Swallow your pride or curiosity on rolling your own kit and accept the reality that OneDrive/Google Drive/Dropbox/Box.com/et al are going to be better for your needs. The single biggest benefit of SaaS products are their flexibility to provide service when you can't or don't want to support the deployments yourselves, which is basically startup mode.
Once you're off your feet and have an honest-to-god Enterprise IT team with a budget, let us deal with it. They'll likely keep end-user storage in a Collab Suite (M365, GWorkspace) unless there's a specific advantage or requirement for your business needs in running it on-prem.
Everything is a tool, and the use-case of these tools is in freeing you to solve the really hard problems of startups, i.e. survival, success, and sale/solvency.
Look, my job would be a lot more lucrative if I could convince the C-Suite to self-host everything on-prem again. The reality is that startups need to run lean and mean until they've got reliable revenue coming in, and this is where "off-the-shelf" solutions are going to win out over bespoke offerings.
Is it doable? Sure. Is it affordable? You betcha. Is it sensible for a startup? I'd probably say no.
Your staff may be familiar with Office365, Google Workspaces, Nextcloud, or any number of solutions, or none at all. So you will still have to leovide some level of training and support.
Not being locked in provides startups with ghe flexability they need, and the standards Nextcloud implements makes it even more flexible.
As for locked-in..._eh_, that's both fair (Google and Microsoft both have abysmal data exfil capabilities) but not as valid as it used to be (there are gobs of startups whose entire schtick is automating your exit strategy from a "locked-in" vendor). By the time vendor lock-in for your corporate IT is a concern, you've likely got the capital to splash on an actual, dedicated Enterprise IT team.
Remember that it is easier to unwind a startup if you only need to turn off cloud services instead of disposing of equipment and support contracts.
I recently moved (from Contabo to Hetzner), and I struggled to migrate NextCloud. I ended up creating a new Nextcloud container and re-uploading the files. I did have some more unique setup (trying to move from Caprover to Coolify, and the Coolify NextCloud image was using sqlite instead of MySQL for the NC database). If you migrate, just make sure to back-up the files locally first in a different folder, because the server is authoritative, so if on the new server the files are missing, they will also be deleted locally.
It's nice to hear you are dedicated to self-hosting. If you ever need a self-hosted Hotjar alternative, check out my UXWizz platform :)
In my young year I was pushing for something like this, an off the shelf self hosted system. I was worried that slinging it out to a thirdparty would be both expensive and bad for my career.
However I was wrong. Much as it was a dick to set up, migrating to Google (workspace? fuck knows what it was called) was totally worth the money.
At newer companies I've look at self hosting, but I just don't want to be on the hook for securing the stuff, or dealing with the email deliverability when some marketing prick does something stupid.
Do not self host. The only thing you should ever self host is software your company wrote; unless you have a dedicated team for that specific piece of software alone.
You will never meet the uptime of Google Workspace. Your tools will never have less bugs. You will never meet the security certifications. Your real-time document editing, which may you think doesn’t matter, will never meet your employee’s expectations. You will never have as good tools for automated legal compliance. And if it goes down, which it will, even a day of downtime is more expensive than years of Google Workspace in all but the smallest of businesses. Additionally, every time something doesn’t work (or, heaven forbid, you’ve been hacked), your company’s employees and lawyers can and will blame you instead of an unmovable entity.
That's what it's really about. Modern day "nobody ever got fired for buying IBM". It's not about the better solution, it's about blaming someone else if things go wrong.
Azure has been completely hacked twice now, yet people still move their shit over to Microsoft's cloud offerings. I don't understand how fucked in the brain you have to be to consider this a good idea, except for being able to shift blame.
If you self-host for a business, and your self-hosted instance is hacked when other self-hosted instances of the same software weren’t, you are at risk for legal action and a possible criminal investigation. Was it really the software that was hacked, or were you negligent? Was it truly an accident, or did you have malicious intent? Plus, define negligent - does not having a service like CrowdStrike installed count? (You might say, “obviously not,” but if it takes $50K to convince the court on that point, shallow victory there.)
If you have a family, even if this is only a 10% chance of happening, you would have to be, in your own words, “fucked in the brain” to put your livelihood and career on the line to save a few bucks.
