Hello. I’ve had a request from the finance department to set up another instance of BCA because they have a large volume of reports which need to be ran every end of month. I’m not too seasoned with the setup or installation of BCA.
Does it involve establishing a cluster, making the current instance the primary and my BCA server the backup? I really hesitate to alter the main(current) server’s configuration. Is there another simple way it can be accomplished without setting up a full clustered environment? If that’s the case I’ll have to have BO talk me through setting up the cluster cause that sounds pretty major, at least to someone like me
I’m not sure I quite follow what you mean by another instance, but if I’m following your question correctly, it sounds like you want to share your BCA environement on your WebI server.
This can work, but you need to be careful that you do not allow too many BCA processes to run concurrently or they’ll take away CPU from your WebI users. We started out this way, and it worked fine, but as the user base on WebI continued to grow, we found we had to purchase additional servers and licenses to separate our BCA environment.
Having separate environments for WebI vs BCA is the ideal infrastructure. Most users schedule documents on the BCA that run longer than a standard WebI refresh and do not comprimise your WebI environment during core hours.
Actually we are not using WebI at all here (unsophisticated user base). The finance group has a few reports which run on through BCA that take almost 2 hours to refresh (poor warehouseing if you ask me). If too much other activity is going on the reports will time out.
The current request is that a secondary BCA be set up so that they can run aditional reports concurrently with the others at end of month. I’m trying to avoid having to set up a whole clustered environment to do so.
Be careful of those finance folks! They have a nasty habit of trying to tell you the solution, rather than describing the requirement. I can take a shot at them, because I am in finance, and have to support them as well!
Their requirement is to be able to run more reports during a certain time window. The solution may be in the data warehouse as you suggest, because NO report should take two hours to run. I’m guessing an investment in database tuning, instead of more BCA power, would yield a much better return.
There’s multiple options to offering a solution. I’ll try to give you this without making this into a novel. Our environment currently runs 3 BCA servers on a windows environment (cluster mgr, and 2 nodes). We have 18 BCA schedulers between the 3 servers. We run anywhere from 225-250 jobs daily. The key is to balancing the workload. If you have users wanting to schedule long running jobs, you can do that on a single server with another BCA scheduler, but cut back on the number of concurrent jobs running.
I believe when you set up a new scheduler, the concurrency is 5. Drop this back to 1 or 2. This way only 1 or 2 long running jobs can run at a time, and not interfere with the other jobs on the box, because a job can only take 100% of one cpu, leaving the other cpus on the box to run the other jobs.
The other option is to spread out the work. Suggested option would be to run the long running jobs outside of core hours. Your database staff will appreciate this as well as long as loads are not occuring during their run time.
Setting up a BCA node is not a big deal, and if the box is solely for BCA purposes, it’s pretty stable as well. Weekly reboots are recommended and seem to work out ok.
Our warehouse is a complete slop job. I’d love to rewrite the entire thing but I just don’t have time ATM and they rely on these reports to operate. I agree so wholeheartedly that no report should take two hours to run, but most of the report isn’t data retrival but rather calculations within the report (yes our warehouse has no aggregrates ). I wish you people could see the thing. It’s like an LD kindergardener trying to write his name with his feet. That describes our warehouse in a nutshell.
That being said I recently was given a budget to buy a new test machine for testing the new stuff and now they want to use it as a backup/alternate BCA (which really ticks me off they want to use my testbox for production).
So in line with what you are suggesting, dchrista, I can either install another instance of BCA on the actual BO server ( which is a 4cpu machine), or I can convert the current BO server into a cluster Manager and turn my testbox into a BCA node. Is that a good road to start traveling?
That’s funny…but in all seriousness, we’ve all been there. Personally, I’d let the DBA’s battle the run time issue. Sounds like you’ve got your own environment to maintain. I’m sure it’s only a matter of a few months before you get contacted about a strain on your warehouse, but at least you’ll have some backup to help support the argument of “TUNING THE DATABASE FOR OPTIMAL PERFORMANCE.”
Anyway, I’ll get off my soapbox on that one. Back to the BCA servers. If you’ve got the licensing to have another BCA production box, I’d say spend the money and go for it to set this up as a node. If you can add another 4 cpu’s to your BCA environment, you’re only helping your situation down the road if you don’t see database tuning in the near future. You can always find some old server sitting around to use as a test environment (desktops even work well for testing).
