Implementing a Large Scale Document Storage Service(blog.thegrizzlylabs.com)
blog.thegrizzlylabs.com
Implementing a Large Scale Document Storage Service
http://blog.thegrizzlylabs.com/2014/12/implementing-large-scale-document.html
3 comments
How do you plan on implementing a local export? As you mentioned, cloud services aren't suitable for all users or all use cases.
Numbers are quite impressive: Millions of monthly active users. How do you plan to handle the operations once this is running? If I read correctly you are only two?
Thanks! And great questions.
We are currently two but we will be three starting next year. That will definitely help!
We want the backend to be very transparent for the user. This means that if there is any downtime on the server, you will still be able to use the app, scan documents, create PDFs, export them etc.
However, it has always been our philosophy to take our time to reduce the operational load as much as possible since our time is so precious. We have quite some experience with that having run the business as a side-project for a while before going full time on it. Running a boostrapped, profitable company also helps in that you don't have any external pressure.
We are currently two but we will be three starting next year. That will definitely help!
We want the backend to be very transparent for the user. This means that if there is any downtime on the server, you will still be able to use the app, scan documents, create PDFs, export them etc.
However, it has always been our philosophy to take our time to reduce the operational load as much as possible since our time is so precious. We have quite some experience with that having run the business as a side-project for a while before going full time on it. Running a boostrapped, profitable company also helps in that you don't have any external pressure.
Good read! That is an ambitious project. Very interesting to note that the implementation didn't take more than a few days.
Thanks! Note that it's just a prototype implementation. We wouldn't want to put it like that in the hands of users, but it's definitely good enough to play with and get a sense of the final result.
In terms of scalability, is SQL an option for you or do you plan to go NoSQL?
SQL should be fine with sharding. Evernote has actually a great article on how they sped up their sync by rethinking the sharding. http://blog.evernote.com/tech/2014/01/28/synchronization-spe...