In my ideal world, beer would rain from sky and nobody would ever get drunk. We’re not in an ideal world, and “CYA” is a valid reason until you have a proper, large, dedicated IT team.
Related: https://www.theregister.com/2024/09/20/cisa_sloppy_vendors_c...
I'm surprised someone finally has the guts to state it that clearly.
I think the better question is: Do you care enough to divert some time away from building your product?
Some folks might be really motivated by not going with Google Workspace. Others don't care at all. There can be great motivational effects in doing what you think is right, I wouldn't discount it as a purely economical decision. Sure, tools need to be efficient, but we also need to enjoy them.
I think this viewpoint would be more common here in Europe.
And our conpany's O365 also has significant downtime. We've had several major incidents in the last months alone (one of which was the crowdstrike related one, which we don't use but apparently Microsoft do)
This is not about Nextcloud vs GW. The spooky bit is that you consider taking on additional burden with something that will not improve your product in any way, when your product is the only reason why your startup maybe exists in the future.
That's a particularly dangerous habit to pick up, because you will have 1000 chances to make meaningless decisions and distract yourself every day.
Unless you have an extremely convincing reason to do something — and if you are wondering, you don't — don't.
The price is also not $10 per month. You have to increasingly upgrade for features that Nextcloud offers for free, particularly if they are used by enterprise (like multi user accounts).
In my opinion, nextcloud makes sense when you want to host your data on your hardware.
Or spend your time writing privacy policies and adding 3rd party data-sharing consent banners.
OwnCloud and such wouldn't be considered a mainstream resume skill, and comes with its own upkeep and maintenance, along with ownership of the entire backup/restoration process.
However, ideals aside. In many large companies, using Microsoft or Google products can also be a compliance headache (that is, if you're outside the US). A larger corp is more likely to be hit with such issues than a small startup.
Also, self hosting of course requires resources. I'm not talking about compute, in my experience that is very negligible. It of course requires people to keep stuff up to date and learn how to use it to its fullest extent. A larger enterprise can more easily afford this effort than a small startup. Even considering my idealistic stance, it is hard to ignore the low cost of entry as well as the ease of getting started with the big cloud offers.
that doesn’t only apply to google workspace, it also applies to things like payment processing, auth, etc.
Staten that, as a POOR solution to collaborative works on documents NexCloud/O365/Google suite might be "good enough" for some, NextCloud is not on par, but near, the burden for IT is simply the fact these products are monsters alone, essentially impossible to really knows entirely in codebase terms, complex to heal when trouble happen etc.
So yes NextCloud could be used in professional settings, and owning their own infra is better than living on someone else computer, but the real point is if your employees really need such tools/paradigm or not. Someone in IT could collaborate with a shared repo, notes, makes slides in org-mode and so on. Someone much less skilled could still use BookStack and alike. Only end-users already trained on office-like stuff could like these monsters.
Now, I would not use it unless you have to follow ISO norms/get governmental agreements for any company with enough money. If you're a three-person startup with one client barely paying two salaries, trying to find a bigger market though, go for it.
E.g. here are some specific things and examples of things you'll have to deal with, in no specific order. These are just some things I've had to deal with recently.
- You'll have to educate people in your group that there are at least 3 different ways to share files among each other and that they can all coexist in parallel (Individual Shares vs. Group Shares vs. Group folders vs. Circles/Teams) (I did a german blog post on this: https://bitbetter.de/blog/nextcloud-freigabe-chaos/)
- Handling of file/folder names with special characters is a mess e.g. if you have Windows and Linux clients there will most certainly be conflicts. (Luckily this has been fixed recently by the `forbidden_filename_characters` config option – which is not enforced yet via the Web UI) see https://github.com/nextcloud/ios/issues/2802
- Creating Nextcloud users with spaces in their names, will break CalDAV on iOS Devices (https://github.com/nextcloud/server/issues/15641)
- Nextcloud (aka Collabora) Office is very slow if you want to actually work collaboratively with it (no matter the power of your Collabora server) – unfortunately it's no match for Google Docs or Office 365
Don't get me wrong: It's still a fantastic Open Source project with tons of talented people and it's a beacon of hope in the GAFA world. Everyone should try it out (and help it evolve) so it can be better than the commercial alternatives. But going into this and expecting to get the same kind of product quality like Google Drive/One Drive/Dropbox will lead to disappointment.
