Home Blog Page 285

Profile: Cloud-native storage speed and enterprise smarts key for StorageOS

StorageOS supplies software-defined, cloud-native storage — like many other suppliers. It says what sets it apart is its focus on speed, scale and enterprise features.

The company is based in Farringdon, London, UK, and was started up in 2015 by CEO Alex Chircop, Vice President of Engineering Simon Croome, and initial CEO Chris Brandon. The company began with $2 million in seed funding and had an $8 million A-round of VC funding in 2018 plus a $10 million B-round earlier this year. The investors are Bain Capital, Chestnut Street Ventures, Downing, MMC and Uncorrelated Ventures, and the total funding to date is $20 million.

Brandon, now based in New York, had a storage technology and data centre background at EMC then BMC. Later he was at Xsigo (which was acquired by Oracle), then moved to GreenBytes and then Oracle (which promptly acquired GreenBytes). He left StorageOS in late 2019 to become a CTO in residence at AWS.

Chircop and Croome both had experience at Nomura, with Chircop going on to Goldman Sachs (serving as its Global Head of Storage Platform Engineering) and Gazprom. Croome is a specialist in agile delivery of Unix infrastructure and DevOps concepts, and led global engineering teams at Fidelity and Nomura.

Alex Chircop (left) and Simon Croome (right).

StorageOS has about 30 staff, most of whom are engineers.

The container storage problem

Chircop, Croome and Brandon saw that big enterprise’s code development was moving to containers, and that stateful containers needed storage — preferably software-defined and cloud-native storage. That’s what StorageOS has developed: a software-defined, cloud-native storage platform for containers.

Inside a StorageOS container.

Chircop says shared external arrays simply aren’t working in the Kubernetes environment. People want an agnostic storage setup, not lock-in, so StorageOS is platform-agnostic and ships as a container. It’s also faster and more versatile than external arrays. It can run in the cloud, for example. He says “We have some unique IP and features,” and is happy to list them:

  • “StorageOS is optimised for performance and latency, being many multiples faster than the competition — 20x in one instance.”
  • StorageOS can set up a new Kubernetes cluster in 20–30 seconds — it uses in-memory caching.
  • StorageOS is “one of the only products that deploy storage with no hardware or kernel dependencies.”
  • StorageOS has a great “ability to scale because of its control plane”. This uses a disaggregated consensus algorithm, and every volume can make its own placement decisions, like a self-contained storage array.
  • StorageOS can support a high rate of change, creating and deleting tens of thousands of volumes a day.
  • “StorageOS helps reliability on Kubernetes clusters. Each volume can make its own recovery decisions. This is very very different — many systems have centralised control in case of transient errors.”
  • StorageOS supports AWS, Azure, GCP, IBM, Red Hat, Digital Ocean and other public clouds as well as on-premises environments.
  • Most CSPs have have ephemeral NVMe SSD storage installed on a node, but “We can virtualise these ephemeral instances …  We make them enterprise-class.”
  • “CSPs can’t replicate across availability zones — we can. CSPs can’t apply high availability to ephemeral NVMe storage — we can. CSPs can’t encrypt without having the encryption keys — we can.”
  • “We use in-built functionality in the Linux kernel and have no dependencies on non-kernel features.”
  • “We’ve effectively been sold through the Red Hat marketplace and have integration with the OpenShift Catalogue and the Rancher Catalogue.”

StorageOS provides a storage container presenting a pool of virtual volumes to application containers. These volumes are mapped onto physical drives — NVMe SSDs for example. You can download a platform architecture document to look deeper into the product’s design.

StorageOS cluster diagram.

Customers

StorageOS wouldn’t disclose its customer count, but says that more than 5500 StorageOS clusters are installed globally. Example customers include Accenture, DHL, Lloyds Bank and T-Systems.

17 StorageOS customer logos.

One UK ISP customer is building an IaaS offering based on Kubernetes and Storage OS. Virtual machines are launched with Kubernetes, with Chircop pointing out that Kubernetes is agnostic about what it launches.

Roadmap

Chircop says there are two big trends:

  1. Organisations are deploying larger numbers of smaller Kubernetes clusters — “So we build the ability to consume and replicate storage across these clusters.”
  2. There is a bridge to traditional infrastructure. Organisations are going through transformation to Kubernetes. New programs have downstream dependencies to traditional software components. He said: “We need to expose traditional file (NFS or SMB) and block code to Kubernetes.”

For example, a traditional app could produce a bunch of files. A new Kubernetes app could provide analytics routines, and they need to process the data in these files. There needs to be some kind of bridge facilitating that access.

Chircop also revealed that “StorageOS is planning to launch commercial and technical products later this year”.

Competition

StorageOS is competing with a whole string of suppliers producing storage software for Kubernetes-orchestrated containers: Commvault’s Hedvig, Dell EMC’s Project Karavi, HPE Ezmeral, Ionir, MayaData’s OpenEBS Mayastor, NetApp Astra, Pure Storage’s Portworx, Robin.IO with its $86 million in funding, SUSE Rancher and others.

StorageOS is making progress with sales to enterprise customers. It has to hope that the enterprise IT infrastructure development coalface experience of its founders — which is guiding its development and business strategy — outweighs the attractions of competitors who lack it.

Profile: Flexify.IO puts multiple cloud storage providers into single bucket

Flexify.IO provides cloud storage virtualisation and migration software that offers customers cloud-agnostic object-based data storage. It enables and simplifies migration, and avoids dependency on a single cloud storage provider.  

The concept is straightforward. An application reads and writes data from and to a virtual S3 bucket. The actual data is stored in a physical bucket in one of more than 20 storage locations, including Alibaba Cloud, Amazon Web Services (AWS) S3, Microsoft Azure Blob, Backblaze B2, Dataline, Dell EMC ECS, DigitalOcean, Dropbox, Exoscale, Google Cloud, Linode, Mail.ru Cloud, MinIO, OVH, Wasabi, and Yandex Cloud.

Flexify virtual bucket scheme.

Flexify’s software tracks which physical storage location is used, and carries out API-based storage IO using connector modules to satisfy the read and write requests.

There is no support for file or block storage.

Migration matters

Data can be moved between these cloud service providers (CSPs) and on-premises storage locations, avoiding CSP lock-in. Specifically, Flexify transparently converts the S3 API to the Azure Blob API — so there is no need to convert S3-compatible applications to Azure Blob-compatible ones. It similarly translates between the S3-compatible API and Alibaba OSS APIs.

Data can be split across multiple locations and presented as being in a single (virtual) bucket. The data can also be migrated between these locations, with Flexify’s code taking care of errors and preventing downtime during a migration. Multiple Flexify instances (stateless engines) can be used in a migration process with a so-called slot-based distribution technique. This speeds the data transfer process.

No egress fees are payable when:

  • Migrating from Azure or Google Cloud Platform to Amazon S3 region on the same continent;
  • Migrating from Amazon Web Services or Google Cloud Platform to Azure region on the same continent;
  • Migrating from Azure, Google Cloud Platform or Amazon Web Services North American regions to Backblaze B2 US or from European regions to Backblaze B2 EU;
  • Migrating from Amazon Web Services S3 us- regions to Backblaze B2 US or from Amazon S3 eu- regions to Backblaze B2 EU;
  • Migrating from Microsoft Azure or Amazon Web Services to Google Cloud Platform region on the same continent.

We’re told that a large gaming company recently worked with Flexify.IO to move more than 5PB of data from Amazon S3 to Google. The largest migration to date involved 1.5PB of data spread across two billion objects.

Management console

Flexify provides a management console that includes a migration wizard. Data access speed is improved by an asynchronous transfer design, in which object access commences before a whole object has been fetched by Flexify. The company claims that: “With proper regions configuration (so that you don’t send data across the Atlantic Ocean and back) you can expect delay of no more than a few milliseconds.”

This is a multi-user management console, supporting customers who want to manage multiple departments and users on Flexify.IO through a central administrator.

Flexify.IO is available as a cloud service, managed by Flexify, or it can be installed privately in a customer’s own cloud or on-premises. The Flexify (cloud-native) app is available in the Azure and Digital Ocean marketplaces, and a Community Edition can be downloaded from the Docker Hub.

The Community Edition is limited to a single machine and is not scalable. The main edition is horizontally scalable, with centralised management of geographically distributed engines. 

Sergey Kandaurov.

Flexify founding and funding

Flexify.IO was started up in June 2015 by CTO Sergey Kandaurov and Corporate President & CEO Sergey Smirnov in Nizhniy Novgorod, Nizhegorod, Russian Federation. It raised an undisclosed amount of funding in May 2018, and has an office in St Petersburg, Florida. Its head of development, Alexander Bondin, is located in Novgorod, Russia, and Kandaurov is based in the Philippines.

Future development plans include adding support for smaller cloud providers, storing unstructured data in blockchain, and looking at near-real-time continuous synchronisation to keep two or more data stores in sync.

Check out a series of articles here to look further into Flexify. 

Pricing

Your occasional storage digest with IBM, Red Hat and Backblaze — and Kaseya’s $70M REvil demand

We start with exec changes at IBM, followed by a new version of OpenShift, and Backblaze going to Dell.

In the shorter news items we see both Panzura and Qumulo active in the healthcare sector and the ReEvil gang wanting an eye-watering $70 million to free up Kaseya ransomware attack victims’ files.

IBM exec changes

As part of a set of exec changes rolled out by IBM CEO Arvind Krishna, IBM’s storage unit has a new boss.

The main change was Red Hat’s ex-CEO Jim Whitehurst stepping down from being an IBM President and taking on a senior advisor’s role. So, in a cultural sense, Red Hat has become Big Bluer, but Big Blue has not become Red Hatter.

The change we’re most interested in concerns Tom Rosamilia, who was Senior VP in charge of IBM Systems, which covered mainframes, servers and storage and the global business partners’ organisation. He had been in place since January 2015, and now moves on to become SVP for Cloud and Cognitive SW.

Ric Lewis becomes the new SVP of Systems. He is being hired from a Senior VP/GM role at HPE, looking after the Software-Defined and Cloud Group (SDCG). His product set included Hyperconverged Systems, Synergy/Composable Systems, Helion, Cloudsystem, Oneview, SDS/SDN technologies, and HPE’s Hybrid IT software stack. He had previously run datacentre infrastructure and, before that,  enterprise servers at HPE, where he spent almost 18 years.

Lewis has no direct external storage product set experience. It’s debatable whether IBM Storage, which has suffered a multi-year revenue decline, will be turned around under his reign.

RedHat updates OpenShift

Red Hat announced version 4.8 of its open source OpenShift Kubernetes software platform. It includes:

  • IPv6/IPv4 dual stack and IPv6 single stack support;
  • OpenShift Pipelines for users to declaratively define, version and track changes to their application delivery pipelines alongside their application source code in Git repositories; 
  • OpenShift console, including the ability for Spring Boot developers to code and test locally before sharing the code, and scaling options for the developer console; 
  • OpenShift Serverless functions capability, available as a tech preview, enabling developers to create and run functions, on demand, on OpenShift; 
  • OpenShift sandboxed containers, a tech preview based on the Kata Containers open source project, providing a more secure container runtime using lightweight virtual machines and supporting extremely stringent application-level security.

Red Hat has expanded Red Hat OpenShift Certification to support a broader range of workloads. Partners can now enable and certify software solutions on OpenShift through either Operators or Helm charts.

