3 ms·
"Roll your own" is hubris, unless you have the time, energy, inclination, theoretical knowledge, etc; I wouldn't try and solve solved problems if you can help i
by Cthulhu_ 3y ago
"Roll your own" is hubris, unless you have the time, energy, inclination, theoretical knowledge, etc; I wouldn't try and solve solved problems if you can help it.
But, host your own is definitely recommended IMO; a lot of the GDPR issues are resolved if you just host your own, because no data is shared to a 3rd party. Then you only need to worry about getting some approval and data retention. I'm sure data retention is a non-issue if you process raw analytics data (that can be traced back to a user) into generalized statistics, too.
- gsatic 3y ago> unless you have the time, energy, inclination, theoretical knowledge Why does anyone need all that?
- vntok 3y agoBecause setting up useful analytics for almost any definition of "useful" is not a trivial "I could do this in a WE" project.
- Pannoniae 3y agoThe better question is.... what analytics do you actually need? I don't think cross-page tracking of the user is something most websites need... Useful information is more like, what users clicked, what platform did they use, how much time they spent on the website and things like that. Those aren't that hard to just track yourself.
- quickthrower2 3y agoI reckon server render based analytics (as opposed to cookies and pixels etc. on 3rd party domains) will be better because it wont be blocked by ad blockers. If the page got rendered and sent, well I know about it unless you blocked the site entirely.
- Pannoniae 3y agoI'm much more productive since I realised I don't have to use "standard" tools. It saves me so much time. With that being said, I don't make everything myself, but just the thought that I'm not obliged to use the best-practice standard solution to the problem is liberating to me and it makes me so much more productive in actually doing stuff.
- victorNicollet 3y agoBesides, even if you use the standard tool today, five years from now this will be "the tool that was standard five years ago" unless you actively migrate every so often. There are a few domains where the standard tools remain the same for a long while (Google Analytics being one), but often the standard tooling shifts more frequently than that.
- Pannoniae 3y agoYeah, I heavily suspect the changes in "standard" stuff isn't due to there being revolutionary improvements every 4 years or so, it's just because the new stuff becomes trendy. When I research about programming languages I'm always amazed how almost every feature existed in the 70s as well, just no one bothered to use it.
- capybara_2020 3y agoI would love to know other peoples opinion on this. But I am coming around to the idea that self hosted(atleast partially) might be cheaper and better if you are a single dev/a small team. The learning curve on the tools/hosting providers out there is has become very steep. Plus the costs are unclear with a lot of cloud providers and monthly subscription charges across the services you need can quickly stack up or the prices can suddenly change. I tried deploying an app to AWS a few months back. You get a year's worth of credit when you start out. But the database I used was not covered(I did not realize all the database options were not covered). I got charged a pretty penny(Luckily it was not life shattering, but it was a shock. I hate to imagine what would have happened if I did this for a bigger app) Tried Google Analytics a few times over the last few years. But again it has become complex, so I would had to spend a ton of time learning it to even just get started. Had a few SAAS providers suddenly hike up prices or change pricing models or just shut down. We have been using dedicated servers of late. A single server seems to be able to handle multiple client apps along with hobby/test apps for a fixed price. Yes it is not as easy as putting in an email and credit card and using a service. But the price and peace of mind has been worth it. Plus we just write a few scripts to automate things. In case the work load becomes too much we can hire a person to do that and 1 person will do. Compared to having to hire a specialist for each major cloud service we use.
- viraptor 3y ago> I tried deploying an app to AWS a few months back. AWS is a toolbox you can use to construct your app deployment/hosting environment. Unless you know or want to know how to (for example) setup routing on VPCs, it may be better to go with someone actually hosting apps rather than infrastructure.
- fullstackchris 3y agoI thought (though apparently with these comments I'm in the minority) the most recent conventional wisdom has reverted to agreeing that yes, starting on self-hosted / managed servers is the best path. Seems like naysayers in here don't know how to set up a server with NGINX / SSL / whatever else you need. IMO these are _required_ experiences before trying to do it on a much more complex platform like AWS. These self managed server setup tasks are more than manageable for a small team, or even a solo dev, in a matter of a day or two. My go to 'cloud' path is just a digital ocean droplet, using docker containers to spool up whatever you need and connect it all on a docker network. A single server (2-4 cores) running quite literally ANY modern backend framework (node, go, C#) should be able to handle _thousands_ of requests per second, I'm not sure where or when this idea has disappeared, it seems like everyone automatically assumes their small SaaS or webshop needs an autoscaling kubernetes 20 rack workhorse of a server. Not the case at all!!!! Scale when you need to - if you're getting the kind of traffic where you need to, by then you won't need to worry about the added cost to do the actual scaling / upgrading. Sorry for the winded / ranty answer, I've done this like 20+ times at this point and always had to battle against the "let's put it on AWS with kubectl and 2349023 redundant instances!", when in the long run it was never needed...