Production Deployment Topologies ...

Hi all,
I've been reading up on Coherence deployment topologies and would like some assistance on the best practice for deployment into a Production environment.
In production deployments, what are the recommended deployment best practices:
I understand that:
* Coherence Cache Clusters should be configured to run on a separate Multicast/Unicast addresses to avoid impact with any other applications.
However:
* Should Coherence Cache Servers deployed into their own JVM separate from the Applications that use them?
* Or, Should the Coherence Cache servers be configured to use the same JVM as the applications?
* In a multi-Container environment, where many containers hosting many different applications is possible, what is the best deployment topology?
* If Coherence Cache Servers and Applications are separated into different JVMs, how should they be configured to communicate with eachother (eg. Extend TCP Proxy??)
Any help would be appreciated.

Hello,
I suggest taking a look at this document (especially towards the bottom):
http://coherence.oracle.com/display/COH34UG/Best+Practices
In general we do recommend separate JVMs to host cache servers. As you mention, you have the option of having cache client JVMs either join the cluster or connect to a proxy using Extend TCP. Here are the pros and cons of each approach:
Cluster Membership
Pros: less network "hops" per operation, highest performance
Cons: for best results, requires clients to be "near" servers, preferably in the same subnet/switch; poorly tuned GC on clients can affect cluster
Extend
Pros: allows for more flexible network topology since it uses TCP (i.e. clients can be in a separate network or separated from storage nodes through a firewall), poorly tuned GC will not have as adverse an effect
Cons: requires more configuration, more "hops" per operation (although affinity with a properly tuned near cache can make this moot)
Thanks,
Patrick

Similar Messages

Production deployment - Execution agent , Port connection

Hello Friends,
We are new to ODI. As part of production deployment, we have 1 master rep in PROD env and 1 Work rep per env[Dev & QA]. Like to clarify if the execution port defined (Ex: 20912) be opened between the Server were ODI agent is installed and repository DB Server? (OR) is it also required to open port 20912 to be accessed from were ODI client is installed as well?? in case we plan to use another production agent install as scheduler agent to test large volumn data loads.
Here's our setup:
ODI client install -> Win2003 Server
ODI Agent -> Unix - Solaris 10
Master Rep -> SQL 2005 - Win2003 Server
Thank you.

It should be open where you have scheduled the agent (in your case Unix Solaris 10)

Cloud service restored to the last production deployment

I've update production deployment yesterday morning then I've made changes to service files using remote connection
add and update files and everything was OK.
today morning all the changes I've done after deployment was undone and customers use the old version and this cost us hundreds of thousand of pounds
i need to know what's happen nothing appeared in operations log

Hi,
>> I've made changes to service files using remote connection
As far as I know this is the reason, we could use remote connection to do some development or troubleshoot without redeploy our application, but I don't suggest you use it to update application in product, because the
cloud service will recover if the role needs to be recycled for whatever reason. I suggest you redeploy the application after make some changes
Best Regards,
Jambor
We are trying to better understand customer views on social support experience, so your participation in this interview project would be greatly appreciated if you have time. Thanks for helping make community forums a great place.
Click
HERE to participate the survey.

Recommended deployment topologies

Hello,
I've read the "Recommended deployment topologies" for Oracle AS 10g Release 2. There's mentioned that HTTPS-to-HTTP protocol conversion should preferably be handled by a load balancer. Is there a document describing HOW TO configure Oracle AS to let the load balancer handle HTTPS-to-HTTP conversion?
Kind regards,
Philip

Found the info at:
http://download-uk.oracle.com/docs/cd/B14099_04/portal.1012/b14037/cg_secur.htm#i1036333

Recommendation of NMS product for topology discovery?

I have a LAN/WAN environment where majority of devices are Catalyst 3750 and 4500 switches, and 2800 and 2900 routers. There are some ASA firewalls, UC servers, WLC, and Dell EqualLogic storage arrays as well. Future plan may include Nexus switches as core data center switches (mostly on LAN side, not SAN side).
Currently I'm using SolarWinds to monitor link utilization. We are trying to find a management product that can do the following, hopefully for all the devices mentioned. What are your recommendations? Thanks in advance.
- automatic discovery of network topology;
- link utilization monitoring;
- device utilization monitoring;
- device inventory;
- configuration backup;
- software upgrade.

