What’s “large”? I know people for whom 10K rows of data is “large” and people for whom 1B rows is “large”. However that’s only one “large”, the other “large” is how many marks are being drawn in a given view: is it 100, 1000, 100,000, millions? Also are you trying to have all that data local in flat files or a local database server, or are you connecting to external servers?
A few notes based on my experience w/Tableau for Mac and Windows performance...I’ve been on the same Mac (a mid-2012 13” non-Retina MacBook Pro w/16GB of RAM and an SSD) since Tableau came out with a Mac version and also run Tableau under Parallels and regularly run Tableau on PCs:
First, the hardware:
- Tableau likes RAM, the more the better. With Tableau Desktop my impression is that we get more performance gain having lots of RAM than more CPUs. Given that most MBA’s are generally shipped with smaller amounts of RAM I’d hazard a guess that this is the most likely cause of the lower performance. (Assuming that raw query times on the MBA are the same as when running the same queries on other machines).
- SSDs pretty much always beat spinning HDDs, that’s something to keep in mind.
- Also make sure you have enough swap space on the SSD. When my SSD (that has the OS and Applications folders on it) gets below 10% of available space then performance sl o w s w a y d o w n. The usual recommendation with SSDs is to keep around 15% of available space.
- Watch out for programs like Photos or Parallels that can suck up lots of CPU & RAM in the background.
- I also turned off automatic Time Machine backups years ago because they were slowing down the machine too much, I dunno what that’s like in newer releases of macOS.
- Generally I’m using flat files, when I have to run a database server on my laptop that’s the only time when I might wish I had even more RAM and/or more CPUs.
- Tableau’s native connector to Excel & text files is actually a Tableau data extract under the hood. So for me the only reason to not use a Tableau data extract right away for Excel & text files is when I’m actively changing the data structure of the source file, as soon as I have that locked down then I use extracts to improve that first connection time.
- Unfortunately since somewhere around v9.2 each new release of Tableau for Mac has gotten slightly slower to start up
- When I upgraded to macOS Sierra I started seeing big slow downs in all Tableau apps (v9.3.7 and v10.0.1) on the Mac for start up and querying data sources. I’ve got a call in with Tableau support on that. (FYI there hasn’t been any effect to Tableau on Windows 10 on Parallels, that’s still fast to start and fast to query.)
- Tableau does do hardware acceleration for graphics, however in Tableau 10.0 I had to turn that off due to some rendering issues that will hopefully get fixed at some point.
If you’re at all looking for a portable machine then my suggestion right now would be to wait a few weeks until the rumored new MacBook Pros come out and get one of those. Their CPUs will likely be close to the same performance as a Mac Pro or iMac and the graphics card will likely be fast enough for all but the really large data sets and/or running lots of apps at once (like running a database server on your machine).
- When I upgraded to macOS Sierra I started seeing big slow downs in all Tableau apps (v9.3.7 and v10.0.1)
on the Mac for start up and querying data sources. I’ve got a call in with Tableau support on that.
@Jonathan, this helped me with Sierra: