This chapter is from the book
This chapter is from the book
This chapter is from the book
Foundation Topics
Depending on the way you deliver content, you can classify your data in to three major categories:
Static assets: Any type of content that cannot be opened from the storage environment directly (via block access) but that must rather be transferred locally (downloaded) and only then opened can be considered a static asset. These assets can be any types of files, such as text, videos, images, archives, packages, and other data blobs that reside on a web server and are accessed only via the web service. Because static assets are delivered across the network, the access times for data ranges from tens of milliseconds to seconds to minutes and even to hours (for very large files over slow links).
Dynamic assets: These assets are any type of content that is opened and used from the storage environment directly via block-level access. These can be any types of files that are consumed by services to maintain data records or state, such as databases, log files, and executable files. Usually dynamic assets are opened by a certain process and start writing the records. A dynamic asset is not accessible to other processes on the file system and is accessible only through a service such as a database service or an API. Because static assets can be accessed directly through block access, you usually see the latencies being measured in a few milliseconds to a few seconds.
In-memory assets: An in-memory asset is any type of content that is loaded into memory and used by one or multiple processes within a server directly via access to the memory. Most commonly these assets are any kind of block-level data that is cached for performance, in-memory databases, and other data that needs to be accessed with the lowest possible latency. As the cost of memory per gigabyte is much higher than the cost of disks, it is common to store only “hot data” or caches in memory to increase the performance of traditional disk-based systems. Because in-memory assets are ready to serve, the latencies for delivering data can be decreased down to microseconds.