Technology:Unpacked

This week, Martin and Chris talk to Krishna Subramanian, President and COO at Komprise.  We mentioned the Komprise technology back on episode #75 (It’s ILM All Over Again) as part of a discussion on managing the movement of files to and from archive storage.  Krishna joins this discussion to fill in some background on the challenges of managing unstructured data, including how to implement solutions with as little lock-in as possible.

Why do we need to manage data in the first place?  With around 75% of all data having no access in over 12 months, there’s clearly a cost saving for primary storage.  This extends further into eliminating backup copy duplication.  Of course, efficiencies can also be gained in performance as well as converting capital costs to operational ones.

More details on Komprise can be found at https://www.komprise.com.

Elapsed Time: 00:36:26

Timeline

* 00:00:00 – Intros

* 00:02:00 – Where is new data coming from?

* 00:03:00 – What problems exist with unstructured data?

* 00:05:00 – Unstructured – just file and object?

* 00:06:30 – Start with visibility of the data

* 00:09:00 – Does it make sense to use file virtualisation (like DFS)?

* 00:11:00 – End users need to know that data has been archived

* 00:12:30 – Stubs or links?  What’s the difference and benefits?

* 00:15:00 – How is file searching affected?

* 00:16:00 – Data mover VMs scale to manage data movement in Komprise

* 00:18:00 – Does individual file movement work with all types of data?

* 00:21:00 – Where do we keep our metadata – inception?

* 00:24:00 – Is archive the slippery slope to moving to the public cloud?

* 00:26:30 – Do Opex charges encourage users to delete data?

* 00:29:00 – What strategies are being taken to deal with data sprawl?

* 00:31:30 – Continuous data optimisation is an eventuality

* 00:33:40 – What is the future of data management?

* 00:35:20 – Wrap Up

Related Blogs and Podcasts

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

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

* The $2 billion Gamble on Data Management

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 576B

This week, Chris and Martin discuss the announcements from Dell Technologies World 2019, held in April/May 2019 at the Sands Convention Centre in Las Vegas.  Chris attended as a guest of Dell Technologies, who covered flights and accommodation.

Dell Technologies made a range of announcements, including improved capabilities on the Unity XT and Isilon storage platforms.  A new data protection platform called PowerProtect was introduced and PowerMax gained dual-ported Optane drives that allow a new tier of persistent media storage.

One of the most interesting areas was the announcement of more SaaS and cloud-based offerings, including Cloud Storage Services and Dell Technologies Cloud.  The former is co-located storage that can be used for features like disaster recovery.  The latter is the delivery of Project Dimension and the capability to ship and run a Dell cloud based on VxRAIL, right in the customer’s data centre.

The idea of SaaS-based cloud delivery brings up many questions, not least where companies like Dell and HPE are headed, as cloud becomes more dominant.  What does the future hold for Dell Technologies and other infrastructure companies?

Elapsed Time: 00:34:13

Timeline

* 00:00:00 – Intros

* 00:00:50 – Is the EMC brand vanishing?

* 00:02:30 – How big was DTW 2019?

* 00:04:40 – What did Michael say?

* 00:06:00 – What were the storage announcements?

* 00:10:30 – New co-located storage as a service

* 00:11:30 – Dell Technologies Cloud was new offering

* 00:14:30 – Is on-prem SaaS hardware the right step forward?

* 00:17:28 – Dell, VMware and Microsoft best of friends

* 00:18:25 – What’s the benefit of running vSphere on Azure?

* 00:20:50 – Where is Dell headed?  It’s all about VMware.

* 00:25:25 – Could Dell be the Apple of Enterprise Servers?

* 00:28:00 – Will IoT and 5G be the saviour of infrastructure companies?

* 00:29:00 – What’s the future for infrastructure companies like Dell & HPE?

* 00:33:00 – Wrap Up

Transcript

 

Related Podcasts & Blogs

* #98 – Shared Storage for Scale-out Databases with Walt Hinton from Pavilion Data

* #90 – Dell EMC’s Enterprise NVMe Strategy with Vince Westin

* Garbage Collection #004 – Dell EMC World Preview, DockerCon 2017 and The Return of Violin Memory

* Will VMware lead the newly public Dell Technologies?

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 0431

In this third recording from Dell Technologies World 2019, Chris talks with Walt Hinton, Head of Corporate and Product Marketing at Pavilion Data Systems.  The topic of conversation is the alignment of shared storage with scale-out NoSQL databases that typically place storage components on separate local disks.

As rack-scale computing becomes more widely adopted, local storage will migrate back into bottom-of-rack shared storage infrastructure.  In this instance, does it make sense to put shards of databases back on the same hardware?  Walt thinks it is the next logical step in optimising scale-out applications and introducing what could be classed as SAN 2.0.  Shared rack-scale storage provides the capability to reduce east-west rebuild traffic and the associated risk exposure that failures introduce.

Walt also discusses performance improvements that have been seen with the introduction of release 2.2 of the software that drives the Pavilion storage platform – 90GB/s throughput at 40µs latencies.  You can find out more about Pavilion Data and their NVMe-oF Storage Platform at https://paviliondata.com/.

Elapsed Time: 00:17:58

Timeline

* 00:00:00 – Intros

* 00:00:50 – What is the Pavilion Data Platform?

* 00:01:30 – How are requirements for shared storage evolving?

* 00:04:00 – Rebuild failures become a problem with scale-out architectures

* 00:06:30 – Disaggregation with NVMe enables rack-scale architectures

* 00:08:30 – Should we be reconsolidating sharded database applications?

* 00:11:30 – What are the options in rack-scale designs?

* 00:13:30 – Release 2.2 of Pavilion software delivering greater performance

* 00:15:00 – Disaggregated solutions aim to get out of the data path

* 00:17:00 – Wrap up

Transcript

 

Related Podcasts & Blogs

* #89 – Choices in NVMe Architectures

* #59 – Ethernet vs Fibre Channel

* The Biggest Storage Trends of 2019

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 1C40

This episode was recorded live at Dell Technologies World 2019.  Chris talks to Kam Eshghi, VP of Strategy & Business Development at Lightbits Labs.  Lightbits has pioneered the development of NVMe/TCP, which allows standard Ethernet NICs to be used for NVMe-over-Fabrics storage traffic.

The ability to implement NVMe-oF without custom NICs (like RDMA) offers the potential to deploy NVMe fabrics in hyper-scaler and enterprise data centres with lower cost and complexity compared to existing solutions, but what are the drawbacks?  Kam explains how NVMe/TCP can be implemented on high-speed networks that even include multiple switch hops or operate between data centres.  The solution offers significant benefits with disaggregated and composable infrastructures where compute and storage can be mapped dynamically across the network.

Lightbits implements NVMe/TCP as software or as a storage array (SuperSSD), which includes an optional PCIe card to offload some more CPU-intensive tasks.  You can find more information at https://www.lightbitslabs.com/.

Elapsed Time: 00:20:24

Timeline

* 00:00:00 – Intros

* 00:01:00 – What is NVMe/TCP?

* 00:04:00 – Why use NVMe/TCP instead of RDMA?

* 00:07:00 – NVMe/TCP will work across multiple network hops and switches

* 00:08:00 – How is network performance managed?

* 00:09:30 – How would NVMe/TCP be deployed?

* 00:12:20 – Reference code and an SPDK is available for target-mode NVMe/TCP

* 00:14:00 – How are enterprise customers using NVMe/TCP?

* 00;16:30 – Teaser – vSAN could be enabled with NVMe/TCP

* 00:17:40 – Wrap Up and a view of the future

Transcript

 

Kam’s Bio

Kam Eshghi, VP of Strategy & Business Development, Lightbits Labs  Kam joined Lightbits Labs from DellEMC and has over 20yrs of experience in strategic marketing and business development with startups and public companies. Most recently as VP of strategic alliances at startup DSSD, Kam led business development with technology partners and developed DSSD’s partnership with EMC, leading to EMC’s acquisition of DSSD. Previously as Sr. Director of Marketing & Business Development at IDT, Kam built their NVMe Controller business from scratch. Previous to that Kam worked in data center storage, compute and networking markets at HP, Intel, and Crosslayer Networks. Kam is a U.C. Berkeley and MIT graduate with a BS and MS in Electrical Engineering and Computer Science and an MBA.

Related Podcasts & Blogs

* NVMe over Fabrics – Caveat Emptor?

