Technology:Unpacked

This week, Chris talks to Vince Westin, Technical Evangelist within the PowerMax Group at Dell EMC.  PowerMax is the latest in a storage product line with lineage back to the original Symmetrix systems of the early 1990s.  PowerMax is an all-NVMe storage array, reflecting an industry move towards faster, low latency media.  Vince takes us through the rationale for moving to NVMe across the Dell portfolio and PowerMax in particular.

Some of the more interesting aspects of the transition include the ability to simplify code paths with NVMe compared to back-end SAS.  NVMe introduces a shared PCIe fabric and increases parallelism across the infrastructure.  This results, of course, in lower latency for specific workloads like high random I/O.

In 2019 we can expect to see Storage Class Memory being introduced into the PowerMax family.  This brings challenges around how data should be encrypted and whether data optimisation is worthwhile, bearing in mind the additional performance overhead.

You can find Vince’s personal blog at https://applyit.typepad.com/ and the specific post relating to SCM and PowerMax here.  We also recommend viewing the Storage Field Day 16 videos of Vince’s presentation on PowerMax, which helps to fill in the visual gaps missing from the podcasting format!

Elapsed Time: 00:39:59

Timeline

* 00:00:00 – Intros

* 00:01:00 – NVMe primer and recap

* 00:04:20 – How is Dell and Dell EMC using NVMe?

* 00:06:00 – SCM will be introduced to arrays this year

* 00:08:00 – Managing transitions – NVMe internally first

* 00:09:00 – Moving to NVMe-oF means multi-pathing differently.

* 00:10:00 – How was VMAX managed at the back end to develop PowerMax?

* 00;14:00 – Is NVMe scalable at the back-end like SAS was?

* 00:16:30 – What performance improvements can customers expect?

* 00:18:30 – PowerMax uses dynamic data reduction for least active data

* 00:20:00 – What media does PowerMax use? Tiering is back!

* 00:28:41 – How will customers move to PowerMax?

* 00:32:20 – Is there a future for SAS in Dell EMC systems?

* 00:36:30 – Why to customers continue to use platforms like PowerMax?

* 00:38:00 – Dell Technologies World is around the corner

Related Podcasts & Blogs

* #43 – All-flash Market Review 2018 with Chris Mellor

* #56 – Defining Scale-out Storage

* PowerMax, VMAX, XtremIO – the Agony of Choice

* The Race towards End-to-End NVMe in the Data Centre

Vince’s Bio

Vince Westin is a Technical Evangelist for PowerMax in the Infrastructure Solutions Group at Dell EMC. In this role, Vince focuses on presenting the PowerMax strategy and roadmap to customers, as well as gathering feedback used to set future product requirements. Vince has been with Dell EMC since 1995, holding various positions in pre-sales, engineering, implementations, and product marketing. Prior to joining Dell EMC, Vince worked for technology companies in pre-sales, development, and systems implementation. Vince graduated from the University of California, Berkeley with degree in Electrical Engineering Computer S...

This episode was recorded while Chris was in Silicon Valley for Storage Field Day 18.  Although not part of the event, Chris took time to catch up with Pavilion Data. in particular Jeff Sosa, VP Product Management and Walt Hinton, Head of Corporate and Product Marketing.  Both Jeff and Walt have a long history in the storage industry, including companies like Virident, Western Digital, Fusion-io, Brocade, NetApp and Data Domain – to name but a few!  It’s great to get opinions on the future of storage technology from industry experts, in this case on the subject of NVMe architectures.

NVMe represents a paradigm shift in how storage platforms are implemented.  Where companies moved away from SAN to get local performance and agility, new architectures are enabling modern applications to take advantage of shared rack-scale storage and compute.  Exactly how this plays out is still in process, but we can see a move towards the SAN technology that was so successful in the early part of the new millennium.

More information on Pavilion Data can be found at paviliondata.com.

Elapsed Time: 00:27:37

Timeline

* 00:00:00 – Intros

* 00:02:00 – Are we seeing the industry cycle around again with SANs?

* 00:03:00 – Industrial Military Complex – Microsoft, Oracle!

* 00:06:30 – Why did the new apps not use Fibre Channel?

* 00:09:00 – Understanding storage disaggregation

* 00:10:30 – Could scale out databases still benefit from shared storage?

* 00:13:00 – How can pricing be maintained without data reduction?

* 00:18:00 – Do super-high capacity drives force a return to shared storage?

* 00:21:00 – Should architectures start using cache-less writes?

