scale.bythebay.io: Sijie Guo Interview

my name is Chicago I'm I work for stream my name is Chicago I'm I work for stream Bo I'm one of the cofounders upstream Bo I'm one of the cofounders upstream neo and in extremely or I mostly focus neo and in extremely or I mostly focus on mastering in star H pieces and before on mastering in star H pieces and before stream deal I was their Twitter leading stream deal I was their Twitter leading the messaging infrastructure so the key the messaging infrastructure so the key that key problem are able to do it that key problem are able to do it so the key engineering problem for so the key engineering problem for messaging and storage extremely are we messaging and storage extremely are we are facing is mostly one is the scale are facing is mostly one is the scale scalability issue for messaging system scalability issue for messaging system and also the simplest simplicity the and also the simplest simplicity the unification of messaging and storage unification of messaging and storage system so the scaling issue comes from system so the scaling issue comes from our messaging system of our storage our messaging system of our storage system comes from two aspects is one is system comes from two aspects is one is the scalability of the number of the the scalability of the number of the topics in a messaging system and the topics in a messaging system and the other issue is the scalability of the other issue is the scalability of the system to be able to scale beyond number system to be able to scale beyond number of hundreds or maybe thousands of the of hundreds or maybe thousands of the nodes nodes [Music] I think that the challenging things for I think that the challenging things for the scalability issue on perhaps on the the scalability issue on perhaps on the messaging system its various especially messaging system its various especially for the arise micro service you might for the arise micro service you might need a lot of small a lot of small need a lot of small a lot of small toppings or messaging cues for micro toppings or messaging cues for micro service and the ability to scale that is service and the ability to scale that is you you need to scale the i/o all the you you need to scale the i/o all the data the stupid but you also need to data the stupid but you also need to scale the number the metadata and scale the number the metadata and scaling and metadata is the hardest scaling and metadata is the hardest problem and that is what we are looking problem and that is what we are looking for and be able to provide an actual for and be able to provide an actual scale messaging system for in the market [Music] [Music] from system engineer infrastructure from system engineer infrastructure engineer perspective we write a lot of engineer perspective we write a lot of code around I think a synchronous code around I think a synchronous processing and using a functional processing and using a functional language to do a signal processing which language to do a signal processing which simplified the implementation design simplified the implementation design make code much clearer to for maintain make code much clearer to for maintain and that is the most beneficial and that is the most beneficial advantage that we can get from advantage that we can get from functional program for for systems I think it's well organized the I think it's well organized the extensive of the impressive talks and extensive of the impressive talks and besides I work for Twitter before it's besides I work for Twitter before it's good it's good for coming back [Music]