Big fan of notion. Not a fan of the data lock-in or haphazard security. For tools like this, SaaS will almost always beat open source just by virtue of elbow grease and a product direction. So I’m willing to forego the open source alternatives. But I really wish Notion gave me more options for how to host my data. AFAICT they offer no “dedicated storage” or “on premise” plan, even if you want to pay for it.
In fact I heard a rumor that the entirety of Notion runs on a single unsharded database. If true, combined with the fact that this API took so long to ship, I have to assume that there is a crippling amount of tech debt in their stack. Hopefully it’s growing pains and they’re sorting it out, but it doesn’t make me feel confident about the security of my (most sensitive) data.
It’s maybe also a cautionary tale about what happens when you dismiss too many early decisions as “premature optimization.” You ship fast in the beginning but a few years later you’re buried in all the debt you generated.
Granted, it’s a good problem to have. If the choice is between a profitable, popular company buried in tech debt, or the perfect stack with no customers, I’ll take the debt.
We shared a bit more in the past about how most of our user content was stored in a single database instance (see more here: https://www.notion.so/notion/Focus-on-performance-reliabilit...). That is no longer true and we recently re-architected this database to scale horizontally.
Totally understand that you want more options for how to host your data. Unfortunately, we don't have any plans on that front at the moment.
My company had Notion hit us up for an Enterprise contract. When we asked if they offered enhanced security, single tenancy, or even they’d accept some sort of liability if they had a breach … the answer was no, no, and no.
We ultimately decided to not go enterprise since it offers little value over their existing plan and keep sensitive data out of Notion.
I hope Notion gets to a point where they can offer stronger guarantees to their customers. If they pull it off they’ll easily be mission critical software.
Notion doesn't even need to offer these guarantees in their SaaS service. They can just ship to my Kubernetes cluster, and let me manage the ingress and storage, and that would be enough to make it mission critical software.
In the meantime, we end up using tools with a little bit less good UX, but we just can't share that data publicly. UX can't win over security, and I hate having to choose.
That seems exactly like what customers want — just when you share things publicly it goes to a separate public sandbox? It should be fairly trivial to obfuscate
Thanks for being upfront about “don’t have any plans for that”, instead of a vague “we’re looking into it” which sounds better but effectively means the same thing.
>Totally understand that you want more options for how to host your data. Unfortunately, we don't have any plans on that front at the moment.
This is unfortunate. I tried Notion a few months back, and almost loved it enough to pay for a subscription. Once I saw that, I refunded the subscription and went back to markdown and git. I love your software, but the lock-in is frustrating.
"Lock in" is not a part of our strategy. We try to offer data portability via import and export; you can export your entire workspace as HTML (most metadata, pages link to each other and can be browsed offline), PDF, or Markdown (most easily editable). Given an HTML export and a few hours of scripting, I think you could migrate to any API-accessible document storage strategy.
Except if the export is broken, as it has been for us for months with no answer from support. Our company account got quite large to about 40 GB. Ever since, all export options just result in "export failed"... and as nothing can really be deleted; I'm scared we'll be locked in permanently.
Don't get me wrong. We love Notion. But we also need our data to be portable and backed up to not be liable.
You know, it could be that they just have a product vision and want to build and sell it to people that want it. If it's not what you want, you don't have to use it.
If your product is only profitable because you created artificial friction, your business is unlikely to exist a decade from now unless you are thinking about diversifying already.
Allowing me to host my data myself should not reduce your profits as long as your product is still the best.
If you’re focused on (relative) short term profits to get paid and retire on a beach, and you succeed, good for you. I’m totally jealous and wish I was you.
I don't see how or why these things have to be related in any way whatsoever? Atlassian offers self hosted solutions and presumably do so because it's profitable. I don't see why offering a self hosted solution has to somehow mean less profit, or to the extent of non-profitability for that matter.
You could make it *more* expensive to use the self hosted version. It's straightforward to justify if you consider the additional development overhead, and if you include some kind of a support package.
My question is genuine. I see strange justifications like this often when it comes to self hosted versions of X, Y, or Z product. What exactly is the risk to profit here? I can see extra engineering overhead as a possible risk, but that's a solvable problem. There are many ways to handle software updates. Spinning up new backend environments is often done on devs machines daily when developing. I don't see any novel problem that hasn't been solved.
