Clustered architecture

Hi,
          We have 2 Solaris boxes running WL 8.1 SP5 with a clustered license. We need some loadbalancing technology for our cluster.
          1. If we decide to go with HttpClusterServlet, would it be advisable to deploy it on the AdminServer of the domain or should we rather have a dedicated instance of WL running for HttpClusterServlet?
          2. If our dedicated WL instance also resides on one of the 2 available boxes, do we need another license (we have a clustered license for the 2 boxes already)?
          3. What approach should we use to avoid single point of failure for HttpClusterServlet? We only have 2 boxes for our cluster and proxy.
          Thanks in advance!

Personally I prefer hardware load balancers. I would put 2 of them in front of the WLS cluster for redundancy.
-- Rob
WLS Blog http://dev2dev.bea.com/blog/rwoollen/

Similar Messages

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..

MultiProcessor Clustering Architecture Question

Comments inlines.
          Cheers - Wei
          Jim Zhou <[email protected]> wrote in message
          news:[email protected]...
          > Wei,
          > When servlet calls ejb, if that ejb is deployed on four wls servers,
          > because the round-robin, only 25% of the chance the servlet and ejb are in
          > same JVM. Am I right on this?
          If your servlet and ejb are deployed on different clusters, the answer is
          yes.
          If they are on the same clusters, WLAS will call the ejb on the same local
          machine, if the call fails, that's the time load-balancing algorithms plays
          into the picture.
          Hope it helps.
          > What do you mean on "WebLogic rjvm will try to
          > call local remote object as much as possible."? How do achieve load
          balancing
          > if the servlet always get the local ejb to serve its request?
          > I have a client going to use 50 low-end Sun box as JSP/Servlet
          cluster,
          > 4~6 high-end Sun box as EJB cluster. Just curious how much performance
          penalty
          > I will get. Thanks.
          >
          > Jim Zhou
          > BEA Professional Service.
          >
          > Wei Guan wrote:
          >
          > > If I were you, I would like to create one cluster with four weblogic
          > > servers. Every server will serve both servlet and ejb. The benefits:
          > > 1) High availabilty and utlilization.
          > > 2) RMI calls between servlet and ejb will be optimized. WebLogic rjvm
          will
          > > try to call local remote object as much as possible. Save
          > > serizliation/deserilization overhead and communication cost.
          > > 3) Better load balancing.
          > >
          > > My 2 cents
          > > --
          > > Cheers - Wei
          > > bahar <[email protected]> wrote in message news:[email protected]...
          > > > Hello,
          > > >
          > > > I have a client who needs to cluster two Sun E420R (4 procesor)
          > > > machines.
          > > >
          > > > I have presented the following architecture for a cluster of two
          > > > machines running 5.1.0 and JDK 1.2.2:
          > > >
          > > > Three Apache WebServers are load balanced via a director. On the
          > > > Application Servers, I have created two clusters (on each machine) for
          > > > the servlets and EJBs. So, basically, each machine has two clusters
          > > > (web and ejb) with two JVMs. So, there are two members of the web
          > > > cluster and two members of the EJB cluster. The apache plugin
          contains
          > > > a comma delimited list of the IP addresses of the machines in the web
          > > > cluster.
          > > >
          > > > I have added a Multi-IP URL in the JNDI ProviderURL for the calls in
          the
          > > > Web Cluster to point to the list of servers in the EJB cluster
          > > > (comma-delimited list).
          > > >
          > > > The Http cluster uses In-Memory replication for HttpSessions.
          > > >
          > > > Is this a valid architecture that will support failover and
          clusterable
          > > > service load balancing? If not, what suggestions should I make?
          > > >
          > > > Thanks in advance.
          > > >
          >

JNDI lookup in Multitired clustered architecture

Hi alll,
We are in the design phase of a multi tired clustered application.
This application has jsp/servelet cluster and business application cluster. We are confused abt the JNDI lookup logic need to use in servlet cluster.
Can anyone help me to find out an optimal solution for jndi which gives the complete advantage of load balancing and esp. fail over
of clustering
Thanks in advance
gokul

Hi,
          First of all, I would like to clear the point that load balancing can be done within a cluster not among the cluster.
          So if you have more than one cluster the load balancing
          would be at each cluster level.
          For JNDI lookup you have to provide cluster address i.e combination of Managed Server Address e.g:
          t3://localhost:7001,localhost:8001,localhost:9001...etc
          Thanks,
          Qumar

