Home Blog Page 260

Ditch the switch: stealth-exiting Rockport Networks says it has a better way

Rockport Networks has exited stealth with a switchless datacentre network product that it claims can carry data traffic faster and with better latency than switched networks, yet is a drop-in replacement for Ethernet and InfiniBand cabling and switches with no application software changes.

It says switch-based networks are reaching a dead-end of increased complexity and cost. Matt Williams, Rockport’s CTO, said in a statement: “When the root of the problem is the architecture, building a better switch just didn’t make sense. With sophisticated algorithms and other purpose-built software breakthroughs, we have solved for congestion, so our customers no longer need to just throw bandwidth at their networking issues. We’ve focused on real-world performance requirements to set a new standard for what the market should expect for the fabrics of the future.”

Doug Carwardine, Rockport Networks CEO and co-founder, said: “Rockport was founded on the fact that switching, and networking in general, is extremely complicated. … We made it our mission to get data from a source to a destination faster than other technologies. Removing the switch was crucial to achieve significant performance advantages in an environmentally and commercially sustainable way.”

How it works

Network switch functionality has been distributed to intelligent endpoint network cards — most definitely not called NICs or even SmartNICs — which have an FPGA and connect up to 24 endpoints to a dedicated 1U SHFL (pronounced “shuffle”) optical device using passive cabling. SHFLs need no power or cooling and can be linked together to scale out the network. Ethernet and InfiniBand traffic can be carried over the Rockport network. The network cards are standard, low-profile half height, half length (HHHL) cards fitting in a PCIe slot. 

CTO Matt Williams told us: “People can remove their Ethernet NICs and InfiniBand cards, use the Rockport Network Card, and it just works.”

Video of Matt Williams giving a high-level description of Rockport’s approach.

The Network Card contains a Mellanox ConnectX unit, and that’s what Ethernet and InfiniBand-using applications in servers and other devices see: a Mellanox ConnectX-5 Ethernet NIC, providing verbs (RoCE) and sockets (TCP/UDP) API support. In effect Rockport provides OSI layer 1 and 1.5 networking. As Williams explains: “We transport Ethernet frames but don’t switch them.”

Traffic is divided into small chunks called FLITS by the network card’s Rockport Network Operating System (rNOS) software. These are interleaved and transmitted together so that large frame or packet network transfers don’t hold up smaller ones, eliminating the noisy neighbour problem and resulting tail latency. Rockport’s NC1225 network card has 300Gbit/sec bandwidth and enables each endpoint or node to be connected to 12 other nodes. 

This switchless network is a fully distributed direct interconnect system based on supercomputer topologies. It supports small and large scale clusters with greater bandwidth efficiency and lower power requirements than a comparable traditional switching environment. 

Network technology

We had a look at Rockport’s 6D torus-based networking technology back in 2015 and reported: “Its idea is that storage and server nodes in a network should be directly interconnected using a Direct Interconnect (torus mesh) of optical cables. A torus mesh is a geometric shape analogy of their interconnection scheme, being a ring with a circular cross-section and a hole inside the middle of the ring — think doughnut.”

This design favours east-west network traffic in a datacentre as much as north-south traffic, unlike a traditional switch-based network. Rockport says a hierarchical switch-based scheme rapidly becomes excessively complex as it scales.

Rockport switch network diagram.

It contrasts this with its relatively simple Network Card and SHFL scheme;

Network admins  use the Rockport Autonomous Network Manager (ANM) user interface (UI) to configure, manage, and troubleshoot the Rockport Switchless Network. The ANM continuously captures all aspects of the network (health status, performance metrics, error counters, events, alarms, and the UI status) in a temporal or a time-series database. It stores 30 days of historical data, with the last seven days of history stored at high fidelity within the temporal database, available from the timeline. This lets network administrators scroll back in time to insert themselves into that context for troubleshooting. 

Nodes autonomously detect failed links and automatically avoid the impacted paths in real time at hardware speed. A downloadable Rockport technology primer document explains more.

Benefits

Rockport says that its users are seeing an average 28 per cent reduction in workload completion time because of its faster networking with lower and more consistent latency. It claims a one per cent improvement in workload completion times translates to approximately one per cent improvement in overall AI/ML/HPC cluster utilisation.

It also argues that eliminating centralised switches and using its scheme reduces power and resulting cooling requirements of network equipment by over 60 per cent, with overall datacentre energy savings of 12 per cent or better. Cabling is reduced by up to 75 per cent. The larger the infrastructure, the greater the savings in both material and personpower to install and manage networks. 

This means Rockport’s gear can provide, it says, benefits for customers with one to many switches.

Users

There are 396 Rockport Network Card nodes in production at TACC, the Texas Advanced Computing Centre, is both a customer and a Rockport Centre of Excellence. TACC houses “Frontera” — the tenth most powerful supercomputer in the world — and Rockport’s switchless network is part of the system.

