach.
Thanks.

> I have two questions.
> 1) I haven't gotten data to update properly in the subvi.  The "Get
> Image" always snaps a picture of the default vi and without the
> values.  What's the best way to pass the data in?
The Set Value method should work. Also, if the VI is not looping, you can
simply put it's panel in the sub panel, and call the VI like any ordinairy
VI, using it's connector pane to pass data. That is the best, if it means
"easiest". If best means the most scalable, you'll need a more elaborated
architecture (see 2)).
> 2) Is their a better architecture  to use for this goal?  I'm
open to
> other ideas it their is a more efficient and streamlined approach.
>
Yes. There probably are better architectures. Most of the time I create
seperated user events for each seperate type of info I need to communicate.
I put the user event in a buffer, and each dynamic VI that needs that info,
registers for it's events. Often those VI's are in subpanels, but sometimes
they are VI templates, so I can start multiple instances. Sometimes there
front panels aren't visible at all, for instance DAQ VI's. I don't mean to
say this architecture is the best, but it has been great for all my projects
so far.
Any good architecture will need a lot of effort and experience. Not only
experience with LabVIEW, but also with the kind of projects it is made for.
That is why thee is no "best" or "standard" architecture. A power plant
needs another architecture then a word processing program...
Regards,
Wiebe.

Best OAS R3 Architecture

Similar Messages

Maybe you are looking for