GopherCon 2014

Desarrollando un sistema de analítica con MongoDB y Go

William Kennedy  · 

Transcripción

Extracto de la transcripción automática del vídeo realizada por YouTube.

so a couple of years ago I got hired the ardent studios got hired by financial apps to build a pfm of personal finance management application and that is called power wallet that's been around for a few years and what power wallet it's similar to meant

if you guys know meant so people can come on they log in they download their banking transactions their credit cards they can do their bills and on this system they can categorize all that budget it get bill reminders all of that so they're managing their

financial lives on this system and we realized pretty quickly that we had all this incredible consumer data right we knew what these people we're at least they were buying the things they were buying so to put that into perspective we shot a short video

give you an idea of a user mix so funny enough you can't film in these stores and we got kicked out of Walmart so we wanted to do a scene where he was on the conveyor belt and then we paid and did got kicked out learn the hard way but I'm here so I

didn't go to jail all right so that's a typical consumer for power wallet right they for that day he went out and he purchased a bunch of things and so I have we have this database slide will go sorry one more so I have all of this data that we're

gathering in the application we have in read all the user data and this is typical to most of the applications that are out there right registration we can bring in social you saw some of the financial data that we're pulling in the application itself

generates a bunch of data impressions alerts budget transactions things that users were doing and we wanted to make all of that data that's in yellow and red actionable it was cool to have it but not just for running reports we want to make it action we

want to be able to do things specifically if I know the Gopher is online and I'm looking at the fact that he's now has a girlfriend I want to be able to target offers or advice or things to the go for while he's online specific to what we know

about him as opposed to someone else's online we want to make this data actionable now I don't know about you but my client if financial apps the guy is an idea guy you know the idea guys the guys at every time you you walk in on Monday it's been

thinking about it all weekend and wants to do something like in the next three hours right but you can't do and you're tired of saying no to him right so guys got great ideas and the first version of the system that I built was using a sequel server

and this caused a lot of problems for us because he would come in on Monday and say bill I found this new feed it's got all these great offers I want to start giving them to everybody and I'm like Bob it sounds good but I got to build a table and after

I build a table I got to get that data in there then I got to make sure it's attached to the rest of the system and by the time I do that then I got to build some code and then we've got to get it in there so he'd be like how long is they gonna

take three four weeks maybe four lucky two if he told me to go do it by the time I did it I came in on Monday and we're all ready and guess what Bill and I found this great feed I want to use it right now or what about the one I just coated forget about

that one this one's better so it wasn't working we needed to find a way to be able to bring in these data feeds this inventory to allow others to bring in their own feeds even if they weren't public maybe like advice feeds internally I've been

able to do that almost real-time without having to build the system without needing really a programmer to get involved so we sat down and we said if we had a system that would allow us to bring in dynamic feeds at the time Bob found them to be able to bring

them into the system and use them but use them in the format that they're in so a feed like a daily deal feed is very different than a credit-card feed right so daily do you feel will have their deal ID you'll have a description you have a title but

when you get to a credit card feed now you're talking about interest rates APRs a whole set of fields that were really impossible to map into the sequel server that dynamically and be able to query against it but when we looked at Mongo we realized very

quickly we could build a collection on the fly we could put the data that we're seeing in the format that it's in all of the fields we don't have to lose any of it and put it in there and it would be usable right away so we had everything there

[ ... ]

Nota: se han omitido las otras 2.235 palabras de la transcripción completa para cumplir con las normas de «uso razonable» de YouTube.