Join Log in
Score:0
Gabor Szarnyas avatar Server

Persistence of local SSDs of EC2 instances

de flag
Gabor Szarnyas

Certain EC2 instance types (r5d, m5d, c5d, z1d, etc.) offer local disks coupled to the lifetime of the instance. For example:

With R5d instances, local NVMe-based SSDs are physically connected to the host server and provide block-level storage that is coupled to the lifetime of the R5 instance

Other instance types with a local instance store (e.g. i3) do not have this claim.

Do the persistence guarantees of disks on instances with the "physically connected", "coupled" SSDs (e.g. r5d) differ from those with regular "local storage" NVMe SSDs (e.g. i3)?

I ran a few tests on both an r5d and an i3 instance: I created a partition on the SSD, mounted it to a directory, and placed a file to this directory. When doing a reboot, I could mount the disk and the file was kept on both instances. When I stopped the instance from the EC2 console, then started it, the data was lost in both cases and I was unable to mount the disk. So, these tests did not reveal any difference.

68
0 + 0
Score:1
John Mahowald avatar Server
cn flag
John Mahowald

Your test was valuable, confirming such disks only live until the instance is stopped.

Yes, this is instance store. AWS writes a lot of marketing copy for each instance type, and sometimes reaches for synonyms. "Local" and "NVMe" emphasizes low latency. "Physically connected" or "coupled" is restating that these cannot be attached to another instance. "Coupled to the lifetime of the instance" is a polite way of saying they will be deleted on stop.

EBS also exists, persistent disks with more features. Local SSDs do not need to be anything other than scratch space, when fast SAN storage is available.

0 + 0
ธงชาติไทย
Elon Musk
I sit in a Tesla and translated this thread with Ai:

mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.