Technology:Unpacked

This week, Chris and Martin reflect on changes in the storage industry and the ways in which vendors can make money from storage hardware and software. It’s been a tough time for storage vendors of late and the discussion starts by looking at the challenges of making money in a market where components are commodity and decline in price each month.

How should vendors adapt to these challenges? There are alternative strategies than simply piling it high and selling it cheap (although that is one model). All-flash systems offered the ability to charge higher margins, but even this approach is being challenged with the availability of cheaper and higher capacity flash drives.

Another option is to change the financial model and sell either based on consumption or entirely as a service. These options themselves are tricky to implement because the vendor has to persuade the customer to lengthy commits or second guess how the customer will use the hardware onsite.

Finally, Chris and Martin discuss the idea that perhaps a “pure play” storage vendor isn’t a long-term strategy and really, what customers want to buy is solutions. This means packaging CI, HCI or even containerised solutions that integrate with storage platforms.

Elapsed Time: 00:37:25

Timeline

* 00:00:00 – Intros 00:01:50 – This is a tough market for storage vendors * 00:03:00 – Hardware is good enough to build your own* 00:03:45 – HDDs are capacity-only products* 00:04:30 – The flash market is also on a price reduction curve* 00:05:00 – Are the 3-year refresh cycles over with flash systems?* 00:06:30 – Has Open Source had any impact on enterprise storage prices?* 00:07:40 – So how can vendors make more money?* 00:09:15 – Support organisations have been cut back – why?* 00:11:00 – Is the price of consumer drives over-priced to compensate for hyperscalers?* 00:12:30 – NetApp are releasing a SAN-only version of ONTAP* 00:14:15 – What is “cheap and deep SAN”?* 00:17:20 – On-premises storage allows businesses to “sweat the asset” more* 00:18:15 – Cloud is changing the model of consumption 00:20:00 – How much will vendors make out of hybrid?* 00:23:00 – How about economies of scale – just sell more stuff?* 00:24:40 – New hardware components could allow vendors to add value* 00:28:45 – New pricing models (SaaS, Subscription) could change vendor revenue* 00:30:30 – Perhaps a pure-play storage company is a bad idea? * 00:32:20 – Own the architecture, ecosystem, platform 00:35:30 – Wrap Up

Related Podcasts & Blogs

* Pure Storage Cloud Block Store Goes GA* Building a Golden Data Repository* HDD Capacity Threshold Reaches 20TB* Is AWS passing on the benefits of storage media price reductions?* On-premises infrastructure – as a service* #119 – Storage Hardware is Back!

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #cg08.

This week Chris records a live podcast episode with Niraj Tolia, co-founder and CEO at Kasten Inc. As a follow-up to online discussions that dive into how container storage should be presented to Kubernetes (and other container) environments, this episode reviews some of Chris’ thoughts and Niraj’s opinions on the market as to how persistent storage should evolve.

The discussion starts by reviewing the idea of either block or file storage for container environments and moves on to discuss whether application-based replication could simply replace the need to have a persistent storage back-end. The Container Storage Interface (CSI) is one way vendors are trying to address the issue of storage abstraction, but it seems the technology still has some way to go in terms of metadata management.

The conversation moves next to thoughts around delivering more than more than just block and file storage to container environments. Perhaps there’s a need to deliver higher-level storage functionality (like structured data) to applications as part of the container management framework.

Finally the conversation wraps up with a discussion on Day 2 issues – managing security, audit, compliance and other operational challenges in an enterprise environment.

You can find more about Kasten and their data protection solution for containers at kasten.io.

Elapsed Time: 00:36:03

Timeline

* 00:00:00 – Intros* 00:02:10 – Should we use block or file for container storage?* 00:03:55 – Deeper dive, how are block devices presented?* 00:05:40 – Most K8S block devices are shared in single reader mode* 00:07:20 – Raft & PAXOS provides consensus mechanisms to manage LUNs* 00:08:20 – Should we just rely on application-based replication?* 00:10:15 – CSI – Container Storage Interface* 00:12:30 – Metadata management – still an issue* 00:16:00 – Are virtual storage arrays a solution?* 00:18:55 – Should container environments provide more “plumbing” to the application?* 00:19:40 – Storage arrays should be simpler….* 00:21:00 – Other storage protocols & solutions are available* 00:24:00 – Platform storage – K8S could provide higher level storage services* 00:26:30 – Day 2 – how will enterprises meet security & audit concerns?* 00:32:00 – Google’s Query of Death!* 00:32:45 – Shared DAS – with NVMe – could be the future of container storage* 00:34:50 – Wrap Up

Related Podcasts & Blogs

* #53 – Kubernetes and Persistent Storage with Evan Powell* What is the Container Storage Interface (CSI)?* Should I Be Backing Up Containers?* Block is Not the Solution for Persistent Container Storage

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #tmpf.

This episode was recorded live (as you can tell from the background noise) at Commvault GO, held in October 2019. Chris talks to Glenn Dekhayser, Field CTO at Red8 about his reflections on the announcements from the event.

The biggest news from the show was arguably the release of Metallic, a SaaS data protection offering for midrange customers. You can learn more about the specifics from the links at the end of this post. Glenn walks us through his opinion on how Commvault is adapting to change and releasing new products like Metallic to an evolving customer base.

How will Commvault use products like Activate to move to true data management? Some of this comes from features like the ability to delete data from existing historical backups.

With ESG claiming 56% of respondents in a recent survey indicating that they are considering or actively planning to change their data protection solution, vendors like Commvault need to keep evolving and adapting their solutions to customer needs. Data protection continues to be an active space and one in which that vendors simply can’t sit still and rest on their laurels.

Finally, thanks Glenn for the most succinct wrap up of the year…

Elapsed Time: 00:20:04

Timeline

* 00:00:00 – Intros* 00:01:00 – What was announced at Commvault GO?* 00:03:20 – Perceived complexity or  a lot of technology?* 00:03:50 – Metallic as an idea to expand the TAM* 00:06:20 – Why use partners to deliver Metallic?* 00:08:44 – Metallic allows for an on-site copy of backup data* 00:10:40 – How does Activate fit in?* 00:11:30 – Yes it’s GDPR, Glenn!* 00:14:30 – How is Commvault’s data management strategy evolving?* 00:17:10 – What can customers expect to see? * 00:18:00 – 56% of people intending to change out their backup system in 2019* 00:19:50 – Wrap Up

Glenn’s Bio

For over 25 years, I have been a technical and business process consultant to our clients, with a particular expertise focus in Enterprise Storage, Disaster Recovery, Holistic Data Management and IT Security. My varied background as a developer, DBA, network engineer, infrastructure architect, and CTO allow me to not only understand the need from the client’s particular perspective, but to be able to make the client understand his/her solution options . I find great joy in developing comprehensive yet simple technical solutions to complex business requirements, and having clients tell me years later that the solutions I recommended still apply to their evolved needs. I post thoughts on the industry at http:\\thedatamachine.wordpress.com

Related Podcasts & Blogs

* #124 – Initial Thoughts on Commvault Metallic with Chris Mellor* Commvault Announces Metallic SaaS Data Protection

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #e6bb.

This is the third in a series of podcasts recorded earlier this year at Hitachi NEXT 2019 in Las Vegas. This episode is a discussion on updates to the Hitachi Vantara UCP CI & HCI solutions portfolio with Chris Gugger (Director, Infrastructure Marketing) and Ed Reidenbach (Senior Director, Product Management).

Hitachi produces two updates a year, covering spring and autumn (fall). The fall update 2019 introduces support for Cascade Lake processor architectures, VSP 5000 storage and Arista switches. Hitachi has updated UCP Advisor to cover multi-VMware and bare metal environments.

The interesting aspect of UCP CI & HCI infrastructure is the focus on solutions. Both platforms are used within Hitachi Enterprise Cloud (on-premises private cloud) and now target containerised, virtualised, big data and traditional application workloads. In common with many vendors, the aim is to “own” the infrastructure and be able to provide end-to-end coverage for all customer requirements. In this respect, the ability for management software to work well across all technologies will be a key differentiating factor.

You can find out more on Hitachi Vantara CI & HCI solutions here (link).

