Simplify lambda

[hamstah]

Why not use the same principle as the iterator to store a map of id to URLs inside the handler code so you don't need the separate write and iterator. You could even make it slightly less ugly by storing it in a JSON file in the lambda next to the handler.

[kkuchta]

Hmm... you'd probably have a bit of a scalability issue with that. Each url you add would make the json list take longer to load into memory. The one part of a url-shortener that you really need to be fast (redirecting from the shortened url to the long one) would grow slower linearly with the number of urls you're storing.

[hamstah]

If that becomes a concern just shard to different files. Keep each dict to 100000 entries or whatever a benchmark tells you is slow, but loading text files is probably much faster than the cold boot time for a new lambda. If you use the line number as an index you dont even need to load the whole thing.

…

On Fri, 10 May 2019, 4:06 pm Kevin Kuchta, ***@***.***> wrote: Hmm... you'd probably have a bit of a scalability issue with that. Each url you add would make the json list take longer to load into memory. The one part of a url-shortener that you really need to be fast (redirecting from the shortened url to the long one) would grow slower linearly with the number of urls you're storing. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#1 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AABN7DXVUJA56MGFPNFRNPLPUWFPZANCNFSM4HL6KTVA> .