Technology:Unpacked

In this podcast recorded at NetApp Insight 2018, Chris talks to Santosh Rao, Senior Technical Director at NetApp.  Santosh leads NetApp’s AI and Data Engineering efforts and has a 10-year history at the company, working on initiatives including Clustered ONTAP.

This conversation covers how NetApp is developing an Edge-to-Core-to-Cloud strategy that includes collecting and processing data at the edge, storing it in the core and exploiting the AI tools of public cloud.

End users want to make the most of CPU and GPU solutions, while minimising data movement.  Edge Solutions, including ONTAP Select provide the means to collect and store data with some pre-processing, if required.  The features of ONTAP allow data to be efficiently moved to the core and cloud for further analysis.

More information on NetApp’s AI strategy can be found at https://www.netapp.com/ai.

Elapsed Time: 00:24:01

Timeline

* 00:00:00 – Intros

* 00:01:20 – What is NetApp doing with AI? Data pipelines

* 00:02:50 – What does Edge->Core->Cloud mean?

* 00:05:00 – Entire devices are being used for de-stage of data

* 00:06:10 – How much pre-processing is being done on the ONTAP Select device?

* 00:08:25 – How does data get into the cloud for processing?

* 00:09:40 – Customers are building multi-cloud architectures to use CSP AI technology

* 00:11:30 – Tri-cloud architectures are being affected by GDPR

* 00:12:25 – ONTAP AI – a software stack and hardware offering – include cloud

* 00:14:50 – What are the biggest challenges for a distributed AI architecture?

* 00:16:00 – ONTAP enables “data copy avoid”

* 00:17:00 – Data is now the centre of the IT universe

* 00:19:20 – NASA – 30 years of data, outliving the platform/compute

* 00:21:40 – Where are we headed in the next 3-5 years?

* 00:23:20 – Wrap Up

Related Blogs & Podcasts

* #78 – Thoughts on NetApp with Greg Knieriemen

* Soundbytes #014: A Coversation with NetApp Founder Dave Hitz

* Soundbytes: The Data Fabric Eplained with NetApp CTO Mark Bregman

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode QYNZ.

This week Chris and Martin talk to Steve Hanna, Senior Product Manager for Enterprise SSDs at Micron.  Steve looks after the 5210 ION QLC SSD product line, which was introduced earlier this year.  As an evolution of existing flash technology, QLC offers greater capacities at lower cost.  But what are the issues?

The conversation covers how QLC has reduced the price to 1/100th of that seen in early enterprise SSDs.  Although endurance is lower than previous flash generations, in reality for many use cases the ability to write up to 1 DWPD is enough.  However, we will see QLC co-exist with TLC, rather than become a direct replacement.  Understanding workloads will be critical and the discussion covers this a few times, just for clarity.

Steve points us at a couple of resources.  Check out https://micron.com/5210 for products specs on the 5120 ION drive and also https://micron.com/QLC for all things QLC.  Here you can find helpful information and guides to understanding the technology.

Elapsed Time: 00:30:06

Timeline

* 00:00:00 – Intros

* 00:01:26 – What is QLC Flash?

* 00:02:20 – Diminishing returns from bit density

* 00:03:00 – Price reductions – $20 to $0.20/GB

* 00:05:00 – How does endurance change with QLC?

* 00:07:00 – QLC will be a read-intensive workload product

* 00:09:40 – But QLC will not be an entire replacement for TLC

* 00:10:30 – Will QLC kill the HDD?  I this a TCO play?

* 00:14:00 – Use cases, like CDNs

* 00:15:30 – How will workload mix differ with QLC?

* 00:17:20 – Tiering will be back in SSDs

* 00:18:30 – QLC will be the performance capacity tier

* 00:19:30 – Why SATA though?  Why not NVMe for QLC?

* 00:22:30 – Let’s talk workloads again.

* 00:25:00 – Why are we not starting at 32TB for QLC drives?

* 00:28:00 – Wrap up

Related Blogs & Podcasts

* #54 – Are we at All-Flash and HDD Array Price Parity?

* Garbage Collection – All-flash Market Consolidation

* #74 – All About Serial Attached SCSI with Rick Kutcipal

* #44 – Ultra High Capacity Flash Drives

* QLC NAND – how real is it and what can we expect from the technology?

* Micron users in the era of QLC SSDs

Steve’s Bio

Steve is an acclaimed author, speaker, innovator, and leader. Steve has transformed products, led businesses, and brought new categories to life – like the world’s first QLC SSD, the hardware heartbeat of artificial intelligence, machine learning, analytics, media streaming, and more. Steve studied innovation and entrepreneurship at Stanford and has authored over 50 thought leadership pieces and delivered seminars on enterprise storage, artificial intelligence, machine learning, and building a winning team culture.

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 5CAB.

Last week Chris attended NetApp Insight in Barcelona and had a chance to chat to Greg Knieriemen, Chief Technologist in the Storage Systems & Software business at NetApp.  Greg was formerly at Hitachi Vantara where he held a similar role.  This conversation touches on Greg’s reasons for choosing NetApp and exactly what he will be focusing on within the company.

Elapsed Time: 00:17:07

Timeline

* 00:00:00 – Intros

* 00:01:30 – Why move to NetApp?

* 00:04:00 – Three BUs, Storage/Systems, Cloud Infrastructure, Cloud Services

* 00:07:00 – The Data Fabric is much more real today than two years ago

* 00:08:00 – Acquisitions, StackPoint, GreenQloud, SolidFire

* 00:10:00 – Cultural change – allowing new processes to drive existing solutions

* 00:11:00 – Where are the gaps?  Bringing other traditional products into the strategy

* 00:13:00 – What will Greg be focusing on over the next 6 months? MAX Data

* 00:16:00 – Wrap Up

Related Podcasts & Blogs

* Soundbytes #013: Hitachi Vantara with Greg Knieriemen

Greg’s Bio

Greg Knieriemen is the NetApp Chief Technologist and helps develop and drive the vision and application of NetApp products and solutions. Previously, Greg worked for Hitachi and was the founder of the Speaking in Tech Podcast. Greg has over 15 years of experience using, deploying and marketing enterprise IT solutions.

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode CFWD.

This week HPE announced Performance Insights, an extension to the InfoSight platform that provides storage customers with greater detail on their 3PAR array storage performance.  Performance Insights takes the knowledge of years of HPE 3PAR data and combines it with genuine machine learning and AI to give HPE 3PAR customers actionable insights on the performance characteristics of their storage arrays.  In particular this means being able to both detect performance problems and also do more accurate workload planning.

A few weeks ago, Martin and Chris talked to Ivan Iannaccone, VP and GM for the HPE 3PAR product line.  Ivan takes us through the background of InfoSight, which came with the Nimble Storage acquisition.  The discussion digs down into why performance management is becoming too hard for individuals and how AI can do a much better job.

For more information in the Performance Insights news, check out Calvin Zito’s Around the Storage Block blog and podcast.

Elapsed Time: 00:35:58

Timeline

* 00:00:00 – Intros

* 00:02:30 – InfoSight – Nimble AI Analytics

* 00:05:00 – What does InfoSight actually do?

* 00:07:00 – InfoSight now being applied to 3PAR

* 00:10:30 – InfoSight was the value add to a solid platform

* 00:14:00 – How do we tackle issues with performance?

* 00:19:00 – Using performance metrics to maximise investment

* 00:22:00 – Saving customers real money by identifying problem issues

* 00:24:00 – How will customers see Performance Insights on 3PAR?

* 00:26:50 – What makes Performance Insights really ML/AI based?

* 00:34:00 – Wrap Up

Related Blogs & Podcasts

* HPE Brings Nimble Skynet to 3PAR Arrays

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 89PY.

In this week’s podcast, Chris and Martin talk to Mark Jones from the Fibre Channel Industry Association.  This recording is an introduction to running NVMe over Fibre Channel, setting the scene on how Fibre Channel has evolved and will continue to be a storage protocol for many years.

Fibre Channel reached an important milestone in 2018.  It’s been 30 years since development and 25 years since the first products were shipped.  Speeds started at 133Mb/s, while today, Gen 6 FC runs at 32Gb/s.  Higher speeds are coming in 2019 and can also be achieved with channel bonding.

How does NVMe fit in with Fibre Channel?  FC can be used as the transport layer for many storage protocols, including FCP (SCSI), FICON and now NVMe.  We can expect to see significant performance improvements in performance and reductions in latency compared to FCP.  Even better, with the right level of equipment, both FCP and NVMe run side by side.

For more information on FC-NVMe, check out the FCIA website at fibrechannel.org.

Elapsed Time: 00:28:58

Timeline

* 00:00:00 – Intros

* 00:00:00 – What is the FCIA?

* 00:02:00 – What is Fibre Channel

* 00:04:30 – Yes, Fibre is the correct spelling

* 00:05:00 – The FC-4 layer – where the storage protocols run

* 00:06:30 – Which is fastest, Ethernet or Fibre Channel?

* 00:09:00 – Let’s talk FC-NVMe!

* 00:12:50 – How much of a performance gain is there with FC-NVMe?

* 00:14:30 – Fibre Channel supports multiple storage protocols – at the same time

* 00:16:40 – Is Fibre Channel adoption growing or decreasing?

* 00:18:50 – What’s needed to install and use NVMe?

* 00:23:40 – FC or Ethernet – place your bets

* 00:27:30 – Wrap up

Related Podcasts & Blogs

* #74 – All About Serial Attached SCSI with Rich Kutcipal

* #61 – Introduction to NVM Express with Amber Huffman

* #59 – Ethernet vs Fibre Channel

Mark’s Bio

 

Mark Jones has worked in the enterprise computing industry since 1984.  Currently is the Director of Technical Marketing for Broadcom Inc. Emulex Connectivity Division.  Mr. Jones comes to Broadcom via the acquisition of Emulex Corporation in 2015 for which he has worked since 2002.  For the 18 years prior he worked as a Strategic Solutions Manager for Burroughs/Unisys.  Mr. Jones has a Computer Science degree from the University of Redland and currently serves as the President of the Fibre Channel Industry Association.

 

 

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 22YV.

 

 

Data volumes have always increased over time as we store more information with the hope that one day some of it will be useful.  Even 30 years ago on the mainframe, Information Lifecycle Management or ILM was a thing with tools like DFHSM used to move content around between disk and tape.  This week Martin, Chris Evans and Chris Mellor talk about the new range of data management or ILM products that are looking to resolve the current issues of data sprawl.

How do these products work?  It is all about data ingest, or just indexing?  What happens when data moves between platforms and how do files continue to be accessed?  Companies like Komprise claim to be able to move data without impacting the application or needing to use techniques like stubs.  Others like IBM Spectrum Discover, simply appear to do content indexing.  Ultimately, perhaps we need to move everything to object stores and dispense with file services altogether.  Can we ignore public cloud and can we get the S3 API moved to an open standard?  All questions the team attempt to answer on this week’s podcast.

Elapsed Time: 00:37:26

Timeline

* 00:00:00 – Intros

* 00:01:00 – Why do we need data management or ILM?

* 00:03:00 – How do we manage so many different data silos?

* 00:04:40 – Data management gets conflated with other features

* 00:05:00 – What vendors are there in this space?

* 00:07:00 – ILM on the mainframe!

* 00:09:00 – Should ILM functions be built into the file system?

* 00:11:00 – How is public cloud influencing the ILM process?

* 00:12:30 – Data gravity (or inertia) causes problems moving into the cloud

* 00:16:00 – How can we solve the distributed data problem?

* 00:18:00 – Hammerspace – separating data and metadata

* 00:21:00 – Everyone needs ML & AI in their storage platforms

* 00:22:00 – Have we simply not defined the data management problem?

* 00:25:00 – Remember File Area Networks?