Customer would like to have details on Clustering resources usage

IHAC that currently implemented an application using 6 WLS instances in 6 domains (!) and 2 phisical servers. I mean they have 6 admin with a single managed server each. They did so to implement an High Available system, working Cluster-like.
When I told them that a better configuration should be 1 Admin and 6 managed servers using the standard Clustering WLS features they claimed that it will be resources consuming, more than the current configuration.
Is there any document where we describe the best way to implement Clustering and a benchmerk where we show the differences in system usage when using cluster and not using it, and also the Pro/Cons of having a clustered architecture?
I cannot find a the info in the standard WLS docs.
They have a WLS 9.2.
Thank you !
Chiara
(ACS Service Manager)

The following doc tells more about why one should cluster.
http://download-llnw.oracle.com/docs/cd/E13222_01/wls/docs92/cluster/overview.html#wp1011562
Scalability
High-Availability
Application Failover
Load Balancing
If your customer doesn't need the benefits mentioned in the above doc then, there's no need for them to cluster their app.
I agree with your customer that cluster adds resource overhead. Most customers in production environment choose high availability and load balancing against additional resource consumption at runtime.
Following the performance tuning best practices and understanding customers application, one can decrease the overhead of clustering to some extent.

SQL Server 2000 "Yukon" architecture

Hi,
could you tell me whether Microsoft have achieved cluster technology with "Yukon" or not?
As announced in article: Database Architecture: Federated vs. Clustered-An Oracle White Paper March2002, "...According to Michael Otey, Senior Technical Editor, SQL Server Magazine:�Technically, this setup isn�t a true clustered implementation. According to Microsoft, SQL Server won�t support true clustering until the post-SQL Server 2000 release, code named Yukon.�(Yukon is not expected to be shipped until well into 2003.) - But when I browse msdn, there is nothing on the architecture itself.
I am doing my Master's thesis on 9iRAC, and I need this information to complete my research. I found several Oracle white papers talking about federated vs. clustered architecture. As said in Response to "Reality behind Real
Application Clusters Marketing Messages"- An Oracle White Paper March, 2003"..when in truth Microsoft does not have a cluster database." Is this still the truth? Can somebody from ORACLE provide me with tips/links to complete this topic?
I am also Oracle DBA, Oracle customer, registered on Metalink.
Thank you for the support.

BOL 2014: "
Data Types (Transact-SQL)
SQL Server 2014
Data Type Categories
Exact Numerics
bigint
numeric
bit
smallint
decimal
smallmoney
int
tinyint
money
Approximate Numerics
float
real
Date and Time
date
datetimeoffset
datetime2
smalldatetime
datetime
time
Character Strings
char
varchar
text
Unicode Character Strings
nchar
nvarchar
ntext
Binary Strings
binary
varbinary
image
Other Data Types
cursor
timestamp
hierarchyid
uniqueidentifier
sql_variant
xml
table
Spatial Types "
LINK: http://msdn.microsoft.com/en-us/library/ms187752.aspx
Kalman Toth Database & OLAP Architect
SQL Server 2014 Database Design
New Book / Kindle: Beginner Database Design & SQL Programming Using Microsoft SQL Server 2014

One SAP system in two MSCS clusters for R/3 Enterprise

Is the one SAP system in two MSCS clusters architecture supported on R/3 4.7 ( WAS 6.20), with SAP CI on one MSCS cluster and MSSQL on another pair of database MSCS cluster ?

Hi,
I don't think so. The supported architecture is SAP CI on one node and SQL Server on the other node.
Regards,
Olivier

Muti-tier clustering

Can the sample Avitek medical applcations
          bundled in wls8.1 be deployed into multi-tier
          clustering architecture? I.e. JSP/Servlet
          in one cluster and EJB objects in another
          cluster?
          If so, has anyone done it with medrec example? Or anyother examples bundled
          in wls8.1?
          THanks.
          -kl


Can the sample Avitek medical applcations
          bundled in wls8.1 be deployed into multi-tier
          clustering architecture? I.e. JSP/Servlet
          in one cluster and EJB objects in another
          cluster?
          If so, has anyone done it with medrec example? Or anyother examples bundled
          in wls8.1?
          THanks.
          -kl

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

BODS and Oracle RAC

Do we know of any problems with BODS and Oracle Real Application Clusters architecture? For example, a RAC TNS entry would specify one service name but multiple hosts for load balancing and failover. In BODS the connections are specified in many places (web console, datastores, job server). As far as I remember in some of those places we do have to enter the host name. Would this disqualify BODS from benefiting from RAC or would it (even worse) cause problems? Any official note by SAP on BODS and Oracle RAC?
Thanks.