Dr Dan Stanzione, director of TACC and associate VP for research at UT-Austin, said: “We’re excited to work with innovative new technology like Rockport’s switchless network design. Our team is seeing promising initial results in terms of congestion and latency control. We’ve been impressed by the simplicity of installation and management. We look forward to continuing to test on new and larger workloads and expanding the Rockport Switchless Network further into our data centre.”

You can watch a video session with Stanzione here (but you might need to get a Vimeo login to do so).

Alastair Basden of DiRAC/Durham University and technical manager of the COSMA HPC Cluster, said: “Based on a 6D torus, we found the Rockport Switchless Network to be remarkably easy to setup and install. We looked at codes that rely on point-to-point communications between all nodes with varying packet sizes where — typically — congestion can reduce performance on traditional networks. We were able to achieve consistent low latency under load and look forward to seeing the impact this will have on even larger-scale cosmology simulations.”

Rockport background

Rockport was founded in 2012 by CEO Doug Carwardine, initial CTO and now Chief Scientist Dan Oprea and COO Michael McLay. It has taken in just $18.8 million of VC, grant and debt financing according to Cunchbase, with about $15 million of that raised in 2019.

The company has 165 employees, up from around 80 a year ago, with 100 of them being engineers, and it has filed around 80 patents. It is talking to potential OEMs and tier-1 channel partners and we might expect announcements about partnerships in the future.

2021 is a banner year for SK hynix

Korean DRAM and NAND supplier SK hynix reported record quarterly revenues on the back of strong semiconductor memory demand, and its NAND business made its first surplus.

Revenues in the quarter ended September 30 were 11.805 trillion won ($10.13 billion), up 45.2 per cent year-on-year, with a profit of 3.3156 trillion won ($2.84 billion) – 206 per cent higher than a year ago. 

Kevin (Jongwon) Noh, EVP and Head of Corporate Center (CFO) at SK hynix, issued a results statement, saying: “Despite the concerns over recent global supply chain disruptions, the remarkable quarterly performance signals that the semiconductor memory industry is continuing its growth momentum.”

Revenues have just grown past the Q3 2018 peak, possibly signalling the coming end of the DRAM/NAND growth cycle. Profits are still lower though.

The company increased its production of leading-edge 1nm DRAM, which saw yield improvements, and 128-layer 4D NAND flash, lowering its product costs relative to DRAM and NAND density. It said 128L accounted for 75 per cent of its quarterly bit output. This, and increased DRAM and NAND demand in the server and smartphone markets, enabled its significant quarterly revenue and profits rise.

Notice the Q3 2021 growth acceleration.

It expects demand to stay high, and the NAND business should get even better because of the coming completion of the Intel NAND and SSD business acquisition.

Noh said: “After the completion of the acquisition, SK hynix will establish [a] complementary product portfolio which maximises both companies’ strengths, and seek to achieve economies of scale. For the future SK hynix will expand its R&D foothold and evolve into a global semiconductor memory leader.”

There won’t be a single, combined product portfolio. Instead the individual SK hynix and Intel NAND technologies will stay separate.

Wells Fargo analyst Aaron Rakers added some colour to reported results, telling subscribers: “Hynix reported [a] DRAM bit shipment decline in the low-single-digit q/q range in 3Q21, falling below the company’s expectation of low-single-digit q/q bit growth (driven by PC slowdown).”

NAND bit shipments did the opposite, and Rakers enthused: “Bit ship grew in the low-20 per cent q/q range, which was ahead of the company’s prior expectations of high-teens per cent growth q/q in 3Q21. … Hynix expects its 4Q21 DRAM bit shipments to grow in the mid-/high-single-digit q/q range and NAND bit shipments to grow in the double-digit q/q range.”

More generally, he observed: “Hynix noted that it expects to see improved Chromebook PC demand trends into late-2021/2022, while also pointing to an expectation of improved smartphone demand in 4Q21 and continued strength in server DRAM demand.” The company was: “particularly positive on the demand trends it is seeing in enterprise SSDs (eSSDs), which continues to be somewhat constrained by component supply.”

It looks like the boom times for SK hynix are set to continue; its product and company strategies are winning in won terms.

Storage news ticker – October 26

Robin.io and AirHop Communications today announced a strategic partnership to provide solutions that deliver automated improvements in the performance of Open Radio Access Networks and end-user quality of experience while reducing the costs, complexities and deployment times for Mobile Network Operators and Cloud Service Providers rolling out 5G and other next-gen networks and services.

Kingston Digital announced its KC3000 PCIe 4.0 NVMe M.2 2280 SSD for desktop and laptop PCs. It uses 3D TLC NAND and PCIe Gen-4×4 NVME access to deliver speeds up to 7,000MB/sec read and write from its 512GB, 1TB, 2TB and  4TB capacity points.  Random read or write IOPS range up to 1 million. The endurance in total bytes written terms for each capacity is: 400TBW (512GB), 800TBW (1TB), 1.6PBW (2TB) and 3.2PBW (4TB). It’s intended for personal compute device use and not for servers.

