![]() Again, it helps if you have experience in knowing how long it takes for this step to complete (if not, keep watching the memory!). This results in the operator restarting the Server again, potentially getting into an endless loop until someone is contacted who has SSAS experience (in this example me), and I told them not to panic – it typically takes over an hour to fully restart SSAS on that particular Server. One scenario that I have seen is where the Server hosting SSAS is restarted, SSAS does not respond for some time (because it’s still loading Databases into memory), causing panic. In my experience, the memory required is roughly double the size of the DataDir folder. This comes in handy as you can estimate how long it’s going to take for SSAS to fully start if you know the amount of memory that it will be using. Leave it alone! How long before SSAS loads?īut how long is it going to take? In my experience, there’s a correlation between the size of the SSAS Data (DataDir folder) and the amount of memory used. Find the “msmdrv” service and observe that the memory usage increases over time until it settles – at this point, all of the Databases are listed in SSMS.Īs a rough check, is the memory allocated to “msmdrv” continuing to increase? If it is then it’s still loading and it won’t show any Databases until they have all been loaded into memory. This behaviour can be observed from Windows Performance Manager. So, what can you do?Įarlier in this article, I mentioned that SSAS takes time to load Databases into memory. It is possible to configure SSAS to provide feedback on what it’s doing, but that needs to be set up beforehand. ![]() I have also seen situations where SSAS has gone into an endless loop, trying to start, hitting a corrupt Database, then re-starting. I have observed a situation where engineers have re-started an SSAS Server (because that fixes everything) because it hasn’t started up in a reasonable time. If a Database fails to load then SSAS is unlikely to start. Starting SSAS is very much an “all or nothing” operation. Please feel free to email me if you can help. As mentioned in a previous post, I am learning more and more about SSAS so may well update this in the future as I learn more. SSAS does not tell you what it is doing – it has a Log file (msmdrv.log), but SSAS does not list the Databases as it loads and performs integrity checks as the SQL Server Engine does. If an individual Database fails at this point, it will be marked as “suspect”, but the SQL Server will be ready for use, together with the other Databases. It will report individually as each one is integrity checked and made available. SSAS Loading ProcessĪs part of the startup routine, SQL Server lists each of the connected Databases that it has to load. This is certainly different from how the SQL Server engine operates. It does not update the list of Databases each time another one is loaded, just once it has processed all of the Databases. The SQL Server ERRORLOG displays status information as the SQL Server starts, listing each Database as it goes through the Recovery process to ensure data integrity.įor SSAS, loading the Databases is an “all or nothing” operation. While SQL Server works in the same way, it is clearer to see exactly what is going on with the SQL Server Engine. SSAS is busy, but it won’t tell you what it’s doing and certainly won’t give you a clue on how long it’ll be before you can access data.Ĭontrast this with the regular SQL Server engine. If you try to query the SSAS Database remotely at this point, you will receive errors. SSMS becomes unresponsive, and can also block your usage of other query windows on other Servers (SSAS or SQL Server). At this point, SSAS decides it should load the Databases and perform basic consistency checks. using SSMS to connect to the SSAS Instance, right-clicking on Databases. ![]() SSAS only loads this data into memory when a user attempts to view the list of Databases – i.e. When accessing SSAS, most users want to use data held in a Database (Cube). In my opinion, this status is not entirely accurate. When starting SQL Server Analysis Services (SSAS) Tabular, the Service is quick to report that it has started (and it has). Your experiences might differ, as “classic” SSAS works differently, and you may be lucky enough to be working with relatively small Databases. This article is based on my experiences with SQL Server Analysis Services Tabular, together with large SSAS Databases (Cubes) – i.e.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |