"SAN and NAS are distinct ways of deploying storage. They differ at fundamental levels of storage architecture, according to their relation to file system and block and physical addressing, but also the network or fabric by which input/output (I/O) travels. Here, we look at the key differences in performance and applicability between block and file storage, focusing on key contemporary workloads in artificial intelligence (AI), virtual machines and containerised environments."
"Where the two methods differ is how that plays out. In NAS systems, file system and storage are bundled together in the same box. Users and applications request data from, for example, a letter-designated drive, such as the C: drive commonly referred to on PCs. The request goes to the file system on the NAS box, is translated to physical addressing and the file retrieved."
"In SAN systems, the file system's work is done elsewhere. The application or database requests data, with communications translated to physical addressing outside the SAN, and data retrieved in blocks. Here lies the key difference, and as the name suggests, file storage serves up entire files while block storage delivers blocks. What are the performance implications of NAS and SAN? NAS servers carry out all their file system processing on board."
SAN and NAS represent different storage architectures that expose data at block or file granularity. NAS integrates file system and storage in the same appliance so client requests are handled by the NAS file system, translated to physical addresses, and returned as whole files. SAN exposes raw blocks with file-system translation performed by hosts or external systems, and delivers data as blocks over a fabric. NAS typically communicates via SMB or NFS over IP and incurs onboard processing overhead. SAN avoids that overhead and often uses higher-performance networks, making it better suited to demanding AI, VM and container workloads.
Read at ComputerWeekly.com
Unable to calculate read time
Collection
[
|
...
]