Elapsed Time: 00:25:38

Timeline

* 00:00:00 – Intros* 00:01:00 – HCI/CI – Fall Launch (spring & fall cadence)* 00:02:30 – Is there a division between CI & HCI?* 00:04:30 – Cascade Lake and VSP 5000 support for latest CI solutions* 00:05:15 – Addition of Arista switches to reference architectures * 00:05:40 – Stretched VMware vSAN clusters* 00:06:30 – Enhancements to UCP Advisor, multi-VMware & bare metal * 00:08:55 – Hitachi Enterprise Cloud uses Hitachi CI/HCI technology  * 00:10:20 – Hitachi uses OpenShift for container deployments* 00:12:00 – VDI solutions can use HCP for unstructured data* 00:14:30 – Hardware partitioning servers for SAP* 00:17:30 – New NVIDIA Tesla GPU solutions* 00:20:30 – Both CI & HCI are growing within Hitachi* 00:21:55 – How flexible are workload migrations between CI/HCI? * 00:25:00 – Wrap Up

Related Podcasts & Blogs

* #125 – Hitachi Next 2019: Ops Center* #123 – Hitachi Next 2019: VSP 5000* Hitachi Vantara Introduces VSP 5000 Series Storage Arrays

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #f082.

Disclaimer: Chris was invited to NEXT 2019, with Hitachi covering travel and accommodation costs. There is no obligation to blog or produce podcasts from the event.

This week, Chris and Martin talk to Datera CTO, Hal Woods about the state of the software-defined storage (SDS) market. SDS emerged around ten years ago as a solution to overcome the expense of buying storage appliances. With cheap hardware, IT organisations could simply buy software or use one of many open-source solutions.

Has SDS been successful and how has it evolved? Early modes of deployment simply ran existing storage software on commodity hardware. As the category developed, software defined storage has been built specifically to address a new set of requirements. These are typically based around offering more dynamic and flexible ways to implement persistent storage.

How has public cloud changed the perception of SDS? Clearly, IT teams are much more comfortable with storage as a service and solutions that deliver to standard metrics like latency, throughput and performance. Cloud challenges businesses to look at their TCO and baseline costs to understand the right solution for their needs.

You can find more about Datera’s software-defined software solution at https://datera.io/. Datera is also hosting a symposium to bring together vendors around the topic of the software-defined data centre. More details are available here – https://sddcsymposium.com/

Elapsed Time: 00:30:03

Timeline

* 00:00:00 – Intros* 00:01:10 – Where did SDS start?* 00:02:30 – Were vendors fairly marking up hardware?* 00:03:10 – SDS solutions are now “purpose built”* 00:04:45 – Hardware & software in SDS need to evolve separately* 00:05:30 – We’re seeing new data, not just storage requirements* 00:08:00 – Storage requirements are now much more dynamic* 00:10:20 – Is SDS more flexible in adopting new technologies?* 00:13:41 – How has public cloud affected our view of SDS?* 00:16:20 – Do new purchasing models mitigate appliance-based storage?* 00:18:32 – Hybrid & multi-cloud solutions are hard to deliver* 00:22:54 – Is data becoming the “centre” of IT?* 00:23:51 – What strategy should IT departments take in using SDS?* 00:26:48 – New projects are typically cloud/SDS-based.* 00:29:00 – Wrap Up

Related Podcasts & Blogs

* Will TCO Drive Software-Defined Storage?* Object Storage Leading the Advancement of Software Defined Storage* Defining Software Defined Storage

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #34a3.

This is the second recording from the Hitachi NEXT 2019 conference held in Las Vegas in October 2019. In a previous episode we looked at the new VSP 5000 platform. In this discussion, Chris talks to Stan Stevens, Director of Infrastructure Software Product Marketing, about Ops Center, the new platform for Hitachi storage management.

Ops Center has four main components; Analyzer provides insights into systems through the use of telemetry collated across the storage footprint. Automator provides workflow orchestration and integrates with common automation tools. Administrator provides the tools to provision storage. Instance Director implements Copy Data Management.

It’s reasonable to say that storage management software hasn’t been Hitachi’s strong point. Hopefully, the release of Ops Center is a move towards streamlining and optimising the effort and challenges involved in operating large storage estates. With this in mind, Stan describes how Ops Center allows chatbots to help support staff deal with common problems and challenges as they arise. For more details on Ops Center, check out this link – here.