Kingston KC3000 PCIe Gen-4 SSD.

Edge HCI supplier Scale Computing and data protector/securor Acronis jointly announced the availability of Business Resilience System for MSPs, Purpose Built for Acronis Cyber Protect Cloud — a scalable appliance that allows for instant recovery of all workloads in an environment, regardless of where those workloads are running. This hybrid cloud platform delivers MSPs everything they need to ensure clients avoid downtime, lost revenue, and fallout from cyber attacks.

ionir, which supplies Kubernetes Data Services, announced updates to its cloud-native data services platform with new capabilities focused on CI/CD acceleration and performance. Kubernetes, now considered the operating system of the cloud, reinvents the concept of infrastructure — everything is a resource, with no notion of fixed assets. Resources are fluid, transportable, orchestrated, and automated. There are:

  • Updates to mobility/cloning — Users can instantly clone volumes after they have been teleported to another location, and view the contents on a remote system, reducing cost and facilitating data movement oacross clouds.  
  • Increased performance — A new class of volumes for high performance applications with low latency requirements. A heterogeneous install balances the ionir system on the hosting cluster in a more detailed way, good for efficiency and rightsizing infrastructure for each application. 
  • New solution guides — A program of integrated technical guides provides validated solution architectures for common Kubernetes applications and tools such as Jenkins, MongoDB, JFrog, CircleCI, and others.

Iguazio, a machine learning operations (MLOps) company, announced that it has become a Pure Storage technology partner. With the new partnership, Iguazio offers an MLOps platform that automates and accelerates machine learning and deep learning pipelines. Its serverless and managed services approach shifts the focus from managing a sprawl of Kubernetes resources to delivering a set of higher-level MLOps services, over elastic resources, thereby reducing significant capital and operational costs, and accelerating time to market. Pure provides enterprise-grade data management capabilities and products that are optimal for the challenges of data science, built for speed and simplicity at scale.

Totenpass stores data on a gold-plated nickel slab

Totenpass, a New York state company run by a contrarian investor, gold trader, jeweller, and bitcoin miner, is beta-testing a system for storing data for 100 years on credit card-sized slabs of gold-plated nickel.

The founder and CEO is Roy Sebag, and he explains: ”Our revolutionary storage drive helps you to retain control of your data while reducing your dependence on third parties and the internet, where these files can be easily erased or stolen. We invite you to make your personal history into a tangible and everlasting treasure.”

Totenpass is a German word meaning “passport of the dead” and Sebag says:  “I believe that Totenpass technology will be a giant leap forward for the permanent archival of mankind.“ Others might disagree.

Totenpass examples.

Product

There are two Totenpass sizes. The small one is credit card size, 54mm x 85mm, contains 0.2 grams of gold and has 1TB of digital capacity or 2,196 pages of analog visual or textual data. The larger card is smart phone size, 71mm x 143mm, with 0.4 grams of gold and 4TB or 3,894 pages of capacity. 

Customers get a card created by uploading data through a website, paying $50 for the small card or $75 for the larger one (with credit card or cryptocurrency — of course), and having the finished product posted to them. Digital information on the card is neither encrypted nor passphrase protected, unless requested. Digital data can be read by a smartphone app. Beta users of Totenpass will not yet have access to the reader app, which is expected to launch in 2022.

Analog data is seen by eye, possibly with the aid of a magnifying glass.

Totenpass analog Venus.

The card requires no electrical power for it to function and so is claimed to support environmental concerns; gold is green.

Technology

The technology relies on digital data being written by collimated laser light and a proprietary light diffraction imprinting technology producing “patterns written into multiple dimensions of the gold surface.” The laser has ”a highly focused 405nm wavelength creating feature sizes of 125nm or 1/1000th of a human hair”

For analog data storage, “the diffractive patterns are arranged with billions of diffractive bits to form pixels that comprise the formats familiar to our human experience such as visual documents and photos.” We’re told “textual files including spreadsheets and Powerpoints, photographs, images, documents and even audio files may be etched directly onto the surface of the Totenpass.”

And: “Totenpass technology prints over 10 billion pixels at a resolution of 25,000 dots per inch at a speed of 1,000 pages per minute making it the fastest analog nano printing technology in the world.”

Technology gibberish

Sebag explains his thinking in a lengthy, 22-page white paper full of partial knowledge, questionable comparisons, portentous language and mistaken assertions.

For example, he says that floppy disks can store data for 20 years while SSDs and magnetic tape only last for ten years. We find this floppy disk lifespan claim bizarre.

Sebag writes: “We believe authors, artists, and thinkers will embrace the Totenpass as the primary method for protecting their intellectual property.” Right.

He boasts the “Totenpass is built from solid gold and nickel, which inherently resist destruction, never tarnish and will therefore outlive all other ephemeral forms of data storage, from hard drives to Cloud servers.” The Totenpass cards are claimed to be disaster-proof.

