Technology:Unpacked

This week, Chris and Martin chat to Kevin Deierling, SVP of Marketing for Networking Products at NVIDIA. SmartNICs and DPUs (Data Processing Units) are starting to become mainstream as application use-cases such as AI and analytics drive a need for greater data throughput and performance. Kevin explains the design and thinking behind BlueField, NVIDIA’s family of DPU products that combine offloaded network, storage and security functionality.

Elapsed Time: 00:47:36

Timeline

* 00:00:00 – Intros* 00:01:30 – We’re not a networking podcast!* 00:02:15 – Why will we need DPUs and SmartNICs?* 00:04:30 – Von Neumann is diverging* 00:06:50 – What is the BlueField architecture?* 00:08:45 – A DPU could act as a storage array controller* 00:11:30 – Storage DPUs make devices appear local* 00:13:00 – DPUs enable efficient bare-metal server deployments* 00:15:00 – Storage, networking & security use around 30% of traditional cores* 00:16:00 – Does a DPU represent better or worse performance than CPU?* 00:20:15 – NVIDIA DPUs emulate existing devices, reducing application changes* 00:25:00 – Where does the outboard management take place?* 00:26:00 – DOCA is the application framework for DPUs* 00:28:00 – BlueField 2X combines GPU and DPU on the same card* 00:31:40 – DPUs enable the real-time nature of data processing* 00:35:40 – Mainframe reference!* 00:38:00 – Where will initial adoption take place?* 00:40:00 – How does the use of SmartNICs affect TCO?* 00:45:10 – The future is 1000x improvement with BlueField 4* 00:46:00 – Wrap Up

Related Podcasts & Blogs

* #96 – Discussing SmartNICs and Storage with Rob Davis from Mellanox* #177 – SmartNICs and Project Monterey* #180 – SmartNICs – Pliops Storage Processor* VMware Project Monterey – First Impressions* Fixing the x86 Problem

Copyright (c) 2016-2021 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #ky1r.

This week, Chris and Martin debate the success of software-defined storage or SDS. Over the past 10-15 years, storage systems have evolved into a wide range of software solutions, appliances and cloud-based products. What has driven this move to focus entirely on software and what will the next stage of evolution be? Are we in a position where everything is based on software, in some form?

This discussion also touches on consumption models. Has SDS provided the catalyst for vendors to offer better subscription licences and consumption models? Is the writing on the wall for the traditional storage array?

Elapsed Time: 00:40:51

Timeline

* 00:00:00 – Intros* 00:00:55 – Intel H10 recap* 00:01:43 – Siri interrupts!* 00:03:34 – Has SDS been a quiet success?* 00:05:20 – The five phases of SDS evolution* 00:07:08 – What compatibility issues exist?* 00:07:40 – Phase 2 – bespoke SDS solutions* 00:10:14 – Phase 4 – Infrastructure abstraction* 00:11:50 – Phase 5 – Partnered solutions* 00:13:00 – Is it possible to defined SDS today?* 00:14:21 – PowerStore is SDS (not sold like that)* 00:15:20 – Some SDS packaging is a commercial decision* 00:17:00 – Many storage solutions will run as virtual & cloud instances* 00:19:00 – SDS has been around for many years in different forms* 00:21:30 – Is high-end storage (like PowerMax) SDS?* 00:23:10 – Where is SDS challenged?* 00:26:01 – Is Open Source the biggest challenger to commercial storage?* 00:29:39 – Where will SDS go in the future? * 00:30:04 – Will SmartNICs affect the development of SDS?* 00:32:30 – Do purchasing models need to change for SDS?* 00:36:00 – Vendors may need to rewrite their software* 00:37:00 – Does SDS need to be more “data aware”?* 00:39:00 – Wrap Up

Related Podcasts & Blogs

* #156 – Introduction to Hammerspace* #148 – Unpacking HPE’s Storage Strategy* #141 – Building Storage Systems of the Future* Storage Predictions for 2021 and Beyond (Part III – SDS)* Will TCO Drive Software Defined Storage?* ScaleIO Becomes Software Defined on Hardware

Vendors & products referenced in this podcast:

Nexenta, OpenFiler, FreeNAS/TrueNAS, VMware, NetApp, Isilon, HPE (Primera & 3PAR), Dell PowerStore, Pure StorageCeph, OpenEBS, Rook, Weka, StorageOS, Portworx, Hedvig, Hammerspace, StoreONE, Backblaze, Wasabi, AWS S3,LizardFS, Gluster, GPFS, Spectrum Scale, SVC, PowerMax, Hitachi, Nimble Storage, Red Hat, NVIDIA, Komprise, Qumulo

Copyright (c) 2016-2021 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #ugbm.

