Skrift
Helpline software for anonymous aid via SMS
Long story short, the application has been running horribly the last days. And yesterday everything culminated in a database process thrashing wildly - effectively taking the application offline for half an hour in the middle of our must active hours 😭
This has bascially caused me to lose faith in my hosting provider, but I just can't see myself migrating elsewhere. So for now I'll have to settle for slow deploys and bad performance, which I find so hard to accept.
February has come to an end, and we managed to get a fully functional, working SMS helpline up and running.
Not just that, we even migrated all the existing users and messages from the customers old solution to this new platform.
And not just that, they are already using the new solution for all their users.
And not just that, we pulled if off a full week before deadline! 😅
Skrift has now been running in production with an actual customer with 100+ users for more than a week.
More than 200 young people have reached out anonymously via the mobile phone, and we have processed almost 1500 incoming SMS messages - and sent even more!
It's such a great feeling seeing a product take off like this. 😍
When I was approached by an NGO with a challenge, I was sceptical. Could I really build a SaaS product to replace their existing, tailormade SMS helpline solution in a month?
But this was a prime opportunity:
- I'd get a SaaS product with the first customer already signed up.
- There were other products like this out there, but few with this specific niche.
- This was a chance to help out an organization that wanted to do good in the world.
So I said yes...
3 Comments
Hey @mentalizer, sorry to hear that.
As far as I understood you are managing the database by yourself, is that accurate? If so, are you open to consider providers with managed databases.
If not, maybe running two db instances in separate machines, where the second db instance is synced with the primary node, and stays in stand-by for cases like the one you faced.
Oh no, this was a managed database, which basically made it even harder to figure out/resolve.
Thanks for the suggestion, I'm just not sure my MRR - nor workload - calls for standby replicas yet, though :)
While migrating might be something to plan for considering the issues. Take a look at managed nosql dbs. When I worked on inbound messaging platform before we used AWS dynamodb. Even at millions of relatively small records you could get away with a bill under $10.