The "we have no plans for this", or "we are looking into it" are all often used ways of avoiding the question of "why aren't you doing this?" There are plenty of companies out there that for one reason or another, need to control the environment where their data lives. Why give up that business? Why do people who work at those places have to often settle for worse products because of this? Why not give businesses *choice* of storing their data or having someone else deal with the complexity?
Is it that the presumed market for this is too small? And why is nobody being transparent about this?
You are reasoning about this well. Offering a self hosted solution is a solvable problem that can be profitable. The challenge might be that there are multiple different strategies the folks at Notion could follow, but choose not to. Each is solvable and each can bring more profit. And yet they need to decide what to do and what not to do. You should not chase every rabbit and solve every solvable problem.
You should try obsidian plus syncthing. I use notion at work but hate how slow it is most of the time. For everything else and for all of my own stuff at work I use obsidian.
Even pretty established players like Asana don't offer on prem.
Curious what you'd be willing to pay for this, as it's exclusively an enterprise feature. My guess is an absolute minimum of $50k/year contract value, or 250-350 users at an enterprise per-seat price (just guessing).
Aside from going fully on prem, the idea of strictly dedicated storage is interesting -- like "bring your own S3 bucket" simplicity. I don't see that often outside of $10 desktop apps.
Even if the offer is appealing, this is unlikely to happen. Confluence who used to offer it, has stopped and soon everyone would need to use the cloud version. The thing is that for a collaborative tool like Notion (and others) being Saas, and in the cloud is kind of mandatory. As you need your tool to integrate with other services and thus, can't be static.
The only possibility left is "dedicated storage". That works for static ressources. But the `page` content itself is dynamic and changes often, so you will need more than "bring your own S3" and something like "bring your rds" but that starts to be pretty complicated. Quip though still offer on prem AFAIK.
At the end of the day, the issue with collaborative tools, is that you can't e2e encrypt easily, and almost impossible if you offer full text search. The best solution is to trust your provider for data security. And use maybe some daily backups in your own network.
Atlassian tools still offer on-prem, but you need a data center license now and they doubled the price of that. Unfortunate, but they'd be losing some huge customers that cannot run in the cloud not just as a matter of policy but technology. The entire IC developer platform is using the Atlassian suite, for instance, but on JWICS, a classified network with no connectivity to Atlassian's cloud.
What's going to end up happening if more vendors decide you can only use their service from their cloud is the largest enterprises and governments are going to exclusively use Microsoft and AWS since they're more than willing to just build you your own cloud that meets whatever security and disconnection requirements you have, no matter how stringent. Already no new government projects that still have a choice are going with Atlassian tooling.
We'll see how it goes for them, I guess. Maybe they don't need governments and Fortune 500s and can get by selling exclusively to startups with no real security or lock-in avoidance requirements.
Well, the comment still stands. You can't /reasonably/ deploy Confluence on premise any more when weighing up the cost feasibility of using the on premise service for Data Centre vs Server.
The idea of using their cloud service (Atlassians) wouldn't be that daunting if it wasn't for the consistent feed back I see here and else where that performantly, it absolutely sucks.
You're right about Asana -- but it's worth noting that RoamResearch ($100/yr) offers private, offline dbs, and Roam's OSS "vassal" app (ie, clone) AthensResearch is self-hosted and free-as-in-beer.
Not sure about Roam, I've been using Obsidian, and my purpose is for personal, making note archive offline. So never had any need for collaboration. But with Sync and Publish features, maybe there is a possibility.
I think Roam has been fully collaborative since the launch. The browser keeps the whole database in IndexedDB and syncs it continuously with a WebSocket streaming Datomic-style transactions.
I'm most saddened by the people who spend their time writing plugins for Notion / Roam (and to a lesser degree, Obsidian), providing extra value for free to a non-free product that will profit off the free labor. (or rather, I'm saddened that companies like Roam are so willing to exploit community engagement for a profit)
Time would be better spent contributing to an open source alternative like Athens!
Roam's site[1] only mentions $165/yr. The 5 year plan is $500, but then... if you stop using it after a year, you would have paid $500 for 1 year of usage. If you're thinking of no lock-in, you need to see the monthly price, which is $15 month, way more expensive (for personal use) than anything notion has (free/$4/$8 per month)
The support costs are a solvable problem. The challenge is architecting your stack to accommodate external storage, and delivering a usable experience to your users when they connect their resources.
