scale.bythebay.io: Ben Edwards Interview

I'm Ben Edwards I currently work at I'm Ben Edwards I currently work at Starbucks as as a consultant probably Starbucks as as a consultant probably the the biggest one that we've had is we the the biggest one that we've had is we you know transition a lot of people from you know transition a lot of people from one platform very different platform the one platform very different platform the Microsoft on to on to Scala which is Microsoft on to on to Scala which is very different different ecosystem very different different ecosystem different culture I think especially different culture I think especially because scarless that I think still because scarless that I think still trying to figure out what good Scala trying to figure out what good Scala looks like that's pretty challenging for looks like that's pretty challenging for people to be kind of like thrust into I think what often surprised people is is think what often surprised people is is you know we have great hardware we have you know we have great hardware we have great frameworks and it is amazing how great frameworks and it is amazing how seemingly not not particularly seemingly not not particularly significant things can really can really significant things can really can really take take an application out so one that take take an application out so one that springs to mind is we were seeing springs to mind is we were seeing massive connection pileups huge latency massive connection pileups huge latency problem with some of that web front-end problem with some of that web front-end and it just turned out that we were too and it just turned out that we were too aggressive in tearing down connection aggressive in tearing down connection pools and that akka has a 500 pools and that akka has a 500 millisecond lead time on spinning a new millisecond lead time on spinning a new pool back up which if you've got you pool back up which if you've got you know a thousand at the multiple know a thousand at the multiple thousands of connections like piling in thousands of connections like piling in that that can become that can become a that that can become that can become a huge problem it took us a long time like huge problem it took us a long time like far longer than you you think it should far longer than you you think it should have to figure that out and that's not have to figure that out and that's not you know like connection pooling is not you know like connection pooling is not like overly aggressive tear down a like overly aggressive tear down a connection is not something when saul connection is not something when saul says like scale you know you think i says like scale you know you think i need more memory i need more cpu need more memory i need more cpu whatever you don't think i'll i needed whatever you don't think i'll i needed to tweak this value in a config to tweak this value in a config somewhere that i'd never look at I think I one of the things that has I think I one of the things that has been like an interesting switch that's been like an interesting switch that's happened recently is that yeah it used happened recently is that yeah it used to be that you had to protect disks at to be that you had to protect disks at all costs from access you know like all costs from access you know like there are you there are you almost the entire tenor of database almost the entire tenor of database research has been you know how do you research has been you know how do you make sure that you do the minimum make sure that you do the minimum distance block of data and I think I distance block of data and I think I think that's really been flipped on its think that's really been flipped on its head by the current gen of data center head by the current gen of data center class SSDs and actually that the scale class SSDs and actually that the scale problem is gonna actually it's it's been problem is gonna actually it's it's been turned on its heads rather than feeding turned on its heads rather than feeding CPUs its feeding its feeding storage and CPUs its feeding its feeding storage and I think that we'll yeah it's going to I think that we'll yeah it's going to necessary a whole a whole bunch of of necessary a whole a whole bunch of of new ways of doing things if we want to new ways of doing things if we want to take advantage of the hardware that we take advantage of the hardware that we go [Music] [Music] the endeavor of programming is is is the endeavor of programming is is is writing is writing things that can be writing is writing things that can be clearly understood by people to be clearly understood by people to be executed by machines that to you know to executed by machines that to you know to provide to provide value and functional provide to provide value and functional programming allows you to decompose programming allows you to decompose decompose those that you know that the decompose those that you know that the big problem into into smaller problems big problem into into smaller problems with far greater clarity than I think with far greater clarity than I think any other any other paradigm that I've any other any other paradigm that I've tried over the years favorite thing tried over the years favorite thing about scale by the bay honestly this my about scale by the bay honestly this my first time in San Francisco and it's first time in San Francisco and it's it's it's just been enjoying soaking up it's it's just been enjoying soaking up the you know kind of the rich they were the you know kind of the rich they were out lineage you know like this is you out lineage you know like this is you know this is a an area filled with tech know this is a an area filled with tech giants and it's kind of fun to be to be giants and it's kind of fun to be to be around them like really fun [Music] you