This week, Chris and Martin dig deeper into the adoption of Intel Optane technology in both the consumer and enterprise markets. Optane is a new persistent memory technology that offers greater performance and lower latency compared to NAND flash, but currently is more expensive and offers smaller capacity devices. As Optane becomes more prevalent in both markets, how is it being adopted and what needs to change to increase adoption levels?

Note: Since recording, Chris has tried out the Intel H10 device and the results were disappointing. Read more here.

Elapsed Time: 00:31:49

Timeline

* 00:00:00 – Intros* 00:01:20 – Is Optane poised for data centre and consumer dominance?* 00:02:20 – What are the Optane consumption forms?* 00:03:38 – How can Optane be used at home?* 00:06:10 – Could Optane offer “instant on”?* 00:07:05 – How does H10 manage the Optane/QLC cache?* 00:08:10 – Optane offers benefits for home video editing* 00:10:22 – Optane is 16x more expensive than QLC* 00:10:40 – Could FuzeDrive be a better solution to build a hybrid drive?* 00:12:00 – A home PC can offer greater performance than 10-year old arrays* 00:13:45 – Will we use M.2 in the data centre (or U.2)?* 00:14:00 – Where is Optane being used in the data centre?* 00:16:25 – Operating System support for Optane is already here* 00:17:44 – Will Open Source databases be first to exploit Optane* 00:19:28 – Optane is optimised for byte-level addressing* 00:22:08 – Optane adoption is still tactical, like SSDs* 00:22:50 – How would containers and Optane work?* 00:26:40 – Intel doesn’t appear to sell raw Optane chips* 00:28:35 – Would Martin consider Optane for home?* 00:30:40 – Wrap Up

Vendors mentioned in this podcast: StorONE, VAST Data, Pure Storage, 3PAR/HPE, Vexata, MemVerge, IBM, Western Digital, Infinidat, StorageOS, Portworx.

Related Podcasts & Blogs

* Intel H10 Hybrid Optane M.2 SSD is a Disappointment* When Will Optane SSDs Replace NAND Flash?* What is Intel Optane?* FlashArray//X Gets Optane Acceleration with DirectMemory* HPE Demos 3PAR with Intel Optane (3D-XPoint)* #174 – Introduction to Zoned Storage with Phil Bullinger* #171 – Exploiting Persistent Memory with MemVerge* #184 – MCAS – Memory Centric Active Storage

Copyright (c) 2016-2021 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #kehf.

This week, Martin and the two Chris’s conclude their end of year discussions. This episode covers “Turkeys” of 2020, Missing in Action companies and those technologies that should be put into Storage Room 101. These are the products we never want to see darken our doors again. Finally, the team finish on a discussion of what to expect in 2021.

And for your delight, here are some Hitachi Mr T videos:

* https://www.youtube.com/watch?v=tW1S2tsxVHg* https://www.youtube.com/watch?v=boZyHDJ5qCs

Elapsed Time: 00:49:54

Timeline

* 00:00:00 – Intros* 00:01:00 – What are the Turkeys of the Year?* 00:01:40 – Is secondary data re-use a turkey?* 00:05:00 – Is Snowflake offering real data analytics?* 00:07:50 – We explain ETL processes* 00:12:00 – Teradata did analytics 30 years ago* 00:12:40 – Internet of Things – another turkey?* 00:15:00 – We said goodbye to Stellus, Datrium, Violin* 00:16:40 – Storage buyers’ remorse?* 00:19:00 – Is Optane aimed at being the AMD killer?* 00:24:00 – Missing in action – Formulus Black?* 00:26:00 – Whatever happened to Memristors?* 00:27:00 – It’s time for Storage Room 101!* 00:30:00 – Viruses create holes in floppy disks!* 00:32:00 – What is the point of DNA storage?* 00:36:50 – Johnny Mnemonic!* 00:37:00 – Optical drives – never again!* 00:39:20 – Who still uses bare metal recovery?* 00:41:10 – What will 2021 look like?* 00:43:00 – HAMR disks will appear in 2021* 00:46:00 – PLC NAND will appear in 2021* 00:48:30 – What won’t we see the end of in 2021?* 00:49:00 – Wrap Up