It would be nice if AWS could deliver a smooth OAuth experience so that in a few clicks, a user could provision an S3 bucket and share it with a service provider as easily as they can grant access to a GitHub repository.
> Even pretty established players like Asana don't offer on prem.
Sounds like an opportunity ;)
Starting or rearchitecting a project gives you the rare chance to make decisions that your competitors never even had available to them. You can leverage new tech to move faster, and you can learn from their mistakes to avoid their most costly traps.
“Bring your own S3 bucket” actually sounds pretty awesome and I wish more companies would do this. Everybody wants to be a middleman controlling the transaction, but they’re missing a whole class of customers by not offering self-directed storage options.
This is a great example of a decision Notion could make now, that Asana would have a hard time adapting to. As long as they’re rearchitecting, they should keep an eye out for opportunities like that.
As for how much I’d pay – if I’m the only party with access to my raw data, and I can run the software locally even if Notion goes out of business – for a team of 5-10, I would pay $50-75 per user per month.
(I work at Notion, but not on the Infrastructure team)
We completed our sharding migration almost a month ago (https://twitter.com/jitl/status/1383235281457876993?s=21), and are now using >1 primaries. The project took a long time and really made the “changing the engine while the car is doing 200MPH” analogy true for me.
I hope I didn’t come off as too critical. I do love Notion as a product and my org is a paid member. I just wish there were more options around data hosting.
None of these problems are the mark of bad engineers, either. It’s just what happens when you ship fast and grow fast. Like I said, good problem to have :)
I absolutely love Obsidian except for one crucial flaw: no tabs. They've made every editor view a "pane", which means that on a laptop you can effectively only work on one or two files at a time. There's a plugin that adds tabs, but it's pretty awkward and not what you normally think of when you think of tabs. Otherwise, Obsidian is truly incredible.
Obsidian is not a replacement for Notion tho, no databases, a lot of custom configuration, no nice templates, etc... I can't take how slow Notion is, and I'm looking for a self-hosted replacement, but Obsidian is just not there yet, except for the note taking part, that I dont care about.
You can query Markdown files with the (third-party) Dataview plugin. Wrote about it here[0], very simple query syntax. Only tables and lists, but tables is the one I've used by far the most in Notion.
> no nice templates
Well you can always create your own. There's a first-party templates plugin[1] and a similar third-party one that also allows custom variables and has a showcase[2] (think: you can curl the weather and add it to your journal).
Having wrote that, your "a lot of custom configuration" definitely applies.
except,in Notion, you don't need 3rd party plugins, or "create your own" templates, you already have it out of the box, don't get me wrong tho, I don't love Notion, is very slow. The biggest problem I see with obsidian is, requires a lot of configuration, like you mentioned. I just want something that works.
We use plenty of cloud stuff, I guess that wasn't clear from my initial comment. We use Office 365 & teams for our 'business' teams, Slack for tech teams & Monday.com for our 'Jira'. The business has evolved from on prem exchange so its natural a trust in MS is embedded.
I wanted notion to get out of document hell. Legacy business with many spreadsheets and word documents. With notion it wasn't the cloud being the issue, we run all in house software & DBs in AWS.
But some startup from San Fransisco offering a platform with notable downtime / performance issues and no direct ownership of DBs was an issue for something like a wiki where you want to be able to publish sensitive information, ip etc. We're in a regulated industry so compliance & security input is ultra relevant.
Sharepoint is a fucking abomination, it's what was previously used. Notion makes a perfect wiki solution.
If it had a self hosted option we'd have adopted it. It was turned down, and fairly so, by our ops team for the hosting situation. There's no reason for something as *simple* as a wiki should be completely outsourced. Portability is a factor too.
We chose wikiJS in the end. Great wiki but terrible writing experience. Tend to write everything in an external markdown editor and just copy it in.
In my experience, you will find this kind of org uses Office365/Outlook, and not Gsuite/Gmail, as they have in place a commercial agreement with Microsoft which covers regulatory requirements put upon them (and some they put upon themselves).
Of course they could _try_ to get an agreeable contract in place with Google, but Microsoft is Enterprise-friendly and can make this happen much more readily.
I'd imagine the HN crowd cares less than executives at large organisations and government departments. There's a requirements for industries and businesses to either own their own data or keep that data onshored in their jurisdictions.
