On Tue, Nov 23, 2021 at 5:04 PM Alasdair G Kergon <agk@xxxxxxxxxx> wrote: > > On Wed, Nov 24, 2021 at 12:38:31AM +0000, Alasdair G Kergon wrote: > > There seems to be a general industry movement towards edge, attestation, > > trusted boot, nested systems, confidential computing, containers, > > <insert your favourite buzzwords> etc. and I think this type of > > device-mapper target might play a useful role as one of the low-level > > components involved in building up support for some of those scenarios. > > (Just as we recently began adding support for attestation.) > > For example, I think we've already made good progress towards > standardising the industry around dm-verity. (At least, I keep > on encountering it being used in various different places, rather > than people attempting to develop their own version.) We extended > that concept with dm-integrity but I'm less sure about how widely > that has been taken up so far. > > For edge systems needing unattended remote reliable updates, our > high-level challenge here is: Can we together develop a decent and > efficient solution for OTA updates that the industry will be happy to > standardise around? With OTA, it is challenging to have one general COW format - for instance what is good for the Android ecosystem may not be useful for the enterprise world. For ex: Most of the space savings in Android comes from COPY operation i.e for an incremental OTA, we would have metadata which states: COPY BLOCK X -> BLOCK Y There is no compression involved with these operations. Compression is only when a block is "replaced". All these are too specific to the Android ecosystem. I don't want to hijack this conversation here as the current patch is more specific; however, since you mentioned about the standardization of OTA updates, I want to bring up the aspect of block device in user space which gives you the flexibility for users to define COW format as required. There are already couple of use cases which we are aware off (one being android OTA) and there are few open source work such as [1] . [1] https://github.com/albertofaria/bdus > Alasdair > -- dm-devel mailing list dm-devel@xxxxxxxxxx https://listman.redhat.com/mailman/listinfo/dm-devel