Cloud Storage is many orders of magnitude faster. Read speed will be determined by the client's network speed (Internet connection) and TCP/IP stack configuration. This means using Google's CDN, which caches bucket data around the world at points-of-presence that are closer to the client.
To improve locality, bring the data closer to the client. This reduces the unpredictability of the Internet. Network performance and locality.įor network performance, ensure data travels from the bucket to the user over Google's premium tier network. The key to fast performance for the client is two-fold. Selecting a bucket location will have more of an impact link. You can also configure dual-region storage buckets, which are primarily used for replicating data. There are exceptions such as object name hotspots (sequential object names), but this would also affect your multiple bucket strategy. I do not believe a single load balancer would be able to overload a storage bucket. The downside is increased complexity in your deployments.Ĭloud Storage is already very fast and can handle global access. I have not benchmarked using multiple buckets, but I do not think there will be any benefit.
0 kommentar(er)
