If dashboarding were the only factor, this would be a lot easier.
Had you noticed the lack of a BOMgr function for Deski reports in XI? (I think misstated the busobj.exe/CPU ratio earlier - should be 16 per CPU.) Still, if every Deski session takes up a process, thatâs quite a load. Weâre starting to negotiate licenses; they are suggesting CPU based instead of Named User and Iâm leary of this. Converting to Webi and CPU licensing might be a windfall, though.
You forgot the fact that DeskI is merely a temporary solution that BOBJ came up with. There is no long term plan for DeskI and they will try to phase it out in subsequent releases (Please mark the word âtryâ. Whether they can succeed is another story). But if I were you, I will not be comfortable to move to something that is controversal and might be obsolete in the next 5 yearsâŠwell, unless you have well thought-out plan to move out of DeskI in the next couple years.
IrishStan, thx for the recommendations. Do you know how to check on the items you noted:
[list]Number of CPUâs 4
Type of CPU xeon P4 1.5 Ghz Minimum
Memory 4GB Min
Disk Size 4GB min free space + Storage
File System Type NTFS
Swap Space default
Disk Type SCSI
Network Client to Middle Tier 100 Mbps
Network CArd Type Ethernet
Network Card Number 1[/list]
We are in process of installing BO XI r2 in our company and would like to create a cluster for failover. The cluster type we want to go for is exactly what you have mentioned here⊠2CMS and 2 Web APP servers on two different boxes.
Can you please guide as to how do we start? I mean, do we do âNewâ install on both the boxes and then configure them? Or âNewâ install on Server 1 and âExpandâ install on Server 2?
I guess we would have to do two âNewâ install since expand install does not install app server. What files will we need to configure post installation? I would appreciate your help.
This is what you do with them both pointing to the same repository using the same CMS name. You are simply having 2 servers running the same process (XIr2 server) on 2 different machines.
Consider too installing everything on each of your machines but not enabling each server process. This way should you have processes fail on the other machines, you can quickly enable the servers/processes as needed.
I tried this method of installation. But the problem I am having is when I do expand install on the second server, it does not install the application server (Tomcat). So only the server1 has Tomcat and if that goes down, BO goes down.
Will these XI servers enable themselves if Server1 goes down? How will the failover work? Will I need to configure web.xml file to mention the name of the cluster?
You donât need to use BOBJâs Tomcat install, you can install it using a regular Tomcat d/l for installation.
Enabling each service to compensate for a lost server/machine is a manual process so once you were made aware of a machine going down you would have to manually enable the servers/processes on another machine.
You would need to install and configure each machine identically (great practice )
Then it does not make any sense for failover. I thought the process is automated, that is why you create a cluster. If any time, one of the servers go down, the other server will automatically take over the load.
Anyways, this is my scenario:
Full BO Xir2 install on Server 1 [Tomcat]
Expand install on Server2 [No app server]
It creates a cluster. I see all the servers from both BO in CMC. When I shutdown any one CMS, everything goes down. And after that I cannot bring them back up.
This is just a basic Expand install. Do I need to configure web.xml, ccmconfig.sh or any other file?
Regardless of the number of CMS servers (really services), you can only have one CMS database, so youâre OK there.
Two CMS services on one server will give you the ability to handle more requests than one service could, but wonât give you true failover, because if the server goes down youâll lose both CMS services. If possible, itâs better to split the CMS services (and others) across at least two physical servers so if one server goes down, the other server can temporarily handle all of the requests.
BO has a sizing guide that is an excellent resource:
Thanks. I know the guide but in my mind you could not start 2 CMs services on the same hardware at the same time on the same cluster and CMS Database.
Here running performance test will block around 150 cincurrent users and Iâm not in position today to change the DB from MySQL to SQL Server 2005. Then Iâm more inspecting the CMS services response time to close this point.
A new cluster isnât mandatory, I added a second CMS service to a single-server deployment without any issues. I added the CMS service, changed the port to 6500 (and unchecked the âuse default portâ option), and set the CMS service to log in using a network account.
The service started fine and I continued with some configuration:
Verify the new CMS has been added to the CMSClusterMembers entry in the registry (it should be appended with a â:6500â for the port number of the new CMS)
If youâre running Performance Management, youâll want to update the InitConfig.Properties file
I deleted the LocalHost file under \Tomcat\Work\Catalina
Restarted Tomcat
I stopped the original CMS and was able to login and perform basic functions without any issues, so this tells me the 2nd CMS was configured correctly.
In the BO sizing document I came across this statement
âRun 1 Web Intelligence Server service for every available CPU 1 Web Intelligence Server service having 1 availableCPU can support 25 Â 40 Maximum Simultaneous Connectionsâ
So does this mean if I have 160 active concurrent users then I need to have 4 WebI services running in a Quad processor.
I am really confused. Can anyone help me in understanding the statement, as to me this means with every increase of 40 active concurrent users one processor needs to be increased?
Hi All,
Another thing I need to ask is that for the webi Service also it is saying that memory requirements depend on complexity of reports and if reports need to be refreshed. In our case 54% of the reports are very complex and all of them are refreshed. So what GB of RAM should we acquire? The sizing document does not give any clear idea on this Memory part. Can anyone help on this too?
I have to disagree with this ⊠one of the most powerful features that our users have lost when using Webi instead of DI is being able to load a spreadsheet from their desktop and link this data with existing universes without any reliance upon IT. In webi there is no such facility and all data sources have to come via pre-developed Universes.