Hi Mahir,
As far as I know it's only the web console that requires the hostname, however I am using DS XI 3.1 and when selecting database type Oracle in the web console there is a checkbox for 'Oracle RAC'. If you check this box then you do not require to enter the hostname.

Any Oracle best practice/standards for inter-DataCente links for Oracle RAC

Hello Oracle Experts,
Am working for a customer to set up Oracle RAC architecture hosting SAP/Non-SAP applications per SLA levels(MC/BC/Standard) specs. Currently my network team needs calculation to arrive at whether we will go for a (1), (2) or (3) 10Gig links for inter DC (Data-Center) for Oracle RAC.. below is additional background:
•     Porting all client SAP/Non-SAP Oracle databases to new 2 data-centers.
•     There will be 10 blades (4x BL680s and 6x BL460s) in each DC (can scale-up/out later on).
•     Clusters architecture to support Extended/Stretched RAC cluster feature
•     Clusters 2-node each(1-datacenter1, 1-datacenter2) and nodes distributed across 2 x c7000 such that no cluster has more than one node in an enclosure.
•     Each node will have - 4 NIC ports ( 2 x public and 2 x private) , 2 dual-port HBA
•     Oracle ASM/ACFS (ASM Cluster File System), Voting Disk, OCR and Database files
•     the versions are Oracle 11g RAC, Oracle 10g RAC and Oracle 9i (for DataGuard/Standby) on RHEL 6 on Proliant Blades (x86) + BladeMatrix
My network colleagues considering using DWDM across the 2 DCs(given the lesser cost?). Am still looking around if there are any Oracle/industry-best practices around this and having a calculation to support that..
Many Thanks in advance..
Regards,
Abhijit

Hi ,
There are no specific set of steps / practices for batch loading contents to ucm . It would be very much dependent on how many contents does the user have to load to UCM and how well the server is configured in terms of performance .
You can get more details from the following documentation link : http://docs.oracle.com/cd/E21043_01/doc.1111/e10792/c02_settings009.htm
Thanks,
Srinath

6.2 64bit windows end of life

What is the end of life (mainstream maintenance) for 6.2 64bit windows?

Hi Jeff,
If you check the more information tab in the PAM matrix for WebAS 6.20, you will find the following information.
+Installing the Java Add-In for SAP Web AS 6.40 in SAP Web AS 6.20
Releases 6.30 and 6.40 introduced a lot of radical changes to the Java part of the SAP Web Application Server. For example, the clustering architecture was completely overhauled and remodeled using the concepts that had proved successful on the ABAP side. With SAP Web AS Java 6.30, the supported J2EE standards switched to the popular version J2EE 1.3; at the same time, the former Java development environment was replaced by the SAP NetWeaver Developer Studio, which is integrated with the open-source platform Eclipse.
SAP Web AS Java 6.20 will not be maintained beyond 2006
Another important consideration was Suns decision to only maintain the Java Development Kit (JDK) 1.3 until the end of 2006. SAP Web AS Java 6.20 is invariably tied to this JDK, and therefore SAP will not be able to maintain SAP Web AS Java 6.20 beyond this point. Since the support for SAP Web AS 6.20 must - and will - continue until 2009 according to SAPs general maintenance strategy, we needed to find a replacement for the Java part of this version of SAP Web AS.
We recommend using SAP Web AS Java 6.40 with SAP Web AS ABAP 6.20
For of all these reasons, we want to enable customers to upgrade from SAP Web AS Java 6.20 to SAP Web AS Java 6.40 as soon as possible. Therefore, we have decided to support combined installations of SAP Web AS ABAP 6.20 and SAP Web AS Java 6.40 as of January 2006. Consequently, we provide a new set of installation CDs/DVDs for customers who want to insert the J2EE Add-In into existing ABAP-only installations SAP Web AS 6.20: These CDs/DVDs are now based on SAP Web AS Java 6.40. You can download them from the Software Distribution Center at:
􀂃 http://service.sap.com/swdc 􀃆 Download 􀃆 Installations and Upgrades 􀃆 Entry by Application Group 􀃆 SAP NetWeaver 􀃆 SAP NetWeaver Components ( < SAP NW 04 ) 􀃆 SAP WEB AS
The installation requires that you have done the following:
􀂃 Installed the downward-compatible kernel for SAP Web AS 6.40 as described in SAP Note 664679,
􀂃 Installed all hot packages up to number 41, and
􀂃 Upgraded your database to a recent version (the minimum version is 7.5.0.18 for MaxDB, 9.2.0.3 for Oracle, and V8 Fixpak 7b for DB6).
The procedure for installing the J2EE Add-in is described in the "Installation Guide - SAP Web AS Java 6.40 SR1. You will find the version for your database and operating system at:
􀂃 http://service.sap.com/instguidesnw04 􀃆 Installation 􀃆 SAP Web AS SAP Web AS 6.40 SR1 and Related Documentation
For official information about this change, as well as the latest instructions on how to proceed with the installation, you can also refer to SAP Note 914486.+