Wow Afroz that's a link to an 8-year old thread!
I've used SolarWinds' various products and Cisco's products (Prime LMS, Prime Infrastructure and many others) quite a bit. I have to say each has its strengths. I've not used the entry level SolarWinds products like IP Monitor or Cattools much at all.
The strength of SolarWinds is generally speaking, it's a bit easier to setup and use. It is also multi-vendor from the start. Cisco's multi-vendor support is recent and much more limited. On the other hand, Cisco is the most advanced as far as leveraging the Cisco technology solutions like NBAR, Medianet, etc. (Although SolarWinds NTA does a really good job at helping you deploy IP SLA).
SolarWinds NCM is both more advanced and easy to use (in my opinion) than the equivalent functions in either Cisco Prime Infrastructure or LMS.
Either vendor has free trial versions - check them out and make the best decision for your needs.

Best practice - Production deployment

what is the best practice to deploy custom built application in production?
Where to deploy in Admnistration server or in manged server?
Thanks,
Shankar

Hi Shankar ,
It is always better to deploy your application in the Managed Server. You do not need to restart your admin server while you need to restart your managed server for changing your environment related to your application. Admin Server is meant for administering all the managed server instances. You can have many different managed servers for different specific applications. So it is a best practise to deploy your application in the managed server.
Regards,
Ilaya

Java webservice production deployment

I have installed JWSDP1.0.01 and created a webservice 'VisitantWS' using the same. I can deploy, access and consume web service alright in the JWSDP deployment environment.
Next, I am trying to create a production level environment for which, I have installed IIS and Tomcat4.0 on seperate machine with IIS redirector in place.
For deployment of my webservice in the Tomcat, I am placing the .war file(from \dist directory of JWSDP) in to the \webApps and restart IIS, Tomcat. It extracts the .war file and creates folder hirerchy in \work directory automatically. Now, when I try to check my deployment using
http://localhost:8080/VisitantWS-jaxrpc/VisitantWS
it says 'the requested resource (/VisitantWS-jaxrpc/VisitantWS) not found.
I have changed my server.xml and added the context for VisitantWS to it.
when I check the status of the service using
http:/localhost:8080/manager/list
it says
'/VisitantWS-jaxrpc:stopped:0
/running:0'
trying to reload the service gives me
'FAIL - Encountered exception java.lang.IllegalStateException: Container StandardContext[VisitantWS-jaxrpc] has not been started'
How can I deploy the webservice created using JWSDP for the production use where I dont have to use Ant (which is the part of JWSDP) for the deployment or ship JWSDP alongwith my product. Am I missing something for the deployment of the webservice on Tomcat?
Hemal

It was almost a year back and I have been distracted to other stuff but as I think of it now, having tomcat and IIS on different machines must be the problem as the webserver need the webcontainer to be on the same machine to redirect the requests. not sure but I plan to get back to it soon.

BO 4.0 production deployment recommendations

Hi All,
We would like to understand what would be the good approach to segregate the different services into different physical machine.
When we do the sizing using quicksizer, the output is in SAPS and Memory size for four different tiers.
a) Web App Server
b) Inteligence Tier
c) Intelligence DB Tier
d) Processing Tier
Coming to the questions,
1) The processing tier is expected to consume lots of memory so it is recommended to be put on it's own physical server.
2) Combining the processing server with Intelligence tier - Will this impact the performance of the normal activities like navigating and scheduling in case of processing server getting overloaded?
3) Merging Web App Server and Intelligence services - Will this have any impact in performance of the overall user experience?
4) Do we need to have a separate server for the repository database? Can it not be co-hosted with the other Oracle production databases. What are the things that we need to consider to co-host the database?
5) Usage of VM servers for all these tiers - Any issues in using virtual servers to host all these different services?
In short, we need to be able to bring down the number of servers without compromising the performance and usability. We are looking at deploying BO 4.0 on 100 concurrent user license with all the client tools such as Crystal, WebI, Analysis OLAP, Dashboard and Explorer.
6) Am thinking of proposing one virtual server for (web apps + intelligence) and another for processing server. And recommend to co-host the database repository. The processing server to have 32 GB RAM on 8 or 16 core machine. The web + intelligence services to be hosted on a 8 core 16 GB RAM machine.
7) The processing server - is it good to split them to two different physical server so that we can host different client related services on the different physical server?
I know that I am asking too many questions but I do not find deployment options and recommendations anywhere.
Thanks & Regards,
Mohanraj