I'm actually working on launching a self-serve on-prem version of Retool over the next few weeks. Would love your feedback on whether the way we're thinking about it resonates with how you feel about on-prem/self-hosted tools. If you're open to it, let's connect jamie at retool.com
Do they still not support things like 2FA? At my previous company, we couldn't get buy-in from our security team due to them not supporting 2FA. The team loved Notion, but that was a deal breaker. Also, if I remember correctly, magic email links could not be disabled. There is no excuse for not implementing 2FA w/ an authenticator app -- it took me a couple days to implement it into my product. They've had people asking for 2FA for literal years (just search Twitter!)
That only works if you can put the app on your internal network; Notion is neither self-hostable nor available 'on-premise' w/peering, that's what people are complaining about here.
You can't magically put 2FA in front of some third-party public site. (You could from within your own network.. but that's perhaps worse than pointless!)
As a consumer, I want my software on-prem. But as a producer, I want to deploy as frequently as possible, so that I could iterate without friction, and I would try to avoid platforms/techs that are not hot upgradable.
As a consumer, I want no data tracking. But as a producer, I want data-driven growth, so that I could understand where I did wrong and how to improve it. And data comes from tracking.
This paradox is not just limited to the software industry: For example as a consumer, I prefer reading texts. But as a producer, I found videos are easier to monetize because of the attractive/addictive nature.
More often than not, the formers which benefit consumers are not justifiable in pure economic senses compared to the latters, thus the reality we are living in nowadays.
It’s possible to make self-hostable software that auto updates. Besides, managed deployment and single tenancy are not mutually exclusive.
In this case I’m not even asking for self-hostable. I mostly want single tenant. Yes, self-hostable or “bring your own storage” would be cool, but provider-managed single-tenancy at least eliminates a whole class of data security bugs. This way, even if an application-layer dumps all the rows in the database to the client, at least they’re only my rows.
Sorry I wasn't being clear about this part. What I meant is a comparison:
If there is an on-prem option, the release, the update/migration, the installation environment, and the documentation on deployment should be carefully crafted. All of these not only require more effort, but also require people to keep in mind not to stomp on things. And even worse, they have to search for potential problems with their imagination because they don't have access to the data with edge cases they're dealing with. If people are not careful, it's very easy to seriously screw things and lost some customers forever. On top of that, if fixes are required, the team usually needs to fix multiple versions over and over, of which the codebase might have drastically changed. The teams are usually less and less "brave" as time goes by.
For most SaaS without on-prem options, people can just focus on the functionalities and deploy many, many times a day. For non-critical issues, people don't need to worry about them, just debug what's going on and respond fast, and the issues are very likely fixed in hours. Plus they can use components like Kafka, Elastic Search, or even other SaaS as needed, without having to choose between re-invent things or ask everybody to install a zoo of components in their on-prem solutions.
That's one of the key reasons why web application is so popular nowadays compared to what we've done in the 90s desktop applications. Because it enables the "move fast and break things" mentality because more things are fixable in web applications. On-prem is like 90s desktop applications in disguise. Just try to imagine how horrifying if someone breaks the upgrade functionality. Or if one needs to fix a bug in Windows 10, they also need to fix it from like Windows Vista to Windows 10, where the codebases are drastically different.
In conclusion, on-prem sounds applaudable to the audiences on HN, but the reality is it doesn't work well for most for-profit businesses.
I have more respect for and trust in a product that I know is keeping things simple and straightforward vs. fucking around with Big Data consistency problems prematurely.
I’m curious: notion lets you export everything as markdown and CSV. If you had a sync/backup job running, that, say, imported the csv to a Postgres database every night, how would you feel about your concern?
In fact I heard a rumor that the entirety of Notion runs on a single unsharded database. If true, combined with the fact that this API took so long to ship, I have to assume that there is a crippling amount of tech debt in their stack. Hopefully it’s growing pains and they’re sorting it out, but it doesn’t make me feel confident about the security of my (most sensitive) data.
It’s maybe also a cautionary tale about what happens when you dismiss too many early decisions as “premature optimization.” You ship fast in the beginning but a few years later you’re buried in all the debt you generated.
Granted, it’s a good problem to have. If the choice is between a profitable, popular company buried in tech debt, or the perfect stack with no customers, I’ll take the debt.