Results
3
comments of
jasonxie
公司内部这几种都有做过对比。第二种随模型迭代演进，后期维护成本非常高。还是看好第三种，希望大厂能开源共建。
We have similiar scenario. We're using EKS+Fargate to deploy spark job, which will be scaled to Fargate. The spark job requires more than 50GB local disk to calculation, and is...