Vendors mentioned in this podcast – Delphix, Actifio, Catalogic, Cohesity, Snowflake, Rubrik, HYCU, IBM, Veritas, Commvault, Stellus Technologies, Datrium, Violin Systems, StorCentric, Vexata, Retrospect, Kaseya, DDN, Tintri, EMC, Mozy, Iomega, NetApp, Storwize, Pure Storage, Intel, AMD, Micron, Formulus Black, InfiniteIO, Hammerspace

Related Podcasts & Blogs

* #186 – End of the Year Show 2020 – Part One

This week, Martin and the two Chris’s look back at 2020 for the highlights and lowlights of the storage industry. What’s been hyped and what’s been a success? Will we be travelling in 2021 and what impact has the lockdown had on 2020 sales? This episode is the first of a two parter over the next couple of weeks before we close down for the Christmas period.

Elapsed Time: 00:43:10

Timeline

* 00:00:00 – Intros * 00:00:56 – Chris has not been to China * 00:01:55 – Chris M is on a health kick! * 00:02:48 – Gartner analysts haven’t moved * 00:04:05 – Chris M sees all-flash arrays as a disappointment * 00:05:20 – Sales has had a tough year * 00:07:00 – What about conferences? * 00:08:11 – Chris would consider ExCeL – if it wasn’t closed * 00:10:15 – Is it the end for influencers and big conferences? * 00:12:00 – Just call me Mr Chips! * 00:13:00 – The cloud is great – apart from AWS us-east-1! * 00:14:30 – What happened with fund raising in 2020? * 00:17:30 – How was Zerto funded this year? * 00:21:30 – How did the Kasten/Veeam acquisition work? * 00:24:50 – Veeam must be an IPO target at some point * 00:27:00 – NASUNI could be an IPO target too * 00:28:00 – Jensen likes leather jackets and no socks * 00:29:00 – Datrium lives on in VMware 00:29:35 – CloudJumper or CloudSweater? * 00:31:50 – Whatever happened to Stellus? * 00:33:00 – Remember Whiptail? * 00:34:00 – What tech has been over-hyped in 2020? * 00:41:00 – NVMe Ethernet drives are coming * 00:42:40 – Wrap Up

Related Podcasts & Blogs

* #146 – Coronavirus and Impacts on the Technology Industry* #149 – Coronavirus 2.0

Copyright (c) 2016-2020 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #9d7j.

This week Chris and Martin talk to Rob Walters, GM for Pure-as-a-Service at Pure Storage. The company has recently announced upgrades and expansion of the “as-a-service” model, formerly known as Evergreen storage. The name change and the widening of the portfolio represents an opportunity for customers to see transparent pricing and a greater depth of service offerings.

In this sponsored episode, we review the origins of the service and look at how that has evolved into the capability for customers to consume Pure block and file products as true services. We discuss the challenges in implementation, that are mitigated through customer service and data collections from Pure1.

You can learn more about Pure-as-a-Service here – https://www.purestorage.com/products/staas/pure-as-a-service.html and more on the announcement with the press release here – https://www.purestorage.com/company/newsroom/press-releases/pure-expands-pure-as-a-service-offerings.html

Architecting IT and Storage Unpacked coverage of Pure Storage is available through the Pure Storage Microsite.

Elapsed Time: 00:35:20

Timeline

* 00:00:00 – Intros * 00:01:50 – What is Pure as a Service? * 00:03:00 – What is the demand for storage as a service? * 00:04:55 – How much of “as a service” is being is being driven by Public Cloud? * 00:06:55 – Is opex rather than capex driving service demand? * 00:08:20 – The initial offering was Evergreen Storage Service (ES2) 00:10:00 – What are the details of the offering? * 00:11:30 – Minimum 50TiB and 12 months * 00:12:40 – Customers need to evaluate and understand their requirements * 00:15:00 – PaaS aims to “rightsize” customer requirements * 00:18:00 – Pure1 is an essential component to delivering service offerings * 00:19:44 – How are mixed service levels delivered? * 00:21:30 – How do operational processes change to manage the hardware? * 00:24:20 – Are Pure publishing pricing? * 00:26:30 – How hard is it to find out the details of what a vendor charges for storage? * 00:28:00 – What happens if SLA guarantees aren’t met? * 00:31:00 – Datacentre deployment needs planning * 00:34:00 – Wrap Up

Related Podcasts & Blogs

* Pure Storage Microsite* Democratising Enterprise Storage Pricing* #118 – Pure Accelerate 2019 with Patrick Smith

Rob’s Bio

Rob Walters joined Pure in 2019 to lead the

new Storage as a Service (StaaS) business

unit, responsible for defining strategy and

establishing a high-growth new business

portfolio.