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

* #82 – Storage Predictions for 2019

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 527A

In this podcast, recorded live at Dell Technologies World 2019, Chris talks to Rob Davis, VP of Storage Technology at Mellanox, about the development of SmartNIC technology and storage.  SmartNICs offload tasks such as encryption, compression and protocol traffic from core CPUs.  This allows system vendors to create more efficient and cheaper products.  It can also provide a backward compatibility for legacy applications and operating systems.

Rob takes us through the implementation of SmartNICs on both the initiator (host) and target (array) side of the storage network.  We discuss vendors using the technology and what we can expect in terms of evolution in SmartNICs in the coming years.

You can read more about the Mellanox BlueField  and Innova SmartNIC ranges here – https://www.mellanox.com/products/smartnic/.

Elapsed Time: 00:16:34

Timeline

* 00:00:00 – Intros

* 00:00:42 – What are SmartNICs?

* 00:02:00 – Network offload has been around forever

* 00:02:44 – Additional components can be SoC, FPGA, processor

* 00:04:00 – For storage, SmartNICs could be Initiators or Targets

* 00:06:00 – How is storage accelerated at the initiator end?

* 00:08:50 – Hyperscalers are using SmartNICs with their own software

* 00:10:00 – How much cost do SmartNICs add?

* 00:11:50 – Where are vendors using the technology?

* 00:12:30 – SmartNICs are great for disaggregated storage solutions

* 00:15:00 – Where do we think SmartNICs are headed?

* 00:15:23 – Wrap Up

Transcript

Related Podcasts & Blogs

* #89 – Choices in NVMe Architectures

* #76 – Fibre Channel and NVMe with Mark Jones

* #61 – Introduction to NVM Express with Amber Huffman

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 0E59

This week, Chris E is joined by Chris Mellor and returning guest Eran Brown.  The team discuss the usefulness of storage vendor guarantees like de-dupe ratios and uptime.  Do vendors use these metrics as a marketing tactic to attract new customers or is there any value in them?

The idea of a sales/marketing guarantee sounds appealing.  Storage vendors have been using them for years to entice customers and offer some sort of incentive to buy.  Most recently we’ve seen de-dupe and compression ratios being a battleground for optimising the cost of all-flash arrays.  Digging down, what’s actually on offer?  How do vendors quantify what these offerings mean?  What’s the penalty if the vendor doesn’t meet the agreement?  Should you get these guarantees written into contracts?

More details on INFINIDAT’s Always On guarantee can be found here – https://www.emea.infinidat.com/always-on-guarantee.

Elapsed Time: 00:39:34

Timeline

* 00:00:00 – Intros

* 00:01:00 – What is a guarantee and how do they work?

* 00:03:30 – What are storage vendors offering?

* 00:05:00 – Are these simply marketing guarantees?

* 00:06:50 – With de-dupe – how is the X:1 ratio calculated?

* 00:08:00 – Should all data be accepted for dedupe ratios?

* 00:09:00 – What does 100% data availability mean?

* 00:10:30 – Do guarantees actually mitigate business risk?

* 00:12:00 – How do vendors ensure they can pay out on guarantees?

* 00:16:00 – Vendors can offer safe guarantees if they know their systems are reliable

* 00:18:00 – Uptime and availability are not necessarily the same thing

* 00:20:00 – How should vendors compensate customers for guarantee failures?

* 00:24:30 – Does Chris Mellor have a disaster recovery car?

* 00:28:30 – Financial regulation may impact giving customer money back

* 00:30:00 – How can we know if vendor guarantees are paying out (or not)?

* 00:32:00 – How should guarantees be contractually managed?

* 00:35:00 – Should guarantees affect your purchasing/consumption model?

* 00:37:40 – Wrap Up

Related Podcasts & Blogs

* #58 – Storage Vendor Hero Numbers

* Garbage Collection – Implementing a Multi-Vendor Storage Strategy

* Evolving Enterprise Storage Buying Models

 Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 8D7A

In this episode, Martin and Chris talk to Andy Klein from Backblaze.  Since 2014, Backblaze has been collecting and publishing hard drive reliability information for their storage pods.  The pods are custom-built and used to hold backup data from customers and to support the B2 cloud storage offering.  Rather than buy an expensive storage platform or use the public cloud, Backblaze initially decided to build their own solution at a fraction of the cost of other options.  Once at scale, it seemed logical to use the platform for B2.