Elapsed Time: 00:20:29

Timeline

* 00:00:00 – Intros* 00:01:15 – How good is Hitachi’s record on software & management?* 00:02:30 – Four pieces to Ops Center* 00:05:00 – How much data is being centralised for analysis?* 00:06:10 – What integrations are there with HCI & CI platforms?* 00:07:00 – Platforms can be fully driven by API* 00:08:50 – What community access is there to automation scripts?* 00:12:40 – How will software updates be rolled out?* 00:14:30 – Visualising multiple data centres & devices – ViewPoint* 00:16:30 – Chatbot to leverage APIs!* 00:17:30 – How are policy thresholds set?* 00:18:30 – How does Ops Center fit into a managed services environment?* 00:19:55 – Wrap Up

Related Podcasts & Blogs

* #123 – Hitachi Next 2019: VSP 5000* Hitachi Vantara Introduces VSP 5000 Series Storage Arrays

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode #68a4.

Disclaimer: Chris was invited to NEXT 2019, with Hitachi covering travel and accommodation costs. There is no obligation to blog or produce podcasts from the event.

This episode was recorded “live” at Commvault GO in Aurora, Colorado, close to Denver airport. Chris Evans and Chris Mellor discuss the announcement of Metallic, a new SaaS data protection solution from Commvault targeted at mid-market customers.

Metallic has been advertised as a Commvault “venture”, establishing a separate brand and website for the platform. Initially, Metallic will support Office 365, endpoint, file and on-premises virtual machine data protection. Prospective customers can choose to keep data for on-premises backups onsite as well as creating a copy in the public cloud.

Chris & Chris talk through what it means to establish a separate brand, comparing the recent DDN acquisitions and how those companies are being positioned. The conversation also covers exactly how Metallic has been implemented. It’s not enough just to “lift and shift” software to the cloud. Platforms need to be engineered to be “cloud-native”. The guys also discuss what the implications are for Hyperscale, Commvault’s appliance platform that also has a mid-market focus.

Elapsed Time: 00:19:20

Timeline

* 00:00:00 – Intros* 00:01:20 – What is Metallic?* 00:02:30 – No lift and shift – but using Commvault IP* 00:03:30 – Is the mid market looking to move to SaaS backup?* 00:04:20 – Isn’t this what Hyperscale was aimed at?* 00:06:10 – Why can’t you “lift and shift”?* 00:08:24 – Metallic is a Commvault “venture”* 00:09:30 – Metallic is a partner sale* 00:11:50 – Does separate branding make sense?* 00:13:00 – Has any backup company become a data management company?* 00:15:50 – Does Commvault have to act like a hyperscaler?* 00:17:30 – Are vendors building silos of secondary data? * 00:19:00 – Wrap Up

Related Podcasts & Blogs

* Unpicking the Commvault Acquisition of Hedvig* #38 – Talking Dedicated Backup Appliances with Don Foster

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode fab9.

This is the first of a series of podcast episodes recorded at Hitachi NEXT 2019 in Las Vegas. Hitachi announced the next generation of VSP storage arrays, the VSP 5000 at the show. The new platform provides greater performance and capacity than previous systems at lower latency. Chris meets with Colin Gallagher (VP Product Marketing for the Infrastructure Portfolio) and Dan McConnell (SVP of Infrastructure Product Management) to discuss the details.

VSP 5000 offers up to 21 million IOPS at 70µs and up to 69PB of internal capacity. This level of performance can be delivered with flash SAS drives. VSP 5000 is also capable of virtualising other platforms. This is particularly useful with older systems as it provides workload acceleration and the ability to exploit compression and deduplication onto those arrays.

Finally, the discussion covers the new release of Ops Center, which will be discussed in greater length in another podcast episode.

More details on VSP 5000 can be found here – https://www.hitachivantara.com/en-us/products/storage/all-flash-hybrid-flash-storage/vsp-5000-series.html

Elapsed Time: 00:21:09

Timeline

* 00:00:00 – Intros* 00:01:00 – What is the VSP 5000?* 00:02:00 – Hitachi Accelerated Fabric* 00:03:45 – Hero Numbers, 21 million IOPS, 70µs* 00:05:40 – 70µs with SAS flash drives* 00:07:30 – VSP 5000 as a foundation for other solutions?* 00:09:00 – SVOS could be more software-defined in the future* 00:10:45 – Eight 9’s of availability?  Not 100%? (only joking)* 00:12:00 – 69PB in a single box, 279PB of virtualised capacity* 00:13:00 – How is storage virtualisation being used today?* 00:15:00 – Ops Center, new array management software* 00:17:15 – Automation & “self-healing” as a goal* 00:19:30 – Wrap Up

Related Podcasts & Blogs

* #58 – Storage Vendor Hero Numbers* #103 – Data Management and DataOps with Hitachi Vantara

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode c01b.