This idea of the Totenpass’s ability to “inherently resist destruction” is meaningless. It doesn’t consider attack by file, drill or grinding wheel, to state the obvious. It also neglects to consider aqua regia — a mixture of nitric acid and hydrochloric acid that can dissolve gold. 

Sebag mentions DNA storage but doesn’t consider the alternative of Piql film reels in a Svalbard coalmine for 1,000 year storage.

There’s also this: ”The Totenpass user experience interface is beautiful, simple, and login free thus being privacy oriented.” Er … no — not unless the digital data is encrypted and passphrase-protected. Analog data is immediately visible; no privacy there at all.

We think the Totenpass’s natural marketing home is in the advert pages of in-flight magazines along with earwax removal devices, peletons, and outdoor bluetooth speakers. It will probably do very well there indeed.

Storage news ticker – October 25

Reuters has reported that $20 billion merger talks between Kioxia and its joint-venture partner Western Digital have come to a roadblock. Sticking points are differences in how to value Kioxia, securing Japanese government blessing and an uncertain outcome from an internal strategy review at Toshiba Corporation, which owns 40 per cent of Kioxia.

André Rievers.

NAS supplier OpenDrives has appointed André Rievers as its VP of Operations. It raised $20 million in funding in January and has been recruiting since then. Rievers was most recently Head of Post-Production Department for Activision’s Esports division, supporting and overseeing all digital and broadcast content for “Overwatch League,” “Call of Duty League,” “Hearthstone,” “World of Warcraft,” and “Starcraft.” OpenDrives wants to expand into supplying video management software and Rievers looks a good fit to that aim.

Yilan county in NE Taiwan has been struck by a magnitude 6.5 earthquake at about 1.11pm on Sunday October 24. This could interrupt operations at Micron’s two DRAM fabs in the country. Such fabs typically automatically shutdown if an earthquake takes place. Buildings shook in the quake but only minor damage was reported. TSMC evacuated some of its plants.

Bloomberg later reported that Micron experienced some DRAM production impact in Fab 11 (Tao Yuan) from the earthquake. Micron is evaluating the impact and working out how to return to full production.

Research house TrendForce has ranked the top ten SSD module manufacturers, giving Kingston the top position with a 27 per cent market share. ADATA, and Kimtigo occupy the next two spots. TrendForce says that, looking at the channel market for SSDs as a whole, NAND Flash suppliers accounted for around 35 per cent of the total shipments in 2020, while SSD module makers accounted for the other 65 per cent. The top ten module makers accounted for 71 per cent of channel-market SSD shipments from all SSD module makers. 

TrendForce’s ranking of SSD module makers for 2020 only takes account of products bound for the channel market and under brands owned by the module makers themselves; NAND Flash suppliers were excluded from the top ten ranking.

Multi-cloud service provider Scaleway has made its Kosmos managed Kubernetes engine generally available. Developers can attach nodes on any cloud provider or local resource, whether or not other providers offer a managed Kubernetes solution — with a fixed price per cluster of €99 per month, regardless of the number of nodes. Scaleway has already released its Kapsule managed Kubernetes engine and Kosmos enables developers to extend Kapsule beyond the compute infrastructure to any machine — instances from any provider, bare metal products, colocated machines, or their personal infrastructure.

US SSD and memory module supplier Netlist announced the Federal District Court for the Central District of California granted summary judgements in favour of Netlist and against Samsung for material breach of various obligations under a Joint Development and License Agreement (JDLA), which the two executed in November 2015. A summary judgment is a final determination rendered by the judge and has the same force and effect as a final ruling after a jury trial in litigation. The Court held that Netlist properly terminated the JDLA and confirmed that Samsung no longer has a valid license to Netlist’s patent portfolio. The jury is left to decide on direct damages related to the breach of contract during the trial set to commence on November 30, 2021 in the Central District of California.

Mass-capacity drive demand sends Seagate revenues soaring to the clouds

Seagate has benefitted from an extraordinary secular change in the disk drive market as cloud-scale datacentre demand sends its quarterly revenues up 35 per cent year-on-year — its fourth consecutive growth quarter.

Revenues in the quarter ended October 1 were $3.12 billion, compared to $2.3 billion a year ago, with profits of $526 million — a 135.9 per cent increase year-on-year. Seagate is a cash-generating machine.

CEO Dave Mosley was enthused in the earnings call: “Seagate had an outstanding start to fiscal 2022, underscored by our September quarter results. … Overall, our results reflect record demand for our industry-leading mass capacity products and solid execution on cost reduction plans and our ongoing focus of balancing supply with demand. We are confident in our ability to deliver excellent results for the fiscal year.”

Demand growth is set to continue for the rest of the year and its raised its outlook to greater than ten per cent full year growth from around eight per cent. The company also raised its dividend, confirming its confidence in the future.