The certified OpenShift ecosystem with Kubernetes Operators and Helm certification now includes over 150 partner products, with recently certified Operators including Intel OpenVINO Model Server and OpenNESS, Ionir container-native data platform for Kubernetes, MinIO hybrid cloud object storage, MongoDB Atlas cloud database service, and certified Helm charts for HashiCorp Vault.

Red Hat OpenShift 4.8 is expected to be generally available in July, including the ability to try it on the Developer Sandbox for Red Hat Openshift.

What the Dell? Backblaze Storage Pods

Up until now cloud backup and storage service provider Backblaze first built its own disk vaults, calling them Storage Pods, and then started using contract manufacturers, eventually moving to a hybrid contract manufacturer/self-build model. The most recent design is version 6.0 and has 25 different parts from 15 different manufacturers/vendors, plus the chassis and the hard drives. Each month it builds around 60 Pods, each weighing around 330kg.

Backblaze decided to try using Dell storage servers for version 7.0 Storage Pods in an Amsterdam data centre. It ordered two Backblaze Vaults worth of Dell servers for Amsterdam, and then another six. They are viewed as quasi-Storage Pods, and the idea has worked well enough.

There is more background about this in a Backblaze blog by principal storage cloud evangelist Andy Klein. He sums it up by saying: “Overall, it has been a positive experience for everyone involved — not perfect, but filled with learnings we can apply going forward.”

Shorts

Alluxio, a developer of open-source data orchestration software for large-scale workloads, announced the immediate availability of version 2.6 of its Data Orchestration Platform. It features an enhanced system architecture enabling AI/ML platform teams using GPUs to accelerate their data pipelines for business intelligence, applied machine learning and model training.

DDN is now selling Tintri IntelliFlash systems directly, so customers can buy them either from the Tintri organisation’s channel or from DDN itself. (Tintri is a division of DDN.)

The REvil gang is looking to get $70 million for a universal decryptor that could unlock all the business files locked during the Kaseya VSA ransomware attack. The Record has more details about what could be the highest ransom demand ever paid — if it gets paid. It could get all direct and indirect Kaseya customers affected by the ransomware off the hook. 

Kaseya ransomware demand.

Kingston Digital is shipping the DC1500M NVMe SSD in U,2 (2.5-inch) format with a PCIe Gen-3 four-lane interface. It has support for multiple namespaces and delivers consistent read/write latencies of <110 µs and <206 µs, respectively. The drives have speeds of up to 3GB/sec for reads, 700MB/sec for writes, up to 510K random read IOPS and 220,000 random write IOPS. Capacities are 960GB, 1.92TB, 3.84TB and 7.68TB.

MariaDB is collaborating with AWS to bring MariaDB SkySQL as a database-as-a-service (DBaaS) offering to AWS using Amazon Elastic Kubernetes Service (Amazon EKS), and AWS Fargate. SkySQL can deploy MariaDB on AWS as a distributed SQL database for scalable transaction processing, as a multi-master cluster for continuous availability, or as a multi-node columnar database for data warehousing and analytics requiring high availability and/or scalability.

In-memory computing supplier MemVerge and Hazelcast, which produces real-time streaming and memory-first database technologies, are partnering. Hazelcast technologies are being integrated with MemVerge memory virtualisation and in-memory snapshot technologies. The two will produce Big Memory reference architectures for Financial Services, starting with Basel III Compliance solutions that will be introduced in the second half of 2021. 

Exoscale, the European Cloud Computing platform, is now offering the Enterprise File Fabric from Storage Made Easy through its Marketplace, supporting cloud customers across Europe.

Panzura is partnering Datatility to offer its Global File System-as-a-Service (GFSaaS) to hospitals and other healthcare institutions. GFSaaS is optimised by Datatility as a fully managed and always-on service for healthcare institutions looking to access, protect and manage their unstructured and untapped clinical data.

SingleStore unified distributed SQL transaction and analytics database is available in the Microsoft Azure Marketplace.