Over the years, the Backblaze team has collected statistics on thousands of hard drives and dozens of models and capacities.  The data shows some interesting statistics, but why collect the information and how is it used?

Andy explains the rationale behind collecting the data and some of the benefits the information provides.  Data for 2018 can be found here – https://www.backblaze.com/blog/hard-drive-stats-for-2018/.  You can see the generations of pods on this recent podcast – https://www.backblaze.com/blog/backblaze-storage-pod-museum/

Andy also references the source data, which is freely downloadable (although quite sizeable now).  You can use this shortlink – https://brk.su/bb-hdd-stats or the full link URL – https://www.backblaze.com/b2/hard-drive-test-data.html.

Elapsed Time: 00:43:35

Timeline

* 00:00:00 – Intros

* 00:01:10 – Who is or are Backblaze?

* 00:02:20 – DIY Storage Pods

* 00:05:30 – What software did Backblaze use?

* 00:07:00 – How does drive supply affect builds?

* 00:11:00 – Enterprise vendors use consumer drives

* 00:11:55 – Backblaze stats are collected from SMART

* 00:15:00 – Data is computed using drive days and AFR

* 00:21:00 – What interesting information did the data show?

* 00:25:50 – Do drives correlate in failures or fail randomly?

* 00:28:08 – The bathtub failure curve is true!

* 00:30:00 – How did Backblaze use the data to improve the service?

* 00:33:30 – What relationship does Backblaze have with drive vendors?

* 00:36:00 – Will SSD find it’s way into Storage Pods?

* 00:39:00 – What about using dual actuator drives?

Related Podcasts & Blogs

* #85 – Storage for Home and Homelabs

* #55 – Storage for Hyperscalers

* HDD Capacity Threshold Reaches 15TB

* Backblaze HDD Statistics for 2017

 Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode D06C

This week, Chris and Martin discuss the issues at Myspace, which recently disclosed that 12 years’ worth of user content had been lost during a (failed) server migration.  The once-mighty Myspace was the largest social networking site from 2005 to 2009 (according to Wikipedia) and had estimated revenues of $109 million in 2011.  So, how could a company with such as large valuation and solid revenue manage to lose data so easily?

In 2005, News Corporation purchased Myspace for $580 million, later selling he company in 2011 for a rumoured $35 million.  Would there have been due diligence during the acquisition?

The discussion continues to look at exactly how much trust we can have in free sites that use other revenue sources, like advertising.  What about free tiers from cloud storage providers like Dropbox?  Do they offer the same level of guarantee as paid customers?  Finally, the conversation wraps up with some recommended best practices.

Elapsed Time: 00:24:05

Timeline

* 00:00:00 – Intros

* 00:01:00 – What was Myspace?

* 00:02:30 – What was the data loss issue?

* 00:04:10 – Surely social media companies don’t single instance?

* 00;04:30 – Was this a long-term problem?

* 00:06:30 – Should you have rights on a free service?

* 00:09:30 – Wouldn’t data protection be picked up during acquisition due diligence?

* 00:13:00 – Data protection should be part of the initial platform design

* 00:15:00 – Don’t rely on one copy – keep at least three….

* 00:16:30 – What if a vendor decides to block your data access?

* 00:19:00 – So what are best practice recommendations?

* 00:23:30 – Wrap Up

Related Podcasts & Blogs

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

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

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

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 6453.

This week, Chris talks to Jerome McFarland, VP of Marketing with Elastifile, Inc.  Elastifile has developed a scale-out distributed file system called ECFS – the Elastifile Cloud File System. ECFS runs either on-premises, in the public cloud as a marketplace offering or, most recently as a full native storage service in GCP.  In addition to the base file system, Elastifile provides the capability to move any on-premises NAS data into the public cloud with CloudConnect.  ClearTier provides the facility to use the public cloud for tiering of cool or inactive data.

Jerome explains the reasons for developing a cloud-native file system and how the use of standard instances such as those in GCP allow ECFS to be deployed across any public cloud without additional proprietary hardware.  The conversation expands to touch on use cases, both for microservices and traditional applications.

