This site uses different types of cookies, including analytics and functional cookies (its own and from other sites). To change your cookie settings or find out more, click here. If you continue browsing our website, you accept these cookies.
Can we get a good compare between the two? Is there an additional file type that would be preferred to either as a file type to use for extracts/outputs and then as a source for analysts to query off from? We are trying to create efficiencies for our analyst community by creating single source extracts and would like to know how others are approaching this for their teams.
"For analysts to query off from" and "create efficiencies ... single source extracts".
YXDB is a proprietary format and must be read through Alteryx while SqLITE is open sourced.
YXDB is write-once where SqLITE allows for insert, update and delete.
Who am I kidding? Write it all to YXDB. Why bother comparing the two? If you are creating extracts, it sounds like you are not looking to perform updates to the data. Even if you were, you likely could run the updates through a custom re-build and still provide better efficiencies than SqLITE (which has limits to data). When you "query off from" the data, I do wonder where you are going with this. If the data is clean and ready for VIZ, then you might be choosing QVX, TDE, hyper etcetera.
I've played with using SqLITE before, but always undo that thinking.
Alteryx ACE & Top Community Contributor
Chaos reigns within. Repent, reflect and reboot. Order shall return.
If you're looking at Alteryx formats - you might also consider Calgary files (CYDB). These function like YXDB's but allow you to create indexes for querying. Depending on how large these extracts are it might be worth going this route vs a yxdb.
If at all possible, I keep everything in YXDB or CYDB format. Sometimes that's not possible, however. But if all your users and all processes that will be making use of the data are able to read in YXDBs (meaning they have Alteryx), then by all means YXDBs or CYDBs are the fastest and most condensed file format that I've ever encountered.
CYDBs are, generally, for 200+ million records that you'll be querying on a regular basis, and want to query super quickly. Some downsides, like the index files it creates (one per field) has to be in the same root folder as the CYDB. This can make the folder look a bit messy, but if you'll be querying large amounts of data, it's more than worth it to have a "messy looking" folder. Although truth be told if you sort the folder by type and not name, it's not that messy looking at all.