With data dispersed across many locations that today could include on-premises, the public cloud, edge and co-location facilities, how can businesses ensure that data is in the right place at the right time, without the issue of creating multiple copies? In this episode, Chris talks to Floyd Christofferson, VP of Product Marketing at Hammerspace to learn how the company has abstracted the file system and associated metadata to address the problems of mobility and consistency. One interesting aspect of the Hammerspace approach is that it enables metadata to be extended and inherited in a way that provides workflow and other data management capabilities. These features can then be automated using policies or objectives in Hammerspace terms.
Elapsed Time: 00:00:00
- 00:00:00 – Intros
- 00:01:00 – Data has become highly distributed
- 00:02:25 – Businesses have a more diversified data landscape than they realise
- 00:04:30 – Data outlives the lifetime of hardware and applications
- 00:06:15 – Identifying owners and users of data asset can be tricky
- 00:07:51 – Shadow IT creates new silos of unconnected data
- 00:09:12 – Data mobility breaks when users start copying data
- 00:11:30 – Metadata stuck in a single piece of infrastructure causes problems
- 00:12:40 – Should an abstraction layer redirect or chunk up data on storage?
- 00:16:00 – Hammerspace data “assimilation” processes only the metadata
- 00:18:14 – Inherited metadata applies rules to data automatically
- 00:20:00 – Policy (objectives) allow data workflow management
- 00:24:00 – Hardware optimisation is one benefit of Hammerspace
- 00:25:32 – At scale, data management needs automation
- 00:26:50 – Hammerspace can dynamically scale into the public cloud
- 00:29:20 – Selective data migration lets data be placed efficiently
- 00:34:17 – The data mobility challenge is the greatest issue to hybrid cloud adoption
- 00:37:30 – Owning the file system is the key piece to data mobility
- 00:42:30 – Wrap Up
Copyright (c) 2022 Unpacked Network. Post #bgvg. Do not reproduce without permission, in part, or whole.
Podcast: Play in new window | Download
Podcast (dataunpacked): Play in new window | Download