Have you looked at the Business Intelligence Platform Planning Guide and Admin guide ?
They have information that will allow you to answer your questions.
This sort of questions fall under what SAP Services do and is almost impossible to answer without intimate knowledge of your requirements, business neeeds and resources available....
In general :
1) The processing tier is expected to consume lots of memory so it is recommended to be put on it's own physical server.
Yes
2) Combining the processing server with Intelligence tier - Will this impact the performance of the normal activities like navigating and scheduling in case of processing server getting overloaded?
It might, depends on resources available to it.
3) Merging Web App Server and Intelligence services - Will this have any impact in performance of the overall user experience?
Usualy yes, again - depends on resources available, the bottleneck will be in disk and network I/O
4) Do we need to have a separate server for the repository database? Can it not be co-hosted with the other Oracle production databases. What are the things that we need to consider to co-host the database?
Recemended way is to have repository DB separare, iside your production Oracle DB's.
5) Usage of VM servers for all these tiers - Any issues in using virtual servers to host all these different services?
Possible, if you fully understand limitations and benefits of VM environments.
6) Am thinking of proposing one virtual server for (web apps + intelligence) and another for processing server. And recommend to co-host the database repository. The processing server to have 32 GB RAM on 8 or 16 core machine. The web + intelligence services to be hosted on a 8 core 16 GB RAM machine.
co-hosting repo DB is bad idea, the rest migth work.
better to have web tier, intelligence and processing separate...
7) The processing server - is it good to split them to two different physical server so that we can host different client related services on the different physical server?
Yes, usualy it is a good idea...

BO Production deployment of universes based on SAP BW query

Hi there,
We just set a scenario on a BO development machine, and would like to deploy all these created objects (webi/cr documents, universes based on SAP BW queries... ) to a BO Production machine.
How can this procedure be performed?
As far as I know, import wizard should be used, but what happens with universes that are referring to SAP BW queries? Obviously, universe in BO Production environment should point to the same BW query name, but from SAP BW Production system, not development.
Is import wizard capable of doing this step? If not, what should be done in order to replicate your universes based on BW queries from BO Development to BO Production?
Thanks and best regards,
David.

Hi David,
which version of BOBJ are you using? The LifeCycle Manager (LCM) may be interesting for you. Runs only on Windows and only for BOBJ XI 3.1 though.
Installation guide:
[http://help.sap.com/businessobject/product_guides/boexir31/en/xi31_LCM_Install_en.pdf]
User's guide
[http://help.sap.com/businessobject/product_guides/boexir31/en/xi31_LCM_User_en.pdf]
Regards,
Stratos

Questions for the wiseman before production deployment

Hi, I have a number of questions I would like to ask before deploying the webelements into a production environment. who do I contact to answer my questions and concerns? Thanks.

hey Michele, here's the answers to your questions..
1. not sure what you mean by this but pass through html is a supported method and feature in xi & xir2. webelements is a function library that uses this method. rather than having everyone create their own code each time for pass through html components, we've institutionalized the code in a custom function library.
2. there is support via Global Services should you need help with a project / implimentation. contact your sales rep should you wish to go the Global Services route. there is no support via Customer Assurance as this is considered custom code...i.e. same as if you were using the sdk to do some custom work. there is this forum and hopefully it will ramp up so that more people use it.
3. not at this point. most of the samples though have instructions on them, plus the Users Guide is pretty big and there's also the video of David Taylor's presentation at http://diamond.businessobjects.com/node/1575
4. this tool does not add the ability to write to the database...that ability has been around since v9 of crystal. this tool adds the ability to create an easy interface to do this which was lacking with regular prompt methods.
5. pass through html must be activated on the crystal reports web server...see the user guide for more info.
6. these controls simply append textÂ to a report viewer...so you are going to see no more performance lag on a report than you would if you put a bunch of text formulae on a report.
7. see 2. it would be the same if you're building a custom app...global services support, forums, but no customer assurance support.
8. one could really write their own tool to do this as pass through html is the method that makes this possible. clients have been writing their own controls since v9 of crystal. webelements is simply a larger library wherein the controls all act in conjunction with opendocument syntax and the prompting method in crystal.
9. this comes from business objects.

OWB 10.2 Production deployment article and Licensing caveats!

Hi All,
If you haven't already read Mark Rittman's blog on OWB in a production environment, it is well worth reading:
http://www.rittman.net/2007/02/14/running-owb10gr2-in-a-production-environment/#comments
In particular, read up on the licensing. If you are doing anything non-trivial, chances are that you need the "Enterprise Edition" which came as a shock to me.
As I interpret the licensing (see the license link in the article):
If you are using more than just the "Default_Configuration" you need Enterprise edition.
If you deploy your warehouse to a different instance than where your meta-data resides, you need the Enterprise edition.
If you deploy BI objects to Discoverer from OWB, you need Enterprise edition.
If you have more than one configuration, you need Enterprise edition.
In addition the type 2 and 3 SCDs, Data Quality etc are add-ons. By the time you add all of this up, Informatica isn't looking all that expensive anymore!
Be careful!
- Mike

Hi Mike,
You make excellent points here.
IMHO, The enterprise option is a complete rip off. There is really no functionality provided that you cannot do yourself with OMB scripting and a little ingenuity.
This is really just another example of Oracle trying to suck money out of us any way they can. To make matters worse, most of these features never work as advertised.
Beware Oracle, that little software company in Redmond Washington is catching up - fast! And I don't believe the term "add ons" is in their vocabulary...

