TM1 Health: How to Maintain Your TM1 Server
What do you need to check to keep your TM1 server healthy? Well, we sometimes get asked to review client’s TM1 servers to make sure they are running as well as possible. Here is a list for maintaining the health of your TM1 servers. We will add to as we find new things that should be checked regularly.
Memory usage, along with hard disk space, are the things that need to be monitored most frequently in TM1.
Logon to your server and simply using Task Manager, check how much memory is being used by TM1 on your server. Try to keep it no more than 70 or so percent utilised.
Note that over time, TM1 grabs more memory and does not release it back to the Operating System, so it is probably a good idea to schedule a restart of the physical server on at least a weekly basis.
Hard Disk Space
Once again, back on your server, check the amount of free space available on the drive containing your TM1 models. TM1 really does not like running out of hard disk, so don’t let it get beyond about 80 percent full.
If the data drive does get too full, the first thing we go look at is the model backups folder and the logs folder. Then, if need be, adjust our TI processes that delete old backups or clean out old log files.
If you are able to, you can help have great TM1 server health by setting up alerts for when your hard disk is running out of space, or your memory exhausted o your CPU’s blow out.
Version (or Scenario) Maintenance
Data bloat is one of the things that can, over time, lead to poor TM1 health. Do you really need the budget for 2016 still, or version 5 of the forecast from 2018? If not, clean up your Version dimension and delete the data that is no longer needed.
Next, do you need actuals from 2008 in your model? If not, go to your Year or Time dimension and remove the old elements to get rid of all that history.
For both of these though, make sure you have a persistent backup stored and labelled of the data, in case you need to restore it!
TM1 Log Review
There is no better place to understand what is going on behind the scenes than to review the TM1 logs. We recommend storing the logs in a special Logs folder, alongside the Config, Data and Backups folders. Then on a frequent basis, open the logs and just scan through looking for failures or errors.
Error Email Notifications
Although not strictly a regular monitoring activity, you should set up an automatic email of errors from Turbo Integrator processes. That way you can be notified of problems before your users tell you (and you can avoid the egg all over your face).
Monitor the Internal TM1 Control Cubes
TM1 has a series of internal cubes that can be used for monitoring and optimisation of TM1. These include:
- }StatsForServer and
These cubes need to be enabled, but once turned on give valuable information about the usage of your model. They can greatly assist you keeping TM1 healthy.
TM1 Model Restart Times
It’s only a small thing, but keep a note of the amount of time it takes to restart your models. If they are getting longer and you can’t explain why, jump in and try and understand why and what you can do to improve model efficiency. Remember to that you can have persistent feeders if the restart takes too long.
Regular Process Duration Times
Keep a tab on how long the regular processes take. Then if they are getting longer, try and understand why. It might be completely valid. But then again you might have a problem that can be corrected.
Rule Save Time
When you save Rules and you have changed a feeder, the target cube for the feeder (so either the cube you are changing the feeder for or a secondary cube if you are feeding another cube) will be updated. This exercise takes time. Just keep an eye on how long the rule save takes and if it blows out, then take a real close look at the feeders.
It should go without saying that your TM1 models are backed up to a local zip file (let’s not rely on corporate IT!). It’s no use having a health TM1 model if it then crashes and you struggle to get the models back. Or if you need to restore a point in time, say from last financial year end or when you budget was approved. If you’re not backing them, make it happen using the scripts on this post.
Keep TM1 Up To Date
Planning Analytics now has a number of components that are either updated by IBM approximately on a quarterly basis or monthly. You need to keep these versions up to date. The components you need to check are:
- PA Planning Analytics Server
- PAW Planning Analytics Workspace
- PASS Planning Analytics Spreadsheet Services
- PAX/PAfE Planning Analytics for Excel
Ask Your Customers!
Shock, horror. Yeah, go ask your customers for their feedback about performance, about usefulness of reports and forms, about accuracy, model heath and relevance. You might just be surprised about how perceptions change – and it would be better for you to be on the front foot about how your users (customers) feel about the amazing TM1 models you manage for them.
Systemise your Checking
With all of these things, it is really easy to become complacent about the health of your TM1 server. So create yourself a checklist, assign a frequency to it (say weekly or monthly) and then make it part of your routine.
Then record your findings and if you’re noticing stuff happening frequently, then dive deeper into it and try to solve whatever the underlying problem is.