I wouldn't call hourly snapshots backed up to S3, monitoring and replication "trivial." They aren't difficult, but most people who don't already do this will probably need to spend hours on reading docs, setting it up and testing. Good luck even finding the time if you already have a loaded schedule of development.
"beyond trivial" - perhaps from your perspective :)
These days the vast majority of people who with the job title "developer" or "software engineer" only know one or two things. Given that the number of software engineers consistently grows 20% year-on-year, it means that at any given time the median developer has just 2 years of experience. It's enough to learn a narrow set of tools (bootcamps or self-taught) to become productive and start your journey. DevOps-y stuff is often not one of those things.
So tools that simplify the unfamiliar parts tend to be quite useful
I am getting frustrated by the increase of hn articles about "self-hosting" yet the software described is reliant on a cloud service. To me, "self-hosting" means I can use it on my own computer, without a connection to the internet.
Did you read what the service does? You probably don't need hourly snapshots backed up to S3, monitoring and replication if you are hosting MySQL on your local system. This looks to be more like setting up a managed service for production without paying the hosting provider for the extras.
Right, but using Lightsail as an example to keep things simple, you're picking between scaling at 3x the cost of the server vs db service or a set $5 per database.
I would still trust the AWS service with something mission critical though. Hippobase might be good for stuff which isn't mission critical but otherwise wouldn't have anything the service offers because you don't want to throw the resources at.
The use case we had in mind initially was that teams often set up their production using Heroku db or RDS (naturally, because it's easy and reliable), and then when they need more environments they naturally do the same. But reliability often isn't that critical for dev environments. You lose your DB, so what? Just make a new one.
Then we realised that we could actually near production-grade reliability by doing backups, snapshots etc. So the choice is up to the user
yeah I agree, it's a bit confusing. I meant "host in your cloud account" as opposed to "use a managed service". The vast majority of SaaS seems to be built cloud-first, it's quite easy to forget that other ways of hosting still exist.
Running a prod db on EC2 isn’t a trivial undertaking- I see the value in a service that can affordably recreate the managed DB experience. I have lots of questions about how and what I can/can’t do with the system, like install foreign data wrappers, etc that RDS disallows. Is binlog/PITR backup expected to be available at some point?
We're weighing pros and cons of opening up direct access to underlying instances for fine tuning. The pros are obviously more control, the cons - loss of standardisation and harder to support as a consequence
Sharp eye :) Yes we are. Our dirty secret is that on the backend all of those are using the same engine with minimal modifications that is generating Terraform automatically and running it for you. We are just iterating on use cases.
37 comments
[ 4.7 ms ] story [ 89.8 ms ] threadEven if this takes an hour of your time saving that hour pays for like a year of Hippobase.
So whether or not this is valuable is subjective, depends on what you see as good use of your time.
AWS is unnecessarily complicated for a common set of use cases. Layers on top that provide simplicity, and more importantly, confidence is nice.
So tools that simplify the unfamiliar parts tend to be quite useful
I'm not going to call it differently just beceause some self-appointed OSS hardliner wants everyone to use communist licenses.
I would still trust the AWS service with something mission critical though. Hippobase might be good for stuff which isn't mission critical but otherwise wouldn't have anything the service offers because you don't want to throw the resources at.
Then we realised that we could actually near production-grade reliability by doing backups, snapshots etc. So the choice is up to the user
They seem to be highjacking the term
(saved me a click through!)
Maybe the mods can change the title? I mean the first word is easily.. and then to learn aws is required? haha
maybe hippobase - deploy aws hosted database systems
It’s interesting to see people building improved UX workflows on top of AWS.