Monthly Poll 201508: What Mode Of Security Do You Use?

[Alan Kirk]

We're starting a new "thing" from this month; a monthly voting / discussion poll that will run for a month. It will be stickied for the number of days in the month, and after that will drop off to become a normal thread though the poll itself will continue to run indefinitely for the benefit of new / occasional users. (This one will have some overlap with the September one because it's starting so late in the month, but normally there will be only one at a time.) If anyone has a suggestion for a future poll feel free to post it, although bear in mind that we obviously have only 12 slots per year.

This one stems from a presentation that I'll be giving later this week on using the APIs. I'll be demonstrating using mode 1 authentication (which we do still use primarily because we have so many legacy networks which don't always play well together), but it set me to wondering how realistic that is these days. (Especially as a new implementation we're developing, targeting users on one specific network, is being developed using CAM authentication which I'm not wild about as I find it very slow and cumbersome.)

You can nominate multiple modes if you use different modes on different servers, but because there's no way of telling which votes have selected one option and which have selected many there's a "Multiple Modes" checkbox as well; but please check that in addition to all of the modes that you use.

And of course, any discussion on why you have elected specific modes over other modes can only add to the flavour.

In this poll, votes can be changed if your mode changes.

[rfielden]

As a TM1 only shop, we are finding our upgrade to 10.2.2 using CAM very painful and inefficient compared to our 9.5 instance.

[Alan Kirk]

As a TM1 only shop, we are finding our upgrade to 10.2.2 using CAM very painful and inefficient compared to our 9.5 instance.

I know what you're saying. We have one instance (9.5.2) on mode 1 and a newer (10.2.2) instance on mode 5. I'm really not loving the mode 5 experience. I've noted that the number of people using it has rocketed ahead as the survey has gone on; whether that's down to people being forced into it by using Cognos Bi as well or whether it's a choice isn't clear. But integrated login is definitely not showing up as flavour of the month.

[Steve Vincent]

Its forced on us - we are now trying to move to Cognos BI for reporting, away from the traditional Excel route. To avoid a crazy amount of security admin we need single sign on so CAM was the only choice in reality.

[Alan Kirk]

Its forced on us - we are now trying to move to Cognos BI for reporting, away from the traditional Excel route. To avoid a crazy amount of security admin we need single sign on so CAM was the only choice in reality.

I saw one of our former members at the Conference who, unfortunately, has had too much of the busies to be on for quite a while now. He has moved to Cognos Bi as well and described it, a couple of times, as "the worst decision we've ever made".

While I haven't had the pleasure myself, when I read this the thought "you poor b'strd" flashed through my mind.

[George Regateiro]

For any of the people that have switch have you seen any additional locking in TM1Top with CAM as compared to other modes? We are in the process of switching (Mode 2 to 5) and our first tests went poorly with RunTI and CAM. We were attempting to kick off multiple processes in parallel and because CAM attempts to update some of the security groups and properties this was causing all the parallel processes to lock. They are supposed to be releasing a fix for this in an upcoming release of RunTI but I am getting a little wary that I dont have a good way to test high volume concurrency with CAM (I really need to learn a new load testing tool).

[lotsaram]

For any of the people that have switch have you seen any additional locking in TM1Top with CAM as compared to other modes? We are in the process of switching (Mode 2 to 5) and our first tests went poorly with RunTI and CAM. We were attempting to kick off multiple processes in parallel and because CAM attempts to update some of the security groups and properties this was causing all the parallel processes to lock. They are supposed to be releasing a fix for this in an upcoming release of RunTI but I am getting a little wary that I dont have a good way to test high volume concurrency with CAM (I really need to learn a new load testing tool).

Yeah, this is a real pain when running lots of TI in parallel. The solution is to put in a 1-2 second wait between kicking off each thread. Allowing each thread some breathing room around establishing the login seems to solve the locking issue.

[George Regateiro]

Yeah, this is a real pain when running lots of TI in parallel. The solution is to put in a 1-2 second wait between kicking off each thread. Allowing each thread some breathing room around establishing the login seems to solve the locking issue.

For a work around we wrote a thread manager to be able to multi thread calls using REST. This allows use to authenticate once and then reuse sessions getting around the issue. For normal processing the 1 to 2 second wait would be fine. We had a use case that kicked off a massive export of data from TM1 to feed an SPSS model. So the 2 second wait was not feasible.