Tahoe on AWS

[Greg Troxel]

Lukas Pirl <tahoe-dev at lukas-pirl.de> writes:

> IMHO, physical disks and cloud IaaS are on very different levels of
> abstraction. Your resources may get migrated around in the data center
> anyways, don't they? Hence, there is probably not /the/ physical disk a
> resource lives on.

Certainly.  But a user wants to have non-correlated failures that they
can understand, and the interests of a cloud provider and a user don't
line up here.

> I'd rather understand Tahoe as a way to achieve classical n-modular
> redundancy (yes, *probably* not dying all together…) and especially to
> be able to achieve geo and provider redundancy.

Sure, that's a reasonable approach, which can get you at best
probalistic results.

I was just wondering if anyone had dealt with what cloud providers
provide for data.  Surely anyone trying to get redundancy would have the
same issues as tahoe.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 180 bytes
Desc: not available
URL: <http://tahoe-lafs.org/pipermail/tahoe-dev/attachments/20141221/2f444f8d/attachment.pgp>