What about onlyoffice ?
Can you confirm your experience comes from a nextcloud instance and a collabora instance on two different servers ? What was the bottleneck ? Network traffic ?
I think mostly the sluggishness comes from some input delay in the browser. When you open a Collabora Calc document and select any random cell, it just takes a couple of hundred miliseconds till the cell is actually selected. I can reproduce this with brand new instances, no matter how beefy the server is.
Give that 5th or 6th 9 it's proper due (almost none almost always) and life gets quite simple and manageable and efficient and your tools provide value instead of extracting it.
If you're already using it for large file version control for, e.g., gamedev, and don't mind the cost, how well does it work to store all other company documents? I'd assume it has better scalability and permissions management than Nextcloud (not to mention the version control on par with git).
It's a joy to browse the design library and do code reviews with Swarm as a plus.
IMHO, having the diversity in storage and computing would help the current tech world a lot.
I think selfhosting open-source services is more useful for niche SMB's that specifically require on-prem data.
I've tried several of the "budget" NextCloud hosts. I'm not going to name names, but all of them were very disappointing. File syncing frequently broke in hard to diagnose ways. And from a perspective of overall service, I would get 502s, 503s, or 504s far more often that I should have. This was with multiple budget providers. I didn't try any of the more expensive providers, because I couldn't afford it, so maybe this is a "get what you pay for" situation... But in theory, the size servers I was paying for should have been able to handle our traffic volume.
Anyway, after a couple of years of trying to make hosted NextCloud work for us, I gave up and bought Dropbox's paid service and haven't had any issues with it.
We use Cryptomator on top of Dropbox to ensure data privacy, by the way. Back when we were using NextCloud, we had been using their end to end encryption plugin until we discovered a silent failure mode in which it was uploading documents in plaintext to the "encrypted" folder. I believe nowadays the recommendation even for NextCloud is to use Cryptomator on top, rather than their built-in encryption.
If you are worried about someone stealing your ideas, don't be. First, nobody cares about what you have until you make it big, which is going to happen a long time from now if ever. Second, the major services all provide enterprise deals that ensure privacy, enough that the largest companies in the world and many governments rely on them. You are far more likely to be hacked if you try to roll your own than if you use a popular service - especially if you aren't an experienced IT admin.
If you are in a start-up you want to put your full focus on your product - don't waste time on infrastructure. Use popular online services, use popular brands for hardware, use popular languages and popular libraries. Use anything you can to get you going as quickly and as painlessly as possible so you can focus on building your product and your business. That's going to be hard enough.
I say this as someone who worked with a start-up from inception to being acquired 10 years later. I was the guy building the networks and the servers and the desktops. We cobbled together our own systems from white box parts and using free software that required lots of setup and maintenance. I spent a lot of my time maintaining that stuff instead of working on our products. When we got acquired, the first thing they did was throw all that stuff out and switch to their existing systems that were all the well regarded name brands that you know. Since then, everything just works.
If you / your team know what you’re doing, you can absolutely run your own stuff without it being a nightmare. I’ve worked at a 9-digit-ARR SaaS where we ran our own servers (as in, we owned physical servers, not a VPS), ran our own networks, etc. Everything was in IaC. There were shockingly few incidents compared to literally every other SaaS I’ve worked at since.
We didn’t self-host email / file sharing, to be clear – that’s a fool’s errand due to IP reliability rules. Google Workspace is great.
I’m not trying to specifically call you out here, I’m just trying to counter the general argument that it’s impossible to do what the big providers do while still having a reliable service.
I won't say it's a good solution for everyone. If you have strict documentation needs for industry reasons, it's going to be hell for your workers. Good luck finding anything you need. Good luck getting people to follow a consistent format, or hell, even a compliant format.
... DevOps shouldn't be deploying, administering and maintaining something like NextCloud.
And honestly without any additional input this question sounds like "I worry what I would be in a position when NextCloud wouldn't be able to support the needs of 10000s users. BTW currently it's me myself and my dog in this startup".