Alki, or how we learned to stop worrying and love cold metadata
In this post, we introduce Alki, a new cost efficient petabyte-scale metadata store designed for storing cold, or infrequently accessed, metadata. We’ll discuss the motivations behind building it, its architecture, and various aspects of how we were able to rapidly prototype and then productionize...
Hasnain says:
Interesting technical read.
“Today, Alki serves roughly 350 TB worth of user data (pre-replication and not counting indexes), at about 1/6 the cost of Edgestore per GB per year. This was made possible by leveraging the inexpensive storage costs of cold blob storage.”
Posted on 2020-12-26T23:59:13+0000