* 00:27:00 – How does IBM Spectrum Discover work?

* 00:31:00 – Qumulo QF2 – focused on fast metadata searches

* 00:32:30 – We still have a lot of sticking plaster and bandaids in place

* 00:33:40 – Should we just move everything to object stores?

* 00:35:30 – Please Jeff, can you donate the S3 API to the community?

* 00:36:30 – Wrap Up

Related Podcasts & Blogs

* #68 – Intelligent Object Storage with Scott Baker

* #65 – Challenges in Managing Unstructured Data with Shirish Phatak

* #60 – New Data Economy with Derek Dicker

* Data Gravity Pointed the Way to Data Rather than Storage Management

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode HNWX.

Most people probably don’t think about the protocol that connects their storage device to a PC or server.  However, NVMe has changed that and become one of the storage buzzwords of 2018.  But what about existing devices not using NVMe?  SAS or Serial Attached SCSI has been a mainstay of storage connectivity for 20 years.  With 12Gb/s throughput, SAS provides back-end connectivity for almost all the enterprise storage arrays on the market today.

The SAS protocol continues to be developed, with 24Gb/s speeds due in 2019.  However, SAS is not all about speed, as we find out when Chris and Martin talk to Rick Kutcipal, President of the SCSI Trade Association.  In fact, the SAS protocol has evolved with features to improve reliability and scalability.  New features are addressing storage technologies like SMR and HDD multi-actuator.

As a technology we take for granted, there’s a lot more going on than we realise.  SAS will still have a future in the data centre, as it offers performance, reliability and possibly most important, scalability for vendors to build out large-scale storage solutions.

You can find out more about SAS, at the SCSI Trade Association website or follow them on twitter at @sas_storage.

Elapsed Time: 00:22:21

Timeline

* 00:00:00 – Intro

* 00:01:30 – What is the SCSI Trade Association?

* 00:02:40 – What is a plugfest?

* 00:03:40 – What is SAS (Serial Attached SCSI)?

* 00:05:50 – SAS-4 is due – 24Gb/s throughput

* 00:07:30 – Where does SAS fit in, with the emergence of NVMe?

* 00:09:50 – Is SAS still good enough for most storage requirements?

* 00:11:00 – When will 24Gb SAS be available and what will it give?

* 00:12:15 – New encoding scheme 8b/10b to 128b/150b.

* 00:13:30 – New storage intelligence features

* 00:15:30 – Do end users and buyers care about this level of detail?

* 00:18:00  – Protocol changes will come for SMR and multi-actuator support

* 00:19:30 – So how should we position SAS, SATA and NVMe?

* 00:21:00 – Call to Action – where to find out more?

* 00:21:50 – Wrap Up

Related Podcasts & Blogs

* #61 – Introduction to NVM Express with Amber Huffman

* #59 – Ethernet vs Fibre Channel

* One (Storage) Protocol to Rule Them All?

* NVMe over Fabrics – Caveat Emptor?

Rick’s Bio

Rick Kutcipal – President, SCSI Trade Association; Marketing Manager, Data Center Storage Group, Broadcom Inc.  Currently serving as a Marketing Manager in the Data Center Storage Group at Broadcom, Rick is a 25-year computer and data storage business veteran. He coordinates the majority of standards activities for Broadcom worldwide and serves at the President of the SCSI Trade Association. Rick received his bachelor’s and master’s degrees in electrical engineering from the University of Utah.

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode IJ10.

Most people probably don’t think about the protocol that connects their storage device to a PC or server.  However, NVMe has changed that and become one of the storage buzzwords of 2018.  But what about existing devices not using NVMe?  SAS or Serial Attached SCSI has been a mainstay of storage connectivity for 20 years.  With 12Gb/s throughput, SAS provides back-end connectivity for almost all the enterprise storage arrays on the market today.

The SAS protocol continues to be developed, with 24Gb/s speeds due in 2019.  However, SAS is not all about speed, as we find out when Chris and Martin talk to Rick Kutcipal, President of the SCSI Trade Association.  In fact, the SAS protocol has evolved with features to improve reliability and scalability.  New features are addressing storage technologies like SMR and HDD multi-actuator.