Mosley said: “Without question, the HDD industry is being driven by long-term secular demand for mass capacity storage, a market that we expect to more than double by calendar 2026 to $26 billion, and Seagate is well equipped to answer the call. “

Total capacity shipped was 159.1EB — a 39 per cent rise on the year.

The year-on-year revenue growth was spectacular, as two charts show: 

Note the four recent quarters of consecutive growth and the first $3 billion-plus quarter in five years.
See the sharp revenue uplift in Q1 FY2021 (blue) line.

Seagate revenues have been in a trough since the glory years of FY2012 to FY2015: 

If its growth expectations for the 2022 financial year come true then we could see $12 billion full year revenues, ending a six-year slump since the glory days. If Seagate maintains its share (~43 per cent) of the HDD market, and it does double to $26 billion in 2026, then annual revenues should grow larger still.

Business segments

Seagate has two main business segments: its HDD business that covers mass-capacity disk drives, legacy drives, and the rest; and non-HDD business referred to as Enterprise Data Solutions, SSD & Other. There are three mass capacity HDD product groups — Nearline, VIA (Video and Image Apps) and NAS — and the three account for 71 per cent of its revenues with 132.3EB of shipped capacity ( up 53 per cent year-on-year).

Mosley said: “Revenue from the mass capacity markets exceeded $2 billion for the first time, reflecting broad-based growth across each of the end markets.” These are cloud, enterprise and OEM customers.

Mosley again: “The cloud is the strongest contributor to the mass capacity markets and Seagate’s revenue growth. Ongoing investments to build and equip new datacentres have translated into stable healthy demand for multiple quarters now and we expect this trend to continue. Over the past five years, the number of hyperscale datacentres has more than doubled to nearly 600 worldwide, with approximately 200 more on the way.”

The enterprise and OEM business grew for the fourth time in a row, though not as much as cloud. VIA demand grew significantly driven by higher-definition cameras and more use of analytics. That’s expected to continue this quarter. 

The Lyve Mobile and Lyve Cloud business numbers were not called out, although Mosley said there was progress in the customer qualifications and a video comms provider partnership. He said: “We will continue to be deliberate in scaling infrastructure and developing an ecosystem.” No drama here.

Seagate shipped 106EB of nearline capacity, up 65 per cent annually.

Mosley said dual actuator (MACH.2) drives were shipping at large scale and that technology could become mainstream once capacities reach 30TB and beyond. He said HAMR drives were shipping with good recent test results, confirmed it was the gateway to 30TB and beyond, but didn’t say there would be a fixed cross-over date from current perpendicular recording (PMER) to HAMR drives. 

We see a gradual increase in the proportion of HAMR drives shipping vs PMR drives, as capacities climb the staircase from 20TB to 24TB, 26TB, and on to 30TB. Mosley said he was very confident about the HAMR pathway to 30TB and beyond.

The legacy business — 29 per cent of Seagate’s revenue at $831 million and up five per cent year-on-year — is divided between consumer drives (mostly PC and branded) and mission-critical (2.5-inch) drives. Mission-critical drive sales rose, offsetting a consumer drive fall. Wells Fargo analyst Aaron Rakers told subscribers legacy HDDs dipped below 30 per cent of Seagate’s revenues for the first time. 

The non-HDD business grew from $177 million a year ago to $251 million and Mosley commented: ”The non-HDD business has been a little choppy. I mean, there are certain places where we can use our brand for continued strength. But there are opportunities and we take advantage of them and sometimes those opportunities wax and wane a little bit.” It’s small potatoes compared to the HDD business.

Seagate generated $496 million in cash flow from operations and $379 million in free cash flow in the quarter. It estimates it will make $3.1 billion revenues in the second FY2022 quarter, plus/minus $150 million, which would be 18.2 per cent annual growth at the mid-point.

Dremio software Darts make data lakes more like data warehouses

Dremio’s latest data lake analytics release provides more than five times faster SQL expression processing than its June version.

Dremio supplies in-memory software, powered by Apache Arrow, to analyse data stores using SQL. Its Dart initiative aims to make it possible to run SQL workloads directly on source data lakes instead of on data warehouses fed by extract, transform and load procedures. It has been pumping out software updates to help achieve this.

Tomer Shiran, founder and Chief Product Officer at Dremio, said: “We want to push the boundaries of what’s possible in the data lakehouse and deliver the best [business intelligence] experience for our customers. To that end, the Dart Initiative has been chipping away at the Zone of Confusion between data lakes and warehouses in critical areas such as query performance and acceleration, SQL coverage, and transactionality.”

The “Zone of Confusion” is a Gartner expression referring to the overlap in analytics processing of data lakes and warehouses. We might say Shiran’s data lakehouse idea adds verbal confusion. The point of running ETL procedures is get more structured data into a data warehouse so that analytic routines can deliver results faster.