Pre production deployment checklist

Hi all.
Iam in the process of adding, couple of channels to the running production environment.
Is there anything that i should consider technically before deploying my new channels on the production environment?
I have two challenges :
1) To add channels so that it fits into the exsisting layout exactly. For this, what are the things that i should consider before importing the channels? Should i concentrate on the Tab provider being used currently?
2) how would i make sure, these new channels must appear for all exsisting ldap users.
I appreciate any feedback that get me started confidently.
Thanks in advance
Veera

I would add to the list:
- run reconciliation report beetween AM accounting and GL accounting.
- set Company Code as "in production" for FI
- set Company Code as "in production" for assets
- create variants for the programas that creates DME files (tipically F110...)
- Documents, vendor, assets, customers, range numbers
- check that all the customer and vendor banks are already created in the system
- modify bank descriptions if needed
regards

Staging - Production Deployment

I won't be able to work with the Contribute software for a
day or two, but I'm trying to understand something about it now.
I'm planning on using the two server, staging - deployment
approach. Does Contribute push content from the staging server to
the production server, or only to the staging server? Based on what
I've read in the documentation, I can't tell. Some things make me
think only staging, while others suggest both.

Hi,
According to your post, my understanding is that you wanted to know the best practices to deploy SharePoint Portal in different SharePoint environment.
If the site is not existing in the production server, we can save the site from the development server, and then import it to the production server.
But if the site is already existing in the production server, we should follow these steps to just add the taxonomy and content types to the production server:
Save the site from Dev as a template
Import the template as solution in Visual Studio
Remove unnecessary items from the solution(Please pay more attention on it. If a content type/list... in the solution is existing in the production site too, it will replace the
same object existing in the production after deployment)
Package the solution
Deploy the solution in the production
For more detailed, please see:
http://ahmedmadany.wordpress.com/2012/12/30/importing-sharepoint-solution-package-wsp-into-visual-studio-2010/
There is a similar thread for your reference.
http://social.technet.microsoft.com/Forums/en-US/7dcf61a8-1af2-4f83-a04c-ff6c439e8268/best-practices-guide-for-deploying-sharepoint-2010-from-dev-to-test-to-production?forum=sharepointgeneralprevious
Thanks & Regards,
Jason
Jason Guo
TechNet Community Support

Production Deployment

Hello,
I'm new to stored procedures and packages, and looking for a some direction. My application is constantly changing and enhancing, and I see that when I write my procedures, i explicitly declare what the inputs and outputs are. Considering a very flexible environment, these inputs may change frequently. I know I just need to deploy the new package script to production and we're all good.
The problem is when there are several application servers calling the same stored procedure, the application code won't match what the stored procedure is expecting, until that code is deployed. So the question is, how do I deploy my stored procedures to a production environment gracefully? and without having to take the whole cluster down :)
thanks for any info..
-chris

I would be rather concerned about any application that was changing so rapidly that the API between different layers was changing frequently. That doesn't make sense and implies that you have a general problem with design and architecture. In the vast majority of cases, if an API needs to evolve, you would do so by adding optional parameters to an existing stored procedure or creating a new overload of the stored procedure that accepted a different parameter list. That would allow both the old code and the new code to continue operating.
That said edition based redefinition in 11.2 does give you the ability to have multiple editions of code in place at once and to allow different servers to point at different editions. But this strikes me as overkill for someone that is relatively new at writing stored procedures.
Justin

Production deployment - Stage(or)nostage

hi,
I am using weblogic.Deployer to deploy in our production, with 32 managed servers on 5 boxes , sharing NFS weblogic tree.
I would like to know, how do i deploy the EAR's in stage mode (or) nostage mode ? and also are there any pros & cons of using -nostage in production ??
Thanks
Venkat

hi,
I am using weblogic.Deployer to deploy in our production, with 32 managed servers on 5 boxes , sharing NFS weblogic tree.
I would like to know, how do i deploy the EAR's in stage mode (or) nostage mode ? and also are there any pros & cons of using -nostage in production ??
Thanks
Venkat

** Production Deployment Topologies ... **

Similar Messages

Maybe you are looking for

Production Deployment Topologies ...