* 00:24:00 – How are rack-scale storage solutions being adopted?

* 00:27:30 – Wrap Up

Related Podcasts & Blogs

* #84 – Discussing New NVMe Form Factors with Jonathan Hinkle

* #76 – Fibre Channel and NVMe with Mark Jones

* #61 – Introduction to NVM Express with Amber Huffman

* The Race towards End-to-End NVMe in the Data Centre

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 8D01.

This week Chris and Martin talk to Nigel Tozer, Solutions Marketing Director for EMEA at Commvault.  Nigel was a guest about 12 months ago and on that episode he talked about GDPR.  This time the discussion is about ransomware and what businesses can do about it.

The challenges of protecting data from theft or extortion are greater than ever.  So, can we identify a common attack model?  Are specific operating systems more vulnerable?  Most important, how do you develop a plan that protects your data and systems?  The process is more than just patching primary systems (although we talk about that).  It also means ensuring backup data is safe, because this is the main means of recovery.

Nigel references the National Cyber Security Centre, which can be found here.  https://www.ncsc.gov.uk/

Elapsed Time: 00:32:02

Timeline

* 00:00:00 – Intros

* 00:01:20 – GDPR – was it a storm in a teacup?

* 00:02:30 – What is ransomware?

* 00:04:00 – Is there a common attack surface?

* 00:06:20 – How does Ransomware get installed?

* 00:07:12 – Ransomware as a service!

* 00:08:30 – Should you pay a ransom?

* 00:10:52 – What are the financial impacts for enterprises?

* 00:14:30 – Could backups also get compromised?

* 00:17:00 – How do you protect official exits?

* 00:20:00 – How do you know when you have active ransomware?

* 00:23:00 – Defence in depth, networking protection is needed too

* 00:23:00 – How should patching be done?

* 00:27:00 – The US government shutdown exposed risk to ransomware

* 00:29:00 – So what strategy should be used to deal with ransomware?

* 00:31:00 – Wrap Up

Related Podcasts & Blogs

* Garbage Collection #008 – Chris’ Travels – Commvault, NetApp and SFD14

* #86 – Eran Brown Discusses Storage, Security & Multi-Cloud

* Introduction to GDPR with Nigel Tozer from Commvault

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 7CCD.

In this episode, Chris and Martin talk to Simon Zola from Avtel Data Destruction.  This is a follow-up episode to #85 (Storage for Home and Homelabs) where Martin and Chris questioned the ease of recovering data from pre-owned storage media.  As we find out, the process was pretty easy.  Martin was able to recover significant amounts of personal data with little effort and using standard tools.

Media destruction is one route to safely dispose of drives and guarantees that data is destroyed.  But what are the consequences?  Simply disabling a drive and throwing it into landfill isn’t good for anyone.  An alternative is to recycle the media.  Simon Zola from Avtel Data Destruction takes us through their hardware milling process, capable of turning a hard drive to dust in 7 seconds!  We discuss the environmental and safety aspects of such a process and learn that the downstream waste produced by the process has ongoing value.  Nothing goes to landfill.

You can find more about Avtel on their website – here.  There’s also a video on YouTube that provides an idea of how the process works.

Elapsed Time: 00:27:28

Timeline

* 00:00:00 – Intros

* 00:01:30 – What did Martin find with used SSDs?

* 00:04:53 – Why should we be destroying drives?

* 00:06:44 – Will encrypted drives have more value?

* 00:07:25 – 25 million mobile phones kept in cupboards

* 00:08:30 – How does Avtel destroy media?

* 00:10:00 – Japan Olympics medals will come from recycled mobiles

* 00:12:20 – A chain of custody is as important as destruction

* 00:19:00 – What approach should businesses take with media?

* 00:21:00 – Are desktop drives an exposure for businesses?

* 00:22:00 – Are there any options for home users?

* 00:25:00 – Should vendors be taking more responsibility (or are they)?

* 00:26:40 – Wrap Up

Related Podcasts & Blogs

* #85 – Storage for Home and Homelabs

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 94C8.

This week, Chris and Martin talk to Eran Brown, CTO in EMEA for INFINIDAT.  As we move to a more distributed world, implementing data security becomes ever more challenging.  Eran explains the issues and how his customers are looking to solve the problem.

One of the initial choices is to decide exactly where to encrypt.  We’re familiar with drive-level encryption, even host-level.  Eran sees hypervisor and application encryption being equally if not more important.

What are the implications on technology?  Encryption reduces savings from de-duplication and compression, changing the cost model.  Encryption and data security needs to be applied equally to secondary backup data, which remains a big exposure for many companies.

As the conversation moves towards public cloud and micro-services, it’s clear that data gravity poses a challenge for building out a secure data framework.  It’s about choosing the right centre of gravity for data – not applications.

You can find more about INFINIDAT at their EMEA homepage here.  Eran discusses two technologies; Oracle TDE (Transparent Data Encryption) and vSphere encryption.

Elapsed Time: 00:34:07

Timeline

* 00:00:00 – Intro

* 00:02:00 – Is co-location and cloud being heavily adopted?

* 00:03:00 – How “high up” should data be secured?

* 00:04:00 – The issues revolve around skills, rather than technology

* 00:06:00 – What are the security skills we need?

* 00:08:00 – Where can encryption be done?

* 00:10:26 – How do we map risk to applications?

* 00:12:50 – How do we deal with legacy systems?

* 00:13:30 – How will key management work in distributed environments?

* 00:14:40 – What breaks when we start encrypting upstream?

* 00:17:30 – Is backup secure enough?

* 00:19:34 – How will we manage trust for short-lived micro services?

* 00:24:00 – Data Gravity will influence where data is managed from

* 00:28:20 – How should end users develop a data security strategy?

* 00:32:00 – Wrap Up

Related Podcasts & Blogs

* #82 – Storage Predictions for 2019

* Data Protection in a Multi-Cloud World

* Evolving Enterprise Storage Buying Models

* Enterprise-Class Public Cloud

Eran’s Bio

Eran Brown is CTO, EMEA, at INFINIDAT.  With over 15 years in the storage industry, Brown has an extensive technical background in networking, security, virtualisation, Big Data tools and other critical infrastructure components. Having gained broad experience as a Pre-Sales Engineer at NetApp to a Senior Product Manager at INFINIDAT, Brown has proven results in leading technical sales activities and delivering new products to market. In addition to this, he has familiarised himself with the multiple verticals throughout his career including the financial sector, oil & gas, telecoms and software industry – assisting in the planning, design and deployment of scalable infrastructure to support business applications.

Copyright (c) 2016-2019 Storage Unpacked.

This week Chris and Martin discuss the subject of storage for home and homelabs.  With so much content being personally generated these days, is it better to store data on local hardware or use the public cloud?  It turns out that many of the home storage devices we’re familiar with are pushing more to SMB and enterprise-class features.

Alternatively, if you’re like Martin, you could build your own solution from drives, Linux and an ATX chassis.  A quick check online shows both HDDs and SSDs for sale.  Are users being careful about wiping their data before selling?  We’ll revisit this subject in a few weeks and find out.

For homelabs, there are so many options.  Home NAS platforms will export iSCSI and NFS.  Alternatively, solutions can be built from open source software.  There are even simulators for many of the older storage platforms.  These provide for the ability to learn about storage solutions or test things out in a non-production environment.

Here are links to the simulators and other software we discussed:

* Dell EMC Unity

* Dell EMC Isilon Simulator

* NetApp ONTAP Simulator (login required)

* Ceph

* Gluster

* Lustre

* Synology Cloud Sync

Elapsed Time: 00:28:15

Timeline

* 00:00:00 – Intros

* 00:00:20 – Blue Monday!

* 00:02:00 – Why use home storage with public cloud availability?

* 00:03:30 – People are starting to store terabytes of personal data

* 00:04:00 – Is storage too cheap to bother being organised?

* 00:04:30 – What are the home NAS options?  Buy or build?

* 00:07:30 – Make sure you backup and backup again!

* 00:09:30 – Can home NAS devices tier to the public cloud?

* 00:11:30 – Turns out Synology is pretty fully featured!

* 00:14:00 – What about encryption and data destruction?

* 00:16:00 – Default encrypt when uploading to cloud?

* 00:17:00 – Checking out cheap SSDs on Ebay

* 00:19:00 – What are the storage options for a home lab?

* 00:21:00 – What could you build with open source?

* 00:22:30 – There are also storage software simulators

* 00:27:00 – Wrap Up

Related Podcasts & Blogs

*  Home Lab: Building VM Images

 

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode F8A8.

This week, Chris and Martin talk to Jonathan Hinkle, Principal Researcher – System Architecture and Master Inventor at Lenovo.  As solid-state storage and NVMe become more prevalent, the legacy 2.5″ HDD format is increasingly impractical for getting the best density, power and cooling in servers.  Jonathan explains how his work led to the founding of the Enterprise and Data Centre Small Form Factor working group and new device formats being introduced into the industry.

EDSFF is responsible for managing device standards like Intel Ruler (see a previous blog episode covering this technology).  How did the standards evolve, which vendors have contributed and how can we expect vendors to integrate these standards into their products?

More information on EDSFF can be found at https://edsffspec.org/

Elapsed Time: 00:33:02

Timeline

* 0:00:00 – Intros

* 00:02:00 – EDSFF – Enterprise and Data Centre Small Form Factor Group

* 00:02:30 – The legacy hard drive format

* 00:04:50 – Remember Fusion-io?  The first AIC card

* 00:06:10 – The HDD form factor was limiting for solid-state devices

* 00:06:50 – The HDD “box” was bad for airflow

* 00:08:30 – Power can be done differently with SSDs

* 00:10:00 – New form factors enable much better server configurations

* 00:10:50 – Where did the new form factors and EDSFF come from?

* 00:13:00 – Lenovo, Intel and Samsung align, MSFT, HPE, Facebook & Dell

* 00:15:00 – Intel brought the Ruler format into the group

* 00:18:00 – How far can the new standard last? 5 years, 7 years?

* 00:20:50 – What determined the device dimensions?

* 00:22:50 – What does that mean for devices in a single server?

* 00:23:00 – Use cases, HCI, hyper scale, disaggregated solutions

* 00:27:00 – Disaggregated solutions like E8 Storage and Excelero could use this tech

* 00:28:30 – Could compute be added to the Ruler format?

* 00:30:00 – Are AIC and HDD solid-state formats effectively dead?

* 00:31:00 – Could new formats be used for old uses, like NICs?

Related Podcasts & Blogs

* #61 – Introduction to NVM Express with Amber Huffman

* Samsung Introduces 8TB NF1 NVMe SSDs

* Has NVMe Killed off NVDIMM?

* Flash Capacities and Failure Domains

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode B222.

On this week’s podcast, recorded live at NetApp Insight 2018, Chris talks to Greg Knieriemen and Rob McDonald about the introduction of Memory Accelerated Data, commonly called MAX Data.  The MAX Data solution is a software product that implements a local file system on a server using local persistent memory such as Intel Optane.  Of course, this is what DAS (Direct Attached Storage) used to offer 20 years ago and this is definitely not what MAX Data provides.  Protection against loss within a server is achieved with a feature called MAX Recovery that synchronously replicates data to another backup server.  Data is also tiered to NetApp AFF storage as an additional level of protection.

The question is how the technology could be used.  The most obvious benefit is to speed up existing applications, however, performance improvements could also result in reducing licensing costs (e.g. enterprise databases) and more efficient use of server hardware (see episode #82 on storage predictions and SCM).  Greg neatly sidesteps the question of using MAX Data and NetApp AI, but this seems like probably the most obvious use case.

For more details on MAX Data, check out the landing page that Rob recommends.  Rob also references a blog post that talks a bit further about database acceleration.  You can find that here.  We talked to Greg about his move to NetApp in episode #78.  You can find that here.

Elapsed Time: 00:25:03

Timeline

* 00:00:00 – Intros

* 00:01:40 – What is MAX Data?

* 00:03:30 – What persistent memory does MAX Data use?

* 00:06:16 – Application transparency, appearing as a local file system

* 00:07:10 – How is MAX Data not DAS?

* 00:10:00 – MAX Data is a real tier of storage

* 00:13:40 – MAX Data integrated with existing shared AFF systems

* 00:15:30 – Where else could the software be used?  Cloud? AI?

* 00:18:30 – Applications don’t need to be rewritten to use MAX Data

* 00:22:10 – Greg ducks the question of using MAX Data with NetApp AI

* 00:24:10 – Wrap Up

Related Podcasts & Blogs

* #82 – Storage Predictions for 2019

* #78 – Thoughts on NetApp with Greg Knieriemen

* #80 – Discussing NetApp’s AI Strategy with Santosh Rao

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode IDUF.

The idea of this episode was to put some structure around a set of enterprise storage predictions for 2019.  As you will hear from the dialog, that’s not quite what we achieved!  However, Chris Evans, Chris Mellor and Martin Glassborow do raise some interesting points on the direction of the industry in 2019.  The conversation starts with a look at media.  QLC flash is likely to be a hot topic, but what about storage class memory?  Have hard drives had their day or is the technology moving into a state of equilibrium?