X509 certificates, hostname verification and SunCluster 3.1 failover.

Hi,
A newbie question - having an existing non clustered architecture and trying to decide how to use the SunCluster features.
I have some self signed x509 certificates that are used by a process. When this process is (going to be) failed over to another machine, and the filsystem that contains the certificates also follows, what is the recommened way of ensuring that I can use the same certificates and that hostname verification etc still works.
When I define a resource group for the filesystem and network interfaces required by this, can I also create a virtual hostname that will work on either of my cluster machines and will not confuse my SSL code when it verifies the certificates and the host?
I think this is not a question of DNS, but a question of what happens when I want to type 'hostname' and would like to get the same result on either box that is part of our cluster. This way my certificates and application configuration would not need to be changed during a failover event.
Thanks!

Forget about the local hostname question - all that is important at the moment is that my keystores and truststores (created using Sun JVM keytool) are transportable and usuable on the other host without change. The network resources associated with the names in the certificates are planned to move across as part of the resource gorup).
In theory I guess this shoud work, but I wanted to know if anyone has had any experience of doing this and there were any gotchas.
Thanks.

SAP EP Sizing Queries

All,
We have designed a clustered architecture of two SAP J2ee Engines ( one central instance
and a dialog instance ) with no fail over option . I used quick sizer tool for initial sizing .
My queries are
1) In the quick sizer , I got 2800 SAPS for Enterprise Portal . Does it mean that the J2ee Engine ( Central instance ) has to be powered
with 2800 SAPS ?
2) What about the other J2ee engine ( dialog instance ). Is it have to be the same 2800 or 2800/2 ?
3) In your experience , do you give equal power ( SAPS ) to both dialog and central instance ?
4) There are other components in the portal like ESS/MSS , LSO .So do we need to calculate the SAPS for each of these and add to
central instance and Dialog instance ?
5) The portal Database in a separate box , Quick sizer doesn't give any idea how to size this .Any clue about this ?
Aneez

Hi,
If 2800 is the total SAPS required, then one would distribute the SAPS between database host, central instance and application host. there is no rule of thumb that CI should have more SAPS than DI, it depends on how you architect the design. ALso hardware vendors inputs would be helpful. Each server would be sized for optimum SAPS>. If you have not considered EsS/MSS or LSO during the initial sizing then ou need to reconsider it or look for additional hardware for the aextra SAPS.
Between the DB and CI, one could distribute the SAPS 40/60.
Also for each CPU, you should consider 2 - 4 GB of RAM, dependending on what hardware you choose.
Hope this helps,
Chetan

Installing HTTPServer in Diff m/c from the OC4J in a Oracle9iAS Rel2 Cluster

Hi
I want make Oracle9iAS Rel2 Cluster.
Some of oracle9iAS Documentation(Clustering architecture pictures) showing these 2 are installed in different machines.
Can I install OHS in one machine and OC4J Processes in other machine.
If possible how do I do that.
Thanks in Advance
Madhusudhana Rao. P

Hi
I want make Oracle9iAS Rel2 Cluster.
Some of oracle9iAS Documentation(Clustering architecture pictures) showing these 2 are installed in different machines.
Can I install OHS in one machine and OC4J Processes in other machine.
If possible how do I do that.
Thanks in Advance
Madhusudhana Rao. P

Clustered architecture

Similar Messages

Maybe you are looking for