The SNIA Swordfish storage management standard extends the DMTF Redfish Scalable Platforms Management API Specification to define a comprehensive, RESTful API for managing storage and related data services. Version 1.2.2a is now available for immediate download here. Updates include:

  • NVMe drive profiles and corresponding details added in the Swordfish NVMe Model Overview and Mapping Guide;
  • A Swordfish Property Guide to easily look up property definitions;
  • DMTF Redfish references.
Larry White.

Toshiba America has a new CEO: Larry White, who replaces the retiring Scott Maccabe.

Customers

Dutch service provider Bytesnet is using DDN storage — specifically EXAScaler parallel file systems. Bytesnet is using DDN DataFlow software to design, deploy and operate high-performance data workflows at scale from EXAScaler arrays.

Scale-out file system supplier Qumulo says it is popular in healthcare, with 150 per cent year-over-year growth in its customer roster. Customers include Biotechnology and genetic sequencing companies like Progenity, global healthcare research organisations like the Institute for Health Metrics and Evaluation (IHME), and healthcare providers such as Dayton Children’s Hospital.

Spectra Logic announced that post-production studio ColorTime deployed Spectra’s StorCycle Storage Lifecycle Management software to automate cleanup of its production storage area network (SAN) and free up space for incoming digital content. ColorTime also installed a BlackPearl Converged Storage System, a 500TB BlackPearl Object Storage Disk System for quick recall of active projects, and a Spectra Stack Tape Library to archive high-resolution digital media.

Virtana says its Migrate SaaS offering is gaining momentum. Partner Presidio delivered cloud migration efficiency and cost optimisation for a national insurance company, accelerating its move to Amazon Web Services (AWS) by 70 per cent.

Out with the old, in with the new — Pavilion Data hires a CEO

Pavilion Data has dropped its product guy, Gurpreet Singh, after four years as Chief Exec, and exchanged him for experienced business CEO Dario Zamarian.

This abrupt change happened on July 1, when Pavilion issued a statement: “Mr. Zamarian brings to Pavilion extensive experience with enterprise technology go-to-market motions from his roles at Cisco and Dell, in addition to being embedded in the investment world with a background in private equity advisor positions.”

Dario Zamarian posing for Pavilion PR shoot.

Mike Gustafson, Pavilion’s Chairman, said in a canned quote: “The Board sought out Dario Zamarian because his proven experience and leadership will serve as a catalyst to drive Pavilion forward as it enters its next phase of accelerated growth. Dario’s track record with building enterprise companies, coupled with his leadership experience, makes him the right person to be Pavilion’s next leader.”

Gurpreet Singh.

No hard feelings

Our private understanding is that this means transitional CEO Gurpreet Singh was not considered capable of getting Pavilion’s growth rate up to the level that Zamarian could. Singh is quoted as saying: “Pavilion is in a great position for its next phase of growth. I am excited to be handing over the CEO reins to Dario, who I’ve known for more than ten years. Dario is well equipped to meet the tremendous opportunities in front of Pavilion.”

Better than Singh, apparently, and it’s nice of Singh to say so. 

Pavilion, founded in 2014, produces a multi-controller, all-NVMe SSD array — the HyperParallel Flash Array — and positions it as the fastest array in the industry. It is a proprietary hardware design, competing with software-defined storage products as well as all-flash arrays from Dell, HPE, IBM, NetApp, Pure Storage, StorONE and VAST Data, and the memory-cached Infinidat arrays. That’s a lot of competition for a hot box with a parallel file system.

Pavilion’s HyperParallel all-NVMe SSD, NVe-oF array.

With Singh as its CEO, Pavilion matured from pre-product stage to significant early revenue growth from its Hyperparallel Data Platform product. Singh also led the company’s early marketing and sales wins, and took the company through multiple financing rounds.

About that financing …

Funding VCs came out with enthusiastic encomiums for Zamarian and Pavilion’s prospects.

