In this podcast, recorded on 6th September 2017, Chris, Martin and Gavin discuss the subject of disaggregated storage. With the move to NVMe, traditional dual controller architectures are hamstrung by the need for all data to go via the processor(s). Disaggregation takes the component parts and makes them scalable, while removing the controller bottleneck. The guys discuss why these solutions are needed, who the players are in the market and what we can expect to see as this section of the storage industry develops.
Elapsed time: 00:21:02
Timeline
- 00:00:00 – Introductions – What do we mean by disaggregation?
- 00:01:10 – Legacy architectures and dual controllers
- 00:03:57 – NVMe or not – it’s irrelevant
- 00:04:28 – Violin is back!
- 00:05:00 – Lost NVMe performance
- 00:06:00 – Thinking about tomorrow’s workloads and IoT
- 00:07:00 – A Raspberry Pi Diversion
- 00:08:28 – The old Mac versus PC debate again
- 00:09:00 – Vendor roundup
- 00:10:13 – Much of this looks like HCI – is the problem the interconnect?
- 00:11:30 – The Storage admin is dead
- 00:11:53 – Howard Marks, Pure and Nutanix
- 00:13:20 – X-IO and interconnects
- 00:14:51 – OK, let’s summarise the architectures
- 00:17:30 – Pure Storage – disaggregated or not?
- 00:19:05 – Wrap Up
Companies/Products Mentioned in this Podcast
Copan Systems, Cohesity, Coho Data, Pure Storage, Violin Systems, Excelero, Micron, Kaminario, Datrium, EMC DSSD, VMware Virtual SAN, X-IO, SymbolicIO, E8 Storage, Vexata, EMC Symmetrix.
Copyright (c) 2016-2017 Storage Unpacked. No reproduction or re-use without permission.
Podcast: Play in new window | Download