The conversation moves onto standards and protocols – NVMe in particular.  There’s a real challenge around using Fibre Channel or Ethernet for block storage (Ethernet has already won in the file and object space).  But how exactly will NVMe adoption proceed, especially for public cloud?

Probably the biggest talking point is data management.  Technology improves consistently, however, the challenge of managing data still persists.  What can we expect in 2019, do we need an Open Data Management model?

The wrap up looks at a few vendors and who will be big and who is in trouble.  Possibly the biggest challenge is over-supply in the SAN and filer market, according to Chris M.  We will see!

Elapsed Time: 00:35:56

Timeline

* 00:00:00 – Intros

* 00:00:38 – Who’s Gavin?

* 00:01:15 – Media – QLC gains more traction

* 00:03:00 – Will 2019 be the arrival of SCM?

* 00:04:50 – Has 3D-Xpoint development stagnated?

* 00:05:30 – SCM will drive container adoption

* 00:06:30 – Will HPE put SCM into Synergy?

* 00:08:10 – 2019 – The year of optical!

* 00:09:00 – NVMe (over Fabrics) will be big in 2019

* 00:10:00 – How about NVMe/TCP in Public Cloud?

* 00:13:00 – How does NVMe-oF compare to FCoE?

* 00:18:00 – The Storage Admin is not dead – yet!

* 00:18:30 – An Open Data Management model – like OSI?

* 00:22:00 – Do we need a dominant player in data management?

* 00:23:30 – S3 killed CDMI

* 00:25:00 – Can secondary data solution vendors solve data management?

* 00:27:00 – Which vendors will do well during 2019?  Pure Storage?

* 00:30:00 – Could better focus be important for storage in 2019?

* 00:32:00 – Chris sees some saturation in SAN and filer markets

* 00:33:30 – Summing up – data management is the biggest challenge

* 00:35:10 – Wrap Up

Related Podcasts & Blogs

* #81 -Storage or Data Asset Management

* #79 – Learning More About QLC Storage with Steve Hanna

* #76 – Fibre Channel and NVMe with Mark Jones

* The Biggest Storage Trends of 2019

* Processors back under the spotlight for 2019

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 39A2.

This week Chris and Martin talk about the evolution from storage management to data management.  This follows from recent vendor events where data management featured highly, but still seemed to focus on infrastructure products.  Is there a definition that can bridge the gap – something like Data Asset Management?

The team start by trying to get a handle on what storage and data management actually mean.  In the data protection world, for example, DLP – data leakage prevention, or data loss prevention refer to more advanced versions of simple backup.  Things get more complex when looking at whether the data as a single entity or the content within the data is the thing under management.

What data management work are secondary storage vendors doing?  Can backup appliance vendors be classed as data managers?  Why are storage vendors not making more of their own software and helping to drive product strategies?

In reality, some of this work is already in place.  At the time of recording, episode #77 (HPE Performance Insights) had not been published.  The discussion with Ivan Iannaccone discusses how HPE in particular is using the data from the field to drive better product solutions.

From Dell’s perspective, the acquisition of Data Frameworks represents a step towards data asset management.  A link to the Dell Analyst presentation in Chicago that is mentioned in the podcast can be found here – Unlock The Power of Data.  Companies onstage at the Dell event were:

* GraphCore

* Immuta

* ZingBox

* JASK

* Noodle.ai

Elapsed Time: 00:23:16

Timeline

* 00:00:00 – Intros

* 00:01:30 – Do vendors offer real data management or something else?

* 00:02:49 – How can we define data and storage management?

* 00:04:10 – Data protection, more an advanced storage management task?

* 00:04:40 – DLP – Data Loss/Leakage Prevention

* 00:05:50 – A definition of data management

* 00:07:40 – Are we blurring boundaries between storage admins, DBAs and app owners?

* 00:08:30 – Dell invests in startups – how many are they actually using in their business?

* 00:10:00 – Yet storage vendors have lots of data that could be managed

* 00:12:00 – Do storage vendors drive their own products strategy from data?

* 00:13:00 – How will SDS vendors use storage analytics data?

* 00:15:00 – What about the secondary storage companies, like Cohesity?

* 00:16:30 – Do we need a new term like Data Asset Management?

* 00:18:40 – Vendor acquisitions Boomi & Data Frameworks

* 00:21:00 – Wrap Up

Related Podcasts & Blogs

* #77 – HPE Performance Insights with Ivan Iannaccone

* #75 – It’s ILM All Over Again with Chris Mellor

* #72 – Hitachi Vantara Looking Forward with Shawn Rosemarin

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode FD9C.