Wen Hsieh, General Partner at Kleiner Perkins, gushed: “We believe Pavilion is poised to be one of Silicon Valley’s foundational companies because of its unmatched ability to accelerate compute, thereby enabling customers to get to answers faster.” 

Brian Wilcove, General Partner at Artiman Ventures, was similarly glowing: “With the key wins Pavilion has secured in Media & Entertainment, Financial Services, Healthcare, and High Performance Computing as a foundation for accelerating growth, now is the time to bring in Dario with his wealth of experience scaling companies. His capital market experience will also support Pavilion’s continued momentum.”

That point about capital market experience sent us looking into our funding history file for Pavilion. It last raised $25 million in a C-round in September 2019, and total funding is just $58 million. That seems light compared to multi-hundred-million-dollar funding totals for some other storage startups. For example, Panasas gathered $155 million, Pensando managed $278 million, and Pure Storage scraped together $470 million before its IPO.

We think Pavilion will need to raise more money. Zamarian has been brought in to get the company growing fast enough for the VCs to feel happy about its potential — and pump in more cash.

Kaseya VSA vulnerability opens a thousand-plus business doors to ransomware

Kaseya’s VSA remote monitoring and management tool was used as an attack vector to inject ransomware into the systems of fewer than 1,500 end-customers of some 30 managed service providers (MSPs) at the start of the USA’s Independence Day weekend.

Update: 6 July 2021. Updated numbers of affected customers added. More affected customers identified.

A Kaseya statement explained: “Kaseya’s VSA  product  has unfortunately  been  the victim of a sophisticated cyberattack.   Due to our teams’  fast response, we believe that this has been localised to a very small number of on-premises  customers  only.“

VSA, the Virtual System/Server Administrator, is software used by Kaseya customers to monitor and manage their infrastructure. It is supplied either as a hosted cloud service by Kaseya, or via on-premises VSA servers. These SaaS VSA servers can be deployed by end-users or by MSPs. Kaseya sends out updates to these VSA servers and, on Friday July 2, an update was distributed that contained REvil ransomware code. It affected fewer than 60 Kaseya VSA customers — but around 30 of them were MSPs, and the code was then sent on to their customers. Potentially thousands of MSP client businesses were infected.

This is known as supply chain attack, and is similar in its basic methodology to last year’s SolarWinds attack, with malware installed via an update server.

Kaseya response 

Kaseya has been putting out regular updates about the attack and its ongoing response efforts. It has turned off its own hosted and also SaaS VSA servers, and states emphatically that customers should turn off their own VSA servers until further notice.

A Kaseya statement on July 4 said: “Our security, support R&D, communications, and customer teams continue to work around the clock in all geographies through the weekend to resolve the issue and restore our customers to service.” 

Customers who receive a ransomware demand should not click on links in the message, as the links may themselves be weaponised. How they should actually respond to the ransomware demand is not spelled out. Meanwhile, they could well find that some of their files are encrypted and will need to recover their contents somehow.

Kaseya advised: “We are in the process of formulating a staged return to service of our SaaS server farms with restricted functionality and a higher security posture (estimated in the next 24–48 hours but that is subject to change) on a geographic basis. More details on both the limitations, security posture changes, and time frame will be in the next communique later today.”

Kaseya CEO Fred Voccola.

Kaseya is working with the FBI, the US Cybersecurity and Infrastructure Security Agency (CISA) and cybersecurity forensics firms, and its executives are contacting affected customers. The company’s R&D engineers are diagnosing the incident breach point and investigating how the attack code has affected customers. A Compromise Detection Tool was rolled out late on June 3 to almost 900 Kaseya customers who had requested it, and an update to this developing tool has also been distributed.

CISA tweet.

The tool can be used to find out if a customer’s VSA server has been infected. Some 2,000 Kaseya customers have downloaded the tool as of July 6.

It is developing a patch that will need to be installed prior to customers restarting their VSA servers, along with a set of recommendations on how to strengthen their security posture. 