You can find more about Elastifile at www.elastifile.com or try out Elastifile with either of the two links:

* Try Elastifile as a fully-managed service on Google Cloud

* Try Elastifile on AWS Marketplace

Elapsed Time: 00:31:17

 

Timeline

* 00:00:00 – Intros

* 00:01:00 – Where did Elastifile come from?

* 00:04:30 – Why has scale-out been such a challenge to develop?

* 00:06:00 – Three components ECFS, ClearTier and CloudConnect

* 00:07:20 – How does Elastifile Cloud File System work?

* 00:09:00 – ECFS on GCP uses standard instances – no bespoke hardware

* 00:11:00 – How does scalability and adding capacity work?

* 00:12:30 – How does ClearTier work?

* 00:15:00 – CloudConnect provides a way to move data from existing NAS

* 00:17:00 – How are customers aggregating data into the public cloud?

* 00:18:30 – Customer examples – media and entertainment

* 00:21:30 – How does Elastifile support micro-services?

* 00:24:30 – Is there a place for supporting traditional applications?

* 00:27:00 – What’s in the future for Elastifile – object support?

* 00:30:00 – Wrap Up

Related Blogs and Podcasts

* #56 – Defining Scale-out Storage

* Elastifile Cloud File Service Now Available on Google Cloud

* Elastifile presentations at Tech Field Day

Jerome’s Bio

Jerome leads Elastifile’s marketing mission and dedicates himself to sharing the transformative value that Elastifile delivers to its customers. He’s passionate about innovation and has extensive experience marketing and managing cutting-edge software and hardware solutions in the technology sector.  Prior joining Elastifile, Jerome has held senior marketing and product management positions at Maxim Integrated, Diablo Technologies, and OCZ/Toshiba, and began his career as a software engineer at Oracle Corporation. Jerome holds Bachelor of Science degrees in EECS and Literature from MIT and a Master of Engineering degree in EECS, also from MIT.

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 93A6.

This week, Martin leads the conversation as Chris discusses the vendors presenting at the recent Storage Field Day 18 event.  SFD18 is part of the Tech Field Day series delivered by Stephen Foskett and GestaltIT.  We last spoke with Stephen in April 2018 – an episode on why storage isn’t boring.

Although there’s no official theme for SFD events, the companies presenting were nicely delineated into scale-out primary storage and data protection.  WekaIO demonstrated their scale-out file system.  VAST Data came out of stealth and also presented a scale-out file system and object store.  Storpool presented on block storage.  Western Digital talked about their end-to-end storage portfolio.  NetApp previewed NDAS.  IBM talked about Spectrum Protect Plus.  Cohesity demonstrated the latest features of their data protection and archive solution.  Finally, Datera presented their scale-out block storage platform.

If you’re interested in being a delegate, check out the delegates page on the TFD website.  If you would like to be a sponsoring vendor, more details are available here.

Videos from Tech Field Day events are posted online on both YouTube and Vimeo.  The YouTube SFD18 playlist (28 videos) can be found here.

Elapsed Time: 00:36:02

Timeline

* 00:00:00 – Intros

* 00:01:00 – What is Storage Field Day?

* 00:02:30 – Dave Hits’ last day at NetApp

* 00:04:30 – Scale-out storage – WekaIO, VAST Data, Storpool, Datera

* 00:07:00 – What is the enterprise looking for from storage?

* 00:07:30 – It’s 2019, do we need another new storage platform?

* 00:13:00 – And now, data protection, Cohesity

* 00:14:50 – Did IBM think of running applications on Shark? Risky.

* 00:16:00 – Would it be best to run analytics in public cloud?

* 00:19:00 – Is Cohesity developing high performance nodes?

* 00:21:00 – IBM – what is their storage and data protection strategy?

* 00:25:00 – NetApp presented on NDAS

* 00:28:00 – NDAS may expose APIs to mine backup data

* 00:30:30 – What did Western Digital present on?

* 00:33:00 – How can listeners learn more about TFD/SFD?

Related Podcasts & Blogs

* #47 – Enterprise Storage is Not Boring

* #52 – An Introduction to WekaIO Matrix with Liran Zvibel

* VAST Data launches with new scale-out storage platform

* Exploiting secondary data with NDAS from NetApp

Copyright (c) 2016-2019 Storage Unpacked.  No reproduction or re-use without permission. Podcast Episode 607B.