Vertical Scaling

[petehase]

We're looking at vertical scaling as an option to improving performance of our system. We've contacted OT and drawn a blank really - although they push the concept in technical briefings, its not something they seem to want to divulge any specific information about. And as we're just looking at the concept, we don't fancy paying for an OT consultant to come in for a couple of days, to set up something that could be done ourselves in 2 minutes, which we might not even use anyway.

Has anyone had any practical experience with vertical scaling, and could provide me with any information regarding the do's and don'ts - or is it just a case of stacking up a number of instances on a machine, load balancing across them, and seeing how hard we can push before it all falls over??

 

[DrDDT]

What Livelink version are you using?

< 9.5 has issues with services names (service name has to be the same as the name in IIS, like '/livelink/'.
With vertical scaling using older versions, you cannot use the same name.

I also had weird problems when using two different Livelink versions on the same machine.

 

[Syntergy]

You might check out Syntergy's latest newsletter which has a section on Architecture strategy.

http://www.syntergy.com/general/newsletter/livelink_newsletter.html

See the article "Does Your Organization Have the Right Livelink® Architecture Strategy?" It's has a very slight sales job but all the key issues associated with Livelink architecture are outlined and specific recommendations are made where appropriate. You are probably not getting the answer you want from OT because there really isn't a good one. There are a lot of variables involved and spending two minutes to set up another instance probably won't solve your problems.

One of our favorite stories is a customer where one user logging into Livelink took 2 minutes for the Enterprise workspace to display and another user logging into Livelink on the exact same machine took 5 seconds for the Enterprise workspace to display. It was all because of the group structure and how permissions were defined. They could have added 5 new instances and still had the same problem.