I’m putting in more data, will try to start some knowhow tool in Anytype. For me, it is surprisingly stable with a missing feature set. A working delete would be perfect
Anytype is lot more reliable now . So I started daily journaling. I take notes daily . All my notes have codes, screenshots, videos . No problems in syncing so far.
https://community.anytype.io/t/-/2242 Maybe you have a proxy? I would need a proxy for use in company
I have noticed a few performance issues that seem to scale exponentially with either the number of objects or the number of tags in use. If this isn’t a known issue, maybe I can do a timed experiment and show what I mean. Two months ago, it took me about an hour for Anytype to export my data to Markdown, and maybe three hours to sync to a new machine. I’d say I’ve only grown my data by around 20% since then, but exporting is now around three hours and syncing takes the better part of a day. I have a lot of objects, but they’re all pretty small and simple.
Performance definitely has decreased as I’ve accumulated about ~500 objects (is viewable if you go to Relations → Created Date). It takes about half a second of loading time to open a page containing lots of blocks or a Set, and a spike in CPU usage can be observed for about 20 seconds after opening the application. The Anytype Helper (renderer) process also takes up a significant amount of memory on my system - usually above 2 GB.
Still, I am satisfied with the performance so far. It’s a major improvement over web-based apps. Yet, I can’t help but be concerned about how Anytype will scale as the amount of relations and objects that I create increase to thousands, or perhaps tens of thousands.
I would predict the time needed to open Sets and their respective Views would increase dramatically as the amount of Objects increase, as the application has to manually filter and sort through every individual object until the ones which meet the filtering and sorting criteria are met. I’d trust that the Anytype team would have found a suitable solution by then - this app is still in development and there are more significant and pressing concerns, after all.
I’d like to try to revive this thread, to see if more people have been experimenting with large volume of data in AT. I’m particularly interested in performance with Sets having a huge number of entries (hundreds, thousands and more) - how fast they display, filter etc.
I think the vast majority of my Sets would only have a few hundred entries, but many would be huge.
I’d like to know if its reasonable to expect AT will one day be able to deal with this volume of data in a performant way, or if I should be looking at a different tool for that.
Would be great to have an honest answer/comment from someone from the team on this as well, even if its to say “we’re not sure yet”.
PS: here’s a related post which I missed when I did this post originally
With the painstaking process of manually adding entries to sets, I have no idea how anyone would have thousands of entries in one set yet. Also it is my impression that it currently is not possible to import to sets. Correct me if I’m wrong.
You cannot. But I am certain that people will be adding in a lot more data after local API is available
Good point!
Though I thought it was already possible to import from Notion, which I assumed would involve interpreting properties as Relations. But I don’t use Notion, so not sure.
The Notion import was designed a long time ago, and sets didn’t exist back then.
Yes, but Relations did, no?
Again, I haven’t used it so not sure if there’s anything in Notion that would convert to Relations (properties maybe?).
From my usage experience so far, it seems as if individual Sets slow down with usage. One Set that I’ve been using for three months or so had deteriorated to the point where opening it would result in more than 20 seconds of loading time, across both Desktop and Android. Fortunately, creating a new Set with the exact same filters, views and sorts as the old, deteriorated Set fixes the issue and loads essentially immediately.
Right now, with ~1500 objects (about half or so are media attachments like images and files, and the other half are pages), Anytype seems to still be decently performant, aside from the aforementioned issue with Sets.
Thanks for that!
That’s weird! If you haven’t yet, please report this bug.
I’ve started using Anytype as my daily driver around 2 months ago and so far I’ve been pretty pleased with the experience. I didn’t have any issue with the performance and to be honest I usually use my Anytype on my mac but I also have it on my android phone. Sure the sync may get a couple of minutes to fully complete but it’s not an issue for me. Also, My sets appear with a near instant speed so no complaints there.
As for the size of the database, I don’t know the exact number but I can say I use relations quite heavily. I will attach my knowledge graph bellow and I would be happy to answer any of your questions.
Thanks for that! Very useful. Hoping more people with big knowledge bases share theirs as well.
I wonder if there’s a way to show the total number of Objects…Would be nice to know.
What would you say is the average size of your objects, in terms of volume of text, images etc?
I have a single object that take a lot of time to sync and create/delete bocks.
I don’t think it’s related to the amount of blocks inside the objects, since I have a couple of others similar where the slowness doesn’t occur.
It’s one of the first object I’ve ever created, with a lot of edits through time, so I think the long history of commits it’s slowing down the syncronization.
But for the rest of my experience I’m really satisfied
Interesting. If possible, would be useful to share that problematic object with the devs (not sure how) to see if they can get some insight from it (and hopefully avoid it happening again)
I think I have around 300-400 objects in my anytype. most of them has 15-20 lines worth of information, and around 20-30 of them are relatively long. all of my objects have its own custom icons uploaded by myself and under 10 images besides them. Lastly, nearly all of my objects are connected with each other as you can see having around 5-10 relations each.
Its an impressive graph view but I wouldn’t consider 400 object large, specially with just 15-20 lines of information in each object. I would hope anytype doesn’t suffer performance.
Starting at 10.000 objects or more I would consider it large and it should still not suffer performance then (or even higher 15.000 or 50.000).
If I look at my intended usecase for Anytype (or other note taking apps) I use it for daily notes (365 notes/year), my collections (movies (1000+), games (1000), Lego (100), etc), inspiration boards (500), life organizing (250), inventory (1000), etc.
This will quickly ramp up to 5000+ objects within one or two years if Anytype would become my daily driver which at this moment it isn’t for obvious reasons being so early alpha.
So I hope that Anytype does get tested in the 10.000 or even 100.000 objects and still hardly suffer any performance!