Disclaimer: Chris was invited to NEXT 2019, with Hitachi covering travel and accommodation costs. There is no obligation to blog or produce podcasts from the event.

This week, Chris and Martin discuss the management of edge computing and storage infrastructure with Phil White, CTO at Scale Computing. Edge Computing isn’t new and there are plenty of examples of platforms and systems that have been running compute at the edge for many years. Just think of technologies like ATMs that can run with a degree of autonomy.

What’s different now is the velocity or real-time nature of computing, combined with the volume of data created and processed in edge locations. Phil describes how Scale Computing deploys thousands of edge clusters to customers, building solutions as efficiently as possible.

Data and storage management is a key challenge, as many sites won’t be visited for up to 6 months, depending on issues experienced. This requires effective monitoring tools and the ability to manage minor issues without site visits.

For more details on Scale Computing, check out their website at https://www.scalecomputing.com/. You can follow the company online at https://twitter.com/scalecomputing and Phil in particular at https://twitter.com/realphilwhite

Elapsed Time: 00:32:56

Timeline

* 00:00:00 – Intros* 00:02:30 – What is Edge Computing?* 00:04:15 – Edge computing has always existed, surely?* 00:06:00 – The velocity (real-time) nature of data processing needs edge* 00:09:45 – How does large-scale edge infrastructure get deployed?* 00:13:10 – Remote management can be a challenge* 00:15:10 – That classic edge computing device – the smartphone* 00:17:15 – How is end user data management – it’s application dependent* 00:19:00 – With thousands of systems, how are failures managed?* 00:22:30 – Good visualisation is necessary to show “fleet” status* 00:24:00 – What happens with mobile equipment – like ship systems?* 00:26:15 – Could Public Cloud help with the data inertia challenges?* 00:31:45 – Wrap Up

Related Podcasts & Blogs

* #71 – IP Expo 2018 – Scale Computing, StorMagic, Nasuni & Nephos* #57 – Storage on the Edge with Scott Shadley* The HCI Definition Obsession

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode d4fa.

This podcast episode continues the discussion with Dr J Metz and focuses on the detail of NVMe 1.4 and NVMe over Fabrics 1.1. Both standards are complete and have been published in the last few months. The conversation today digs deeper into what the new standards offer and specifically how NVMe/TCP becomes “officially” part of the NVMe standard.

The conversation on NVMe/TCP is particularly interesting as it highlights how the protocol will offer significant benefits over iSCSI. The details in the podcast are quite technical and so it’s worth checking out some of the additional background material online. You can find out more on NVMe at the dedicated Architecting IT NVMe microsite.

For more information on the new standards, check out NVMExpress.org. More about Fibre Channel is available at FCIA.org and of course don’t forget to read up on J’s own NVMe content at jmetz.com.

Elapsed Time: 00:32:22

Timeline

* 00:00:00 – What is in the specifications?* 00:01:20 – More internal device & system information – logs* 00:07:00 – Namespace changes, space reservations and so on* 00:09:00 – NVMe 1.4 and PCI 4.0 – what performance improvements will we see?* 00:10:00 – Where has NVMe-oF 1.1 reached?* 00:11:40 – NVMe/TCP becomes “official”* 00:15:55 – SPDK from Intel – drivers in user space.* 00:18:00 – How does NVMe/TCP compare to Fibre Channel or iSCSI? 00:20:25 – How will standards be refactored?* 00:23:05 – Will we see specific enterprise & consumer devices? 00:24:55 – Enterprise devices need I/O protection* 00:27:10 – New features are designed for hyperscalers* 00:28:45 – What’s in the future?  Efficiencies and use case features * 00:30:00 – Wrap Up

Related Podcasts & Blogs

* #120 – NVMe 1.4 Deep Dive Part I with J Metz* #76 – Fibre Channel and NVMe with Mark Jones* #59 – Ethernet vs Fibre Channel* The Expanding Storage Hierarchy

Copyright (c) 2016-2019 Storage Unpacked. No reproduction or re-use without permission. Podcast episode ff52.