This latest software release provides near real-time metadata refresh for datasets by refactoring metadata processing to become a parallel, executor-based process, with metadata stored and managed in Apache Iceberg tables. This delivers metadata refresh times up to 20x faster than previous versions and performance improves as the dataset size increases. This graph illustrates the point:

The company says Arrow component Gandiva is an LLVM-based toolkit that enables vectorised execution directly on in-memory Arrow buffers, by generating code to evaluate SQL expressions that fully leverage the pipelining and SIMD capabilities of modern CPUs. This latest Dart Initiative release enables Dremio to accelerate expression processing rates by over 5x, and in some cases by 30x, as another chart indicates:

The latest release adds Pivot/Unpivot and filtered aggregates functions. Dremio says risk analysis in insurance, maximising revenue in travel and transportation, improving clinical trials in pharma, and enabling credit risk assessment in banking are among the use cases that can benefit from this.

Let the blind see: Datadobi opens up vast file system window to customers

Datadobi has opened up its petabyte-scale file mapping technology to customers with a Dobi Query Language.

Datadobi is a data migration technology supplier, with DobiMigrate moving NAS files and objects, and DobiProtect protecting and recovering them. Both products rely on DataDobi’s so-called Data Mobility Engine and are designed to scan large file systems (data lakes) containing billions of files. This produces a catalogue containing huge lists of file paths and their metadata in proprietary format.

Carl D'Halluin
Carl D’Halluin.

Historically, these scan files were only used for doing data migration or protection by Datadobi, but now customers can query them directly using Dobi Query Language (DQL) as part of a file assessment service.

Carl D’Halluin, Datadobi’s CTO, writes in a blog: “The volume of data is only expected to grow over the next few years. IT administrators need a data management solution that can transform data into digestible material in order to allow curated decisions on storage options for migration and protection to be made.”

Step forward DQL. It is a query framework that can look for inside data lakes to: 

  • Identify cold data sets — data that is infrequently accessed;
  • Identify old data sets — data that was created or modified some time ago;
  • Identify data sets owned by a specific user or group, e.g. by users who no longer work at the company;
  • Identify shares, exports or directory trees that are homogeneous (cold, old, owner, file types) and can be handled as one data set, e.g. upon which to take specific lifecycle actions.

Datadobi created the file system assessment offering last year as a customer service to help plan a data migration or reorganisation. DQL allows these assessments to be customisable.

DQL code sample.

DQL is a form of file system analytics. It opens up a door to future possibilities such as  an alerting software layer that can run DQL to find out file system and subset status, and suggest actions — such as a file/object subset move to faster or slower/cheaper storage based on its activity level. A step beyond that would be to set up policies and then have an additional layer of software take automatic actions if threshold values were reached.

For example, move all files or objects of a particular type to archival or cold storage if certain access rate requirements were met (say, if the data were not accessed within 60 days). Datadobi has interesting possibilities here.

Tick the NVMe-oF box – StorONE adopts NVMe-over-Fabrics tech as no-fuss upgrade

StorONE has added NVMe-over-Fabrics connectivity to its S1 Enterprise array software, speeding data access to sub-millisecond levels.

The S1 file, block and object storage software has a cacheless architecture which delivers high performance through a ground-up redesign of the storage stack using commodity hardware. NVMe-over-Fabrics (NVMe-oF) provides local PCIe-attached drive performance across a network link between a host and an external storage system. The NVMe-oF specification extends NVMe onto suitable storage fabrics such as Ethernet, Fibre Channel and InfiniBand.

Gal Naor.

Gal Naor, CEO and co-founder of StorONE, said in a statement: “By integrating the NVMe-oF protocol into our existing platform rather than creating a new product for organisations to purchase, we are better able to fulfill our mission of turning storage from an IT cost centre to a resource that provides critical competitive advantages.”

StorONE says its NVMe-oF implementation supports NVMe-over-TCP and Ethernet RDMA (RoCE) to reduce IO latency while supporting all block IO features, including high availability and asymmetric active-active failover. Existing customers, after upgrading, can take advantage of NVMe-oF using existing hardware, such as iSCSI hardware.

As far as StorONE is concerned NVMe-oF is a feature, not a product. From its viewpoint any block-access all-flash storage array not offering NVMe-oF access is behind the times and at an increasing disadvantage. It has adopted datacentre-class Ethernet lossless (RoCE) and ordinary Ethernet (NVMe/TCP) methods but not NVMe-over-Fibre Channel. 

There has been a flurry of recent NVMe/TCP support announcements, including Lightbits, Fungible, Dell EMC, VMware and NetApp. It’s becoming table stakes in the enterprise storage array game.

Storage news ticker – October 22

HPE plans to expand its Hewlett Packard Pathfinder venture capital program to help accelerate HPE’s edge-to-cloud strategy. It expects to ramp its rate of investment significantly over the next year starting with three new investments that solve key customer challenges across edge, cloud, and data. Three recent investments support HPE’s vision: 

  • Cellwize is a Radio Access Network (RAN) automation platform that enables telecommunications companies to accelerate 5G network deployment. 
  • vFunction helps companies modernize their legacy applications to run in containerized environments using a lightweight microservices architecture. 
  • SingleStore is a one-stop, single database that unifies data models and data types and offers low-latency, consistent queries and processing for data-intensive workloads across hybrid, on-premises, and cloud environments.  

