Hello! I run a small & free hobby project called Property X-Ray for analysing flood, bushfire and other type of data relating for Australian properties.
Weeks earlier I got quite interested in trying out the project haystack tagging system and was given some direction on the forum there to build out the haystack modelling. So I thought maybe lets give fantom a go too, since the haystack tooling itself is written in fantom; makes life easier.
I am currently looking at revisiting databases, preferably one that can store timeseries data. So far I ha ve been using the sql pod (wraps JDBC) to talk to a SQL database. Looking around for what else is out there, I found the Folio database, it looks like it stores tagged records plus timeseries history. 2 in 1, exactly what I might be able to use as I start to store historical records like property sold price.
Is Folio meant to be usable as a general purpose store, or is it really shaped around building and IoT data? Half my data is append only history (prices changing, development application statuses) so the historian idea caught my eye.
Cheers, Sam
brianYesterday
There are two flavors of Folio. The open source version in the Haxall repo is really just an in-memory document database that stores Dicts. It is great for working with Haystack models. But it does not include a timeseries database/historian.
The version of Folio that ships in the commercial product SkySpark does include an optimized time series database. It also includes more advanced indexing and query planning than the open source version.
Giant TTYesterday
Ahh! Thanks a lot, tat clears up. So the open source Folio is in mem only;half my use cases are the historian part, sold prices and leased_price, statuses piling up as append only history over years, so in memory wont really work for me there.
Sounds like the sensible pattern is to keep the tagged model in Haxall Folio for the live side, and just keep my own SQL database for the historical timeseries, rather than hoping one thing does both.
Either way good to know before I went down a rabbit hole haha. Appreciate you taking the time!
Giant TT Yesterday
Hello! I run a small & free hobby project called Property X-Ray for analysing flood, bushfire and other type of data relating for Australian properties.
Weeks earlier I got quite interested in trying out the project haystack tagging system and was given some direction on the forum there to build out the haystack modelling. So I thought maybe lets give fantom a go too, since the haystack tooling itself is written in fantom; makes life easier.
I am currently looking at revisiting databases, preferably one that can store timeseries data. So far I ha ve been using the
sqlpod (wraps JDBC) to talk to a SQL database. Looking around for what else is out there, I found the Folio database, it looks like it stores tagged records plus timeseries history. 2 in 1, exactly what I might be able to use as I start to store historical records like property sold price.Is Folio meant to be usable as a general purpose store, or is it really shaped around building and IoT data? Half my data is append only history (prices changing, development application statuses) so the historian idea caught my eye.
Cheers, Sam
brian Yesterday
There are two flavors of Folio. The open source version in the Haxall repo is really just an in-memory document database that stores Dicts. It is great for working with Haystack models. But it does not include a timeseries database/historian.
The version of Folio that ships in the commercial product SkySpark does include an optimized time series database. It also includes more advanced indexing and query planning than the open source version.
Giant TT Yesterday
Ahh! Thanks a lot, tat clears up. So the open source Folio is in mem only;half my use cases are the historian part, sold prices and leased_price, statuses piling up as append only history over years, so in memory wont really work for me there.
Sounds like the sensible pattern is to keep the tagged model in Haxall Folio for the live side, and just keep my own SQL database for the historical timeseries, rather than hoping one thing does both.
Either way good to know before I went down a rabbit hole haha. Appreciate you taking the time!