In a career spanning more than 20 years,

Rob has worked in Enterprise IT and built

products and services for both SMB and

Enterprise that span systems and application

management, managed services, and the

public cloud.

Prior to Pure, Rob spent four years at Amazon as General Manager of Search Services for AWS, responsible for the introduction and scaling of Amazon Elasticsearch Service, as well as Amazon CloudSearch. In this role,

This week, Chris and Martin talk to Daniel Waddington, Principal Research Staff Member in the Global Storage Research Group at IBM Research. Daniel is developing MCAS – Memory Centric Active Storage – a high-performance key-value store designed for persistent memory. The aim is to create a platform that can be used to bring storage and compute closer together.

Persistent memory has been a regular topic on Storage Unpacked (see related podcasts below) and we’re always interested to understand exactly how it can be fully exploited in the enterprise. PMEM offers byte-level addressing and is particularly suitable for sparse workloads. It’s hoped that MCAS can fully utilise the unique nature of persistent memory and find uses in traditional storage platforms and the public cloud.

To learn more, check out the MCAS repository on GitHub. If you’re interested in learning more about the Computer History Museum we discussed at the top of the recording – here’s their website.

Elapsed Time: 00:36:41

Timeline

* 00:00:00 – Intros * 00:02:20 – How tall is a 3380? * 00:03:20 – What is MCAS? * 00:05:20 – MCAS uses persistent memory DIMMs using Intel Optane * 00:06:30 – Optane offers much lower I/O granularity – byte addressable * 00:07:42 – Intel Optane provides much more deterministic I/O latency * 00:08:38 – Has Optane been fully exploited in its current form? * 00:11:00 – Could Arm be used with persistent memory? * 00:11:34 – MCAS – deeper dive * 00:12:30 – MCAS will enable “push down” operations * 00:14:15 – MCAS could be used to help eliminate battery backup * 00:19:00 – How will CXL and other remote memory technologies help? * 00:21:30 – Computing architectures are evolving * 00:23:00 – Disaggregation is enabled via technologies like CXL * 00:24:33 – How long will it be before MCAS is available? * 00:26:23 – Exactly how would “push down” work? * 00:31:00 – MCAS will enable technologies like Computational Storage * 00:34:02 – Wrap Up

Related Podcasts & Blogs

* #171 – Exploiting Persistent Memory with MemVerge* #169 – In-Memory Computing and Apache Ignite* #36 – The Persistence of Memory with Rob Peglar

Copyright (c) 2016-2020 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #wxje.

This week Chris and Martin take the opportunity to grill (not broil) Matt Starr from Spectralogic to learn everything about magnetic tape. This episode came out of a discussion with Chris Mellor (episode #176) in which we asked the question why tape media doesn’t wear out. The answer is obviously within the podcast so you’ll need to listen to find out!

In this conversation, the team look back at the evolution of tape but more importantly what we can expect in the future. Magnetic tape is more of an archive offering these days (although ransomware has made it more popular for air-gapping backups). It seems tape has a long future as the lowest cost storage medium with lots of innovations still to come.

You can learn more about Spectra Logic over at https://spectralogic.com/. The featured image for this episode is a single 44-cabinet tape library as mentioned in the podcast that was taken at a Spectra Logic event in 2013.

Elapsed Time: 00:35:47

Timeline

* 00:00:00 – Intros * 00:02:20 – A brief history of magnetic tape* 00:03:00 – Cartridges introduced the ability for automation * 00:05:50 – AT&T test pings and automated tape loading * 00:07:25 – LTO and MagStar are the leading tape formats today * 00:08:00 – What is helical scan, serpentine and linear? * 00:09:30 – Tape damage is an ongoing issue * 00:10:00 – ECC on tape is much better than disk * 00:13:40 – What size are tape heads and how do they fly? * 00:18:51 – What are the current tape technologies? * 00:20:00 – IBM can test features on Jaguar before considering LTO * 00:21:50 – Disk took over the backup mantle from tape * 00:23:40 – Ransomware makes tape more applicable for backup * 00:25:50 – Tape offers the cost-effective location for data * 00:28:00 – Tape users are decreasing but storing more data * 00:29:50 – What is the future for tape? * 00:30:28 – Time to first byte will be important to improve * 00:33:00 – Object to Tape is on the horizon

Related Podcasts & Blogs

* #176 – Opinionated Storage Opinions with Chris Mellor* Garbage Collection #006 – LTO-8 and the Future of Tape* It’s Time for Hard Drives to Join Tape in the Archive Tier

Copyright (c) 2016-2020 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #zdn9.

This week, Chris and Martin chat to Evan Weaver, CTO and co-founder at Fauna. The company has developed a database as a service, available simply through a globally available API. Fauna calls the platform “client serverless”. How can a database be made globally available and still guarantee performance, resiliency and most important, data consistency? Of course, you will need to listen to find out.

As we learn in this podcast, a database accessible via API provides the ability to write applications that talk directly to the platform, with application use cases for mobile and IoT.

Evan references the Calvin algorithm. A detailed paper can be found here. More information on Google Spanner is available through this link. Google Percolator is documented here. Finally, here’s a link to an article that discusses the idea of the Sysplex timer on the mainframe, a feature from 1990, that still remains important today.

If you’d like to try FaunaDB for yourself, get over to https://fauna.com/ and sign up.

Elapsed Time: 00:35:28

Timeline

* 00:00:00 – Intros* 00:02:30 – We’re seeing a return to new “client server”* 00:03:37 – Database users could be devices, not humans* 00:05:30 – We’ve gone full circle to needing powerful backends * 00:07:09 – Fauna developed out of the need to scale Twitter* 00:09:00 – How does FaunaDB manage typical operational challenges?* 00:11:07 – Single clock – mainframe parallel Sysplex* 00:13:48 – Technology reinvents itself* 00:15:07 – What is CALVIN?* 00:19:00 – Web APIs aren’t long-running transactions* 00:23:00 – Partitioning by geography is the typical way to maintain performance* 00:24:21 – What are the use cases for Fauna?* 00:27:00 – How are the operational challenges addressed?* 00:30:00 – Can users try FaunaDB? * 00:33:37 – Wrap Up

Related Podcasts & Blogs

* #140 – Introduction to Document-Orientated Databases* #147 – Introduction to Key Value Stores and Redis* Databases are the next battleground for public cloud

Copyright (c) 2016-2020 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #ecko.

This week Chris and Martin debate the future of on-premises storage infrastructure companies in light of announcements and events from IBM, Dell and NetApp. IBM is splitting in two, while Dell and NetApp have recently held virtual versions of their annual conferences. Is the on-premises infrastructure business shrinking and becoming too low-margin to be worthwhile?

Martin mentions Lou Gerstner’s book – linked here – Who Says Elephants Can’t Dance?

The Simpson’s episode where Mr Burns runs for Governor is “Two Cars in Every Garage and Three Eyes on Every Fish” – without a doubt a classic.

Martin mentions Fabric Pools – here’s a link to the documentation.

Elapsed Time: 00:36:09

Timeline

* 00:00:00 – Intros * 00:02:00 – What’s the future of the infrastructure business? * 00:03:05 – IBM splitting into two companies * 00:06:00 – IBM mainframe storage will stay in IBM? * 00:08:40 – Mainframe retained for z/Linux? * 00:10:00 – IBM continually reinvents itself * 00:11:00 – Where does IBM Research go? * 00:13:10 – What was the value of DTW? * 00:14:40 – So Mr Burns, why are you so popular? * 00:16:40 – Apex – Capex with a silent C? * 00:18:15 – Is “as a service” a real infrastructure business model? * 00:19:30 – Violin Systems acquired by StorCentric * 00:20:40 – Could the IBM spinout be acquired? * 00:21:50 – NetApp is morphing into a cloud business * 00:24:55 – Project Astra – NetApp wants to be cloud glue * 00:26:40 – Is cloud storage enough to re-invent NetApp? * 00:28:30 – Everyone is “service washing” their product lines * 00:29:15 – Recurring revenue is the holy grail * 00:31:50 – Buying MIPS is the future! * 00:32:35 – As-a-service storage needs to be operationally efficient * 00:35:25 – Wrap Up

Related Podcasts & Blogs

* #107 – Should IBM Quit the Storage Business?* #65 – Challenges in Managing Unstructured Data* #130 – Making Money in the Storage Business

Copyright (c) 2016-2020 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #blot.