Attack vector and scope

Kaseya’s own systems were breached through a vulnerability that it became aware of from the Dutch Institute for Vulnerability Disclosure (DIVD) only recently, and which it was in the process of fixing. A DIVD statement on July 3 said: “We discovered severe vulnerabilities in Kaseya VSA and reported them to Kaseya, with whom we have been in regular contact since then.” The date that DIVD reported the vulnerability to Kaseya is not revealed.

Security firm Huntress Labs is updating a Reddit thread on the incident. It contains details of the initial intrusion into customers’ Kaseya VSA servers:

Huntress Labs Kaseya ransomware intrusion details.

The thread includes these points:

  • More than 1000 businesses had servers and workstations encrypted.
  • It’s reasonable to think this could potentially be impacting thousands of small businesses.
  • Based on the forensic patterns, ransomware notes and the TOR URL, we strongly believe a REvil/Sodinokibi RaaS affiliate is behind these intrusions.

Huntress believes that the cybercriminals exploited a SQLi vulnerability and that an authentication bypass was used to gain access into the VSA servers. 

Sophos malware analyst Per Mark Loman is reported to have said that companies who have been impacted are seeing ransom notes of $50,000 or $5 million if their system is part of a larger corporate network.

According to The Record, Kaseya has been targeted twice before. “In February 2019, the Gandcrab ransomware gang abused a vulnerability in a Kaseya plugin for the ConnectWise Manage software to deploy ransomware on the networks of MSPs’ customer networks.”

The Gandcrab gang rebranded as REvil and then mounted a second attack against Webroot SecureAnywhere and Kaseya VSA products trying to get entry through them to MSPs’ customer networks.

In this latest attack the REvil code apparently tries to disable any local anti-virus code and then executes a phony Windows Defender application. This app runs the ransomware routines to encrypt files in the system.

Number of affected customers

Huntress Labs researcher John Hammond thinks there are more than 1000 affected businesses. (Hammond is tweeting about the incident.) These are located around the globe including in Argentina, Canada, Germany, Kenya, Mexico, The Netherlands, New Zealand, South Africa, Sweden, the UK and the USA and other countries..

Kaseya has since said fewer than 1,500 end-customers were affected.

Bloomberg reported that Synnex Corp. and Avtex LLC are two affected US MSPs. New Zealand Kaseya-using MSP Datacom shut down its servers. An un-named German ISP was hit. Dutch ISPs VelzArt and Hoppenbrouwer Techniek are reported to have been affected.

An 800-branch Swedish grocery chain, the Coop, had its point-of-sale terminals disabled because the actual payment code service came from a Swedish MSP called Visma Esscomn. This is a Kaseya customer with a million customers, according to its press notice about the incident.  It distributed the ransomware code to the Coop’s checkout terminals and they could no longer accept payments.

The Coop initially closed all 800 stores, then set about recovery work. Its customers can use a Coop mobile phone Scan & Pay app in some stores.

Sweden’s State Railway organisation was also hit, as was a Swedish pharmaceutical chain.

Comment

This disgusting cyber-piracy seems unstoppable. The only effective defence for victims would seem to be recovery from immutable backups, preferably with granular point in time recovery and fast restores so that data loss is minimised.

It is ironic that Kaseya offers its own Unitrends backup, which has an anti-ransomware capability. Even more ironically, it also provides a Managed SOC service that provides 24/7 threat monitoring with this pitch: “Stop attackers in their tracks with our managed cybersecurity detection and response solution backed by a world-class security operations centre.”

US president Joe Biden met Russia’s President Putin in Geneva in June and told him that the USA would hold Russia accountable for cyberattacks against US companies and organisations if the attacks originated in Russia. He is reported to have directed intelligence agencies in the USA to investigate the attack. 

If a Russian origin to the Kaseya incident is identified, exactly how Russia would be held accountable is not specified. Meanwhile a thousand-plus businesