As specified in the DataMiner System Requirements documentation, the recommended number of users within a DMS are:
- Online Users: 100
- Configured Users: 1000
The basic questions we have are:
- How flexible are the above numbers? Is there any buffer around these?
- When a large active directory (or any other domain system) is synchronized with DataMiner, what puts a limitation to how many users are configured?
- What are the symptoms expected once more than 100 users are online or more than 1000 are configured? Does DataMiner trigger an event?
- Is there any room for enhancing the number of online and configured users at the core level? Would it be possible to take these numbers to another level of scalability (i.e. Configured: 5000 | Online: 200)
Thank you in advance,
Hi Rene,
Those recommendations are quite conservative, and there are no hard limits. They are not enforced and you can go above them, but the system might become slower on certain actions.
Our dcpdma.skyline.be has currently about 4500 users configured on it, and that still works fine (going to the users/groups page takes a few seconds longer although), but that’s also a smaller system. If you would e.g. combine a huge amount of views with a huge amount of elements and a huge amount of users, it could potentially take longer to e.g. create a new view in the system.
To answer your questions:
(1) The numbers are flexible, and there is some buffer, but you have to be careful when exceeding them.
(2) A large active directory is not immediately the problem, it is the number of users you configure in DataMiner which is relevant (by directly adding a user, or via a domain group).
(3) DataMiner will not trigger an event, but might slow down on certain actions. The impact of online users will largely depend on the actions they do and the activity of the DMS. You could probably easily connect with 1000 users on a small DMS not generating any alarms. Larger systems with a lot of alarm activity might have bigger impact of many users being online. I would assume, mainly CPU usage of SLNet.
(4) Potentially the numbers you mention will work without any problem, it all depends on the size of the system and the activity of the users and the DMS. Only then we know what we are talking about…
As you can see, there are many variables involved here. I hope I helped at least a little bit…
Bert
Hi Rene,
I can’t really fully judge on all aspects of your question, so other people can maybe provide further insights on this matter.
But the nbr of Online Users is very likely to be quite flexible. In fact, the actual limits are more likely to be defined by the effective degree of load induced by those users, rather than the absolute number of users online in the system.
A user who is simply connected to DataMiner, and using it to monitor the operation, will hardly cause any load. On the other hand, a user who continuously performs for example large queries, e.g. loading alarm history, will cause more load.
Therefore, I would say that you do not have to be too fixated on that nbr, unless all users would be users that continuously cause a heavy load on the system.
Aside from that, I’m not aware of any important underlying constraints that would prevent a high number of people being connected simultaneously.