Even if you get to add another box, I’d still recommend setting up another scheduler in your BCA environment. We both know that 2 hour run time is unacceptable, so you want to put tight controls on that and only run a few of those jobs through at a time…that is, unless you want to make an immediate point to your DBA’s.
well BO to us pretty much is for finance only. As far as my role I’ve inherited the new role of DBA, lead developer on our ETL processes, Business Objects Admin and Support, and I’m pretty sure I’ll betaking over the universe design as well. This is on top of my normal role supporting and development for our SFA tool and also our corporate intranet content management site. Not to mention this is my first IT job (i’m a 9 month old professional). So when you say things like “tell the DBA to…” I giggle on the inside. I’m going to be making myself worth a ton of money fast.
Wanna see progress? Trace my first few posts here. I’ve gone from report writing to setting up clusters, with no mentor except you guys here at BoB. I can’t count how many times Andreas made me look like a genius. I should order BoB a pizza with all the trimmings.
I’ll have to look into setting up the test box as a node because it’s the quick and easy way to appease the end-users, but I sure would love to develop a different version of the report that takes under 2 minutes to refresh. It’s only a P&L report and we don’t have a product that sells in high volume.
My condolences…would not want to be in your shoes, but wish you the best of luck. Figure out the biggest pain point, and take it from there. The rest can come in phases!
It looks like I’m late on chiming in on this, but I can’t resist a challenge…
What I gather is that Finance has a number of long-running reports that run infrequently and you’d like to move that processing out of your “production” BCA into one of its own. Let me know if I’m missing the point.
If you really want to segment this, you’ll need to create a secondary BCA in a new repository (each repository can have only one BCA). Once done, create a new cluster (a single server might be fine or you can add nodes).
Set up BCA on the new cluster to use the secondary BCA you created in the new repository.
Benefits: 1) complete seqmentation from your other BCA. 2) because its segmented, you can set up the BO Manager to allow for longer run times than 2 hours (see customization documents). You can set up the new BCA for big ugly finance jobs and they’ll stay out of the way of your regular production BCA.
Thanks for the advice, but rather than putting a bucket on the floor I’ve managed to convince them to let me fix the roof, aka, optimise the 2 hour reports. That’s the real problem.
Doing all that clustering and multiple repositories and etc… seems like more set-up, configuration, and administering than a developer like me desires to take on.
I’m a Bo novice…and would really like some help with this problem.
We currently have an instance of BCA running on our server. From here we schedule/run the reports. One of our department wants their one BCa to run the reports. The problem I’m facing is that this is a Citrix enviroment. We have already installed Designer, Supervisor and BO on the citrix, but we haven’t installed BCA.
Does anyone have some tips/tricks that I should keep in mind?
But most importantly; How should I approach/do this install?
Are you needing to install the BCA console software to the CITRIX environment? If that’s the case, all you need to do is go through a normal setup and enter your license key for the BCA console, AND the BO Services Administrator. If you only install the BCA console, you will not get the appropriate components needed to add your PORT and IP address that would need to be called when the BCA console is started. As for your existing software already in CITRIX, it won’t touch it, and that software will function as it did before.
If I install the BCA on the citrix, will they be recognised as two different BCA’s (with the one I have now)? And will they not interfere with each other? And how do I let the BO Admin. console know that there is an other BCA ?
Are the Citrix users on a separate repository? If not, they will get the same BCA, unless you assign a new one to them with Supervisor to their folder structure. You will also need to define this BCA to the Admin Console. Basically, the BCA’s are managed at a folder level, so if users are sharing infomation from the same folders, they will also share the BCA.
Being from a Citrix shop, you would not install BCA on a Citrix server. Citrix servers are for client tools such as full client or the BCA Console, not the BCA product. That gets installed to a dedicated W2K or Unix server.
Server products are installed on servers. Citrix servers service client sessions, not Windows server products. Take a look at the PAR for hardware requirements for BO products.
(BOB member since 2002-08-19)
(BOB member since 2002-09-19)
). I wish you people could see the thing. It’s like an LD kindergardener trying to write his name with his feet. That describes our warehouse in a nutshell.