As a technology we take for granted, there’s a lot more going on than we realise.  SAS will still have a future in the data centre, as it offers performance, reliability and possibly most important, scalability for vendors to build out large-scale storage solutions.

You can find out more about SAS, at the SCSI Trade Association website or follow them on twitter at @sas_storage.

Elapsed Time: 00:22:21

Timeline

* 00:00:00 – Intro

* 00:01:30 – What is the SCSI Trade Association?

* 00:02:40 – What is a plugfest?

* 00:03:40 – What is SAS (Serial Attached SCSI)?

* 00:05:50 – SAS-4 is due – 24Gb/s throughput

* 00:07:30 – Where does SAS fit in, with the emergence of NVMe?

* 00:09:50 – Is SAS still good enough for most storage requirements?

* 00:11:00 – When will 24Gb SAS be available and what will it give?

* 00:12:15 – New encoding scheme 8b/10b to 128b/150b.

* 00:13:30 – New storage intelligence features

* 00:15:30 – Do end users and buyers care about this level of detail?

* 00:18:00  – Protocol changes will come for SMR and multi-actuator support

* 00:19:30 – So how should we position SAS, SATA and NVMe?

* 00:21:00 – Call to Action – where to find out more?

* 00:21:50 – Wrap Up

Related Podcasts & Blogs

* #61 – Introduction to NVM Express with Amber Huffman

* #59 – Ethernet vs Fibre Channel

* One (Storage) Protocol to Rule Them All?

* NVMe over Fabrics – Caveat Emptor?

Rick’s Bio

Rick Kutcipal – President, SCSI Trade Association; Marketing Manager, Data Center Storage Group, Broadcom Inc.  Currently serving as a Marketing Manager in the Data Center Storage Group at Broadcom, Rick is a 25-year computer and data storage business veteran. He coordinates the majority of standards activities for Broadcom worldwide and serves at the President of the SCSI Trade Association. Rick received his bachelor’s and master’s degrees in electrical engineering from the University of Utah.

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode IJ10.

This week, Chris and Martin talk to Subbiah Sundaram, VP of Products at HYCU, Inc.  HYCU is both the name of the company and the data protection product sold by HYCU Inc.

The market already has many data protection solutions, so the unique differentiator for HYCU is focusing on the HCI market, specifically offering backup for Nutanix (both ESXi and AHV).  Of course, HCI platforms already have backup support, but as the discussion explains, having deeper integration into the HCI APIs, specifically those around changed file tracking, provide HYCU with an advantage over products backing up at the hypervisor layer.

As Chris and Martin find out, the name HYCU was originally intended to mean Hyper-Converged Uptime, but the association with Haiku (the Japanese short poems) is much more interesting.  In fact, Martin created a HYCU Haiku during the recording:

Martin’s Haiku: My data flowing, disaster strikes like lightning, polarity reverses

You can find out more details on HYCU at https://www.hycu.com/ and as Subbiah explains, at go to https://www.hycu.com/tryhycu/ for a free 45-day download.

Elapsed Time: 00:34:35

Timeline

* 00:00:00 – Intros

* 00:01:34 – Why does HCI backup justify a new way of working?

* 00:03:30 – HCI introduces a new platform, not just a server to back up

* 00:04:30 – Virtualisation changed the individual server backup paradigm

* 00:05:30 – APIs for taking backups – HCI provide another access point

* 00:07:00 – Where does the name HYCU and the company come from?

* 00:11:30 – Martin creates a Haiku

* 00:13:00 – How is the HYCU software deployed?

* 00:15:00 – How do service providers use HYCU with Nutanix?

* 00:19:40 – What differentiates HYCU from other backup vendors?

* 00:21:30 – Taking data from VADP or from the HCI storage layer?

* 00:23:55 – Applying backup to Nutanix Files (formerly AFS)

* 00:25:30 – Nutanix uses the Changed File Tracking (CFT) API

* 00:27:20 – Cross-cluster backup & restores and cross-hypervisor

* 00:31:50 – How does HYCU licensing work?

* 00:33:00 – Call to action – what’s next?

* 00:33:50 – Wrap up

Related Podcasts & Blogs

* Validating HYCU Deployment By The Numbers

* Comtrade Software Becomes HYCU and releases HYCU 3.0

* Automating HCI Data Protection

* Why HCI Data Protection is Different

* Briefing Document: HYCU

Subbiah’s Bio

Subbiah Sundaram is Vice President of Products at HYCU. He joined HYCU January of 2017 and has been instrumental in enabling the company to deliver the best in class multi-cloud solutions for Nutanix and Google Cloud. Prior to joining HYCU, Subbiah held senior executive positions at BMC, CA, DataGravity, EMC, NetApp and Veritas and has extensive experience in product development, planning and strategy. He holds a MS in Computer Engineering from the University of Iowa and an MBA from the Kellogg School of Management at Northwestern University. ​

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode KUVM.

 

 

This week, Chris and Martin talk to Subbiah Sundaram, VP of Products at HYCU, Inc.  HYCU is both the name of the company and the data protection product sold by HYCU Inc.

The market already has many data protection solutions, so the unique differentiator for HYCU is focusing on the HCI market, specifically offering backup for Nutanix (both ESXi and AHV).  Of course, HCI platforms already have backup support, but as the discussion explains, having deeper integration into the HCI APIs, specifically those around changed file tracking, provide HYCU with an advantage over products backing up at the hypervisor layer.

As Chris and Martin find out, the name HYCU was originally intended to mean Hyper-Converged Uptime, but the association with Haiku (the Japanese short poems) is much more interesting.  In fact, Martin created a HYCU Haiku during the recording:

Martin’s Haiku: My data flowing, disaster strikes like lightning, polarity reverses

You can find out more details on HYCU at https://www.hycu.com/ and as Subbiah explains, at go to https://www.hycu.com/tryhycu/ for a free 45-day download.

Elapsed Time: 00:34:35

Timeline

* 00:00:00 – Intros

* 00:01:34 – Why does HCI backup justify a new way of working?

* 00:03:30 – HCI introduces a new platform, not just a server to back up

* 00:04:30 – Virtualisation changed the individual server backup paradigm

* 00:05:30 – APIs for taking backups – HCI provide another access point

* 00:07:00 – Where does the name HYCU and the company come from?

* 00:11:30 – Martin creates a Haiku

* 00:13:00 – How is the HYCU software deployed?

* 00:15:00 – How do service providers use HYCU with Nutanix?

* 00:19:40 – What differentiates HYCU from other backup vendors?

* 00:21:30 – Taking data from VADP or from the HCI storage layer?

* 00:23:55 – Applying backup to Nutanix Files (formerly AFS)

* 00:25:30 – Nutanix uses the Changed File Tracking (CFT) API

* 00:27:20 – Cross-cluster backup & restores and cross-hypervisor

* 00:31:50 – How does HYCU licensing work?

* 00:33:00 – Call to action – what’s next?

* 00:33:50 – Wrap up

Related Podcasts & Blogs

* Validating HYCU Deployment By The Numbers

* Comtrade Software Becomes HYCU and releases HYCU 3.0

* Automating HCI Data Protection

* Why HCI Data Protection is Different

* Briefing Document: HYCU

Subbiah’s Bio

Subbiah Sundaram is Vice President of Products at HYCU. He joined HYCU January of 2017 and has been instrumental in enabling the company to deliver the best in class multi-cloud solutions for Nutanix and Google Cloud. Prior to joining HYCU, Subbiah held senior executive positions at BMC, CA, DataGravity, EMC, NetApp and Veritas and has extensive experience in product development, planning and strategy. He holds a MS in Computer Engineering from the University of Iowa and an MBA from the Kellogg School of Management at Northwestern University. ​

Copyright (c) 2016-2018 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode KUVM.