Max Number of Pinboard Elements?
I've tried searching but have not been able to find any best practices for pinboards. My team has come across a situation were a pinboard population is delayed due to the number of tables and graphs(approximately 80 visualizations). Even so that when trying to filter, we receive a connection error. I've created a ticket for the error, but wondering in general if there are any guidelines/best practices to follow for pinboards to operate in the most efficient manner.
I don't know if there are any published best practices, but we try to cap our pinboards between 12 and 15 objects. If an end user's question goes beyond that many objects, we feel that we are getting into a different subject area that could merit a separate pinboard, or they can use the drill down feature to dig in more granularly. We've also seen issues with putting too many KPI boxes in a single area. Since TS tries to resolve them all at the same time, it seems like it's caused contention. So displaying fewer visible objects at the same time could help.
In general, with TS or with other BI tools, we try to treat a single page/ pinboard as addressing one specific question. For 80 visualizations, do all users need to see all of them at the same time? You can try to break up the pinboard by user profile or how often they need to view the data.
In regards to the connection error, we experienced something similar with a single query using too much memory. Our SRE was able to assist by limiting the memory a single query can use by adjusting the "falcon_worker_memory_notification_percentage". That seems to have helped.
Thanks for your reply. Certainly makes sense. The audience is a small group but have insisted on having all visualizations present. I agree, drill down or filtering would make more sense and limit the number of objects in the pinboard. I'll get with our SRE to check on the query memory percentage.
That's the exact same question I've asked! The info they are after could easily be consolidated into a dozen or so visualizations and filtering applied. Normally, this is the approach we would take, but this particular ask is an outlier, from upper management, so figured it was time to ask about best practices/system limitations.
Paul - I don't think we have any setting to limit the number of pins on the pinboard, although it does seem to be a good idea to make this a configurable setting. I also agree with other answers on this thread and you should probably go back to the users who want this created and ask them the question - "WHY?". Apart from performance, there is another issue with a large pinboard, which is that ThoughtSpot currently does not have an easy way to quickly jump or search for a pin within a pinboard. The user will need to scroll to get to a specific visual that he/she is trying to see.
But now that you are in this situation there might be something you can try. ThoughtSpot executes a certain number of pinned visualizations in parallel so that we don't overwhelm the server with 80 requests at once. Also, TS will run the queries as the pins fall into the visible viewport of the browser window. We should update this setting to maybe set the max parallel pins to execute in parallel to maybe 3 or 4. This may help with certain timeouts that you may be encountering.
Performance depends on a lot of things and you will also need to see how the search for those pins is built, but limiting the parallel execution count for pinboards may help.
Kunal, I appreciate the info. Would I need to connect with our SRE to set the max parallel pins?
I've recommend that the existing pinboard be split up into multiples or we utilize a combination of filtering and drill down. I'm hoping the business team will agree. I thank everyone for your insight into this issue.