HPE will evaluate introducing SingleStore to the HPE Ezmeral Marketplace to provide a database with tiered storage spanning RAM, NVMe, SSDs, and cloud object storage.

GIGABYTE Technology announced its new R282-Z9G server targets high performance NVMe (Gen-4) SSDs for RAID and incorporates the GRAID SupremeRAID product. The SupremeRAID card works by installing a virtual NVMe controller on the OS while integrating a PCIe device for high performance. Over 100GB/sec of throughput is possible for workloads in HPC, 4K/8K video editing, high-frequency trading, online transaction processing, or database processing.

Couchbase announced the winners of its Community Customer and Partner Awards at the CONNECT 2021 event. A few interesting ones are: Emirates, winning the EMEA Innovating at the Edge category and Project of the Year, for using Couchbase to optimise aircraft maintenance; Citigroup, scooping up the Advanced NoSQL Architecture EMEA prize for designing a system that could store billions of data points within Couchbase; and Amadeus, the third EMEA winner for Cloud Computing, for using Couchbase to develop a cloud-native platform that sped up the time to market of new applications and offerings, as well as data exchange.

VAST Data president Michael Wing made a LinkedIn post showing a complex Dell EMC Isilon/PowerScale data reduction decision tree, posted by an engineer on Slack, contrasting it with VAST’s equivalent tree:

It’s not a high-res image but you can nonetheless see the point he’s making. Wing said the Isilon architecture was “built 20 years ago for hard drives, streaming media workflows and no data reduction. Retrofitting all of this functionality is hard and comes with massive trade offs. Hence the decision tree!”

Seagate has announced the launch of its Lyve Mobile Edge Storage and Data Transfer Services in the UK. The subscription-based service offers businesses a means to transfer massive amounts of data via mobile shuttles and arrays. Customers subscribe through the Lyve Management Portal and a web-based purchasing model gives end-customers or resellers flexible subscription plans ranging from month-to-month plans to longer-term annual plans. Once registered and an account is established, customers can order a project in less than three minutes. Seagate then provisions the devices and ships them to the desired location. Seagate is already providing support to the development of autonomous vehicles at FMCI, Europe’s largest autonomous vehicle testbed, with its Lyve Mobile service.

File lifecycle manager Komprise says St Luke’s Health System — a non-profit in Boise, Idaho, with petabytes of NAS storage — is using its software in a tiered storage architecture. Brett Sayles, storage engineer with St Luke’s, said: “We have 20 years of files from various medical systems and we have been treating all data the same. We are nearing capacity on our NetApp and Pure storage, and we know that a lot of these files can go to cheaper storage.” St Luke’s will use Komprise to archive data from Pure Storage Flash Array and NetApp to secondary storage on Qumulo spinning disk. It expects to save as much as 80 per cent on storage with this initiative.

Database, data integration and analytics supplier Actian announced GA of its DataConnect 12 integration platform, with new functionality for data quality, enhanced scheduling and automation, and the availability of additional software development toolkits (SDKs) to improve development integrations.

Object (and file) storage supplier Scality tells the world it “delivers the most comprehensive and rock-solid portfolio of Disaster Recovery-enabled SmartStore ready deployment options as validated by Splunk.” Scality offers three SmartStore-ready architectures: single site, two-site asynchronous replicated, and three-site synchronous writes, capable of ensuring zero recovery point objectives (RPO) and recovery time objectives. There’s a blog if you want to know more.

Storage news ticker – October 21

.…

Weka IO has hired Frederik Schroeder as VP of Strategic Partners and Sean Hiss as the new VP of GTM Operations. Schroeder joins from Hitachi Vantara, where he was VP and GM of Global Inside Sales. Hiss also comes from Hitachi Vantara where he was VP of Marketing Strategy and Planning as well as Global Head of Revenue Marketing. 

Iguazio, the MLOps (machine learning operations) company, announced support for FSx for ONTAP. This provides fully managed shared file and block storage on AWS Cloud with the data access and management capabilities of ONTAP. 

Yvonne Wassenaar.

Rubrik has appointed of Yvonne Wassenaar, the CEO of Puppet, to its Board of Directors. She serves on the Board of Directors for Forrester, Anaplan, and Harvey Mudd College and, Rubrik says, brings deep experience in SaaS and enterprise go-to-market, having previously held key leadership positions at VMware, Accenture, and New Relic. 

Data protector and manager Quest Software announced the availability of Quest QoreStor v7.1, providing customers with enhanced ransomware prevention capabilities and cloud support. It provides Cloud Reader Mode — an alternative QoreStor instance that can be configured to read a QoreStor Cloud tier that is active, enabling users to test their cloud data. There is integration with Quest’s QorePortal for SaaS-based management of QoreStor instances from anywhere on anything. QorePortal comes with built in organisational structure and role-based access.

File Transfer Protocol (FTP) support has been deleted from the codebase in the latest stable build of the Chrome browser, version 95. See here for more details.

Kubernetes-as-a-Service supplier Platform9 announced a Profile Engine for template-based cross-cluster governance, IDE enhancements for fast access to running workloads, user-specified upgrade strategies (sequential, parallel node percentage or batch , KubeVirt enhancements, and CIS benchmarking compliance. The Profile Engine enables GitOps style automation in multi-cloud environments by allowing administrators to define cluster configurations as desired state templates, and then deploy those to any target cluster.

Amazon Web Services announced the introduction of AWS Data Exchange for Amazon Redshift, making it even easier to find, subscribe to, and use thousands of third-party datasets from within Amazon Redshift with just a few clicks and no need to extract, transform, and load it. Customers can directly query third-party data from AWS Data Exchange and combine it with their own first-party data in Amazon Redshift to generate more relevant, comprehensive insights for their business.

NetApp’s new Spot Security offering provides automates security monitoring in the cloud by scanning cloud resource usage and the relationship between dynamically changing cloud resources to understand the scope and breadth that a compromised resource may have on the overall account. It ingests data from cloud APIs, network traffic, and user activity to monitor cloud infrastructure events like access, utilisation, and configuration changes. Spot Security will be available for private preview soon, to be followed by general availability on AWS and Azure.  Spot Security supports EC2, S3, IAM, RDS,  VPCs and other commonly used services, with more to come. Find out more here.

NetApp Spot Security graphic.

IBM has flat revenue third quarter with a bright spot in – gasp – storage

Revenues in IBM’s third fiscal 2021 quarter, ended September 30, were $17.6 billion — up 0.3 per cent year-on-year. There was a profit of $1.1 billion — 33.5 per cent lower than a year ago — but storage revenues rose 11 per cent annually.

The revenues were effectively flat and profit was down — not a lot to write home about. SVP and CFO James Kavanaugh’s results statement emphasised the positives and said: “We again had solid cash generation for the quarter and over the last year, while maintaining a strong balance sheet and the liquidity to support our hybrid cloud and AI strategy.”

IBM’s Chairman and CEO Arvind Krishna’s results statement ignored the headline numbers and looked ahead: “With the separation of Kyndryl early next month, IBM takes the next step in our evolution as a platform-centric hybrid cloud and AI company. We continue to make progress in our software and consulting businesses, which represent our higher growth opportunities. With our increased focus and agility to better serve clients, we are confident in achieving our medium-term objectives of mid-single digit revenue growth and strong free cash flow generation.”

Kyndryl is the name of the to-be-separated-out IT Infrastructure Services unit of IBM’s Global Technology Services business unit, and will become an independent company next month. The corporation reckons it will exit this fiscal year able to deliver mid-single-digit revenue growth and $35 billion free cash flow in 2022–2024. 

IBM’s business segment results:

  • Cloud & Cognitive software — $5.7 billion, up 2.5 per cent;
  • Global Business Services — $4.4 billion, up 11.6 per cent;
  • Global Technology Services — $6.2 billion, down 4.8 per cent;
  • Systems — $1.107 billion, down 11.9 per cent;
  • Global Financing — $220 million, down 19.2 per cent;

The Systems segment revenue decline was driven by declines in Z mainframes (down 33 per cent) and Power systems (down 24 per cent). Storage Systems grew 11 per cent. Cloud revenue was down 42 per cent. The growth in storage was driven by entry-level and tape systems. 

IBM launched an entry-level FlashSystem 5200 in February — faster and higher in capacity, yet 20 per cent lower in price than the prior 5100 — and it has obviously done well. The growth in archival data storage needs has probably accounted for the tape systems revenue increase.

FlashSystem 9200.

The mainframe and Power revenue declines were attributed to cycle dynamics, implying a revenue uplift should  hopefully be coming from new hardware and software.

Systems provided $0.8 billion in hardware revenues and $0.3 billion in operating systems software revenues. IBM has provided percentage change numbers for the storage part of its systems business, and our records anchor these to an actual storage revenue number provided many quarters ago. We have recalculated the quarterly storage revenue number using these percentage changes and expect we are progressively becoming less accurate as a result.

This proviso is spelled out before we include our chart of IBM’s quarterly storage revenues by fiscal year to show the historical trends:

One quarter’s growth after many declining quarters does not make a trend, but IBM can be hopeful that a corner may have been turned.

We must note that most of IBM’s storage software revenues are not included in the Systems storage segment number. IBM is odd in that it does not provide a central and all-inclusive storage revenue number — at least in its public numbers. We simply do not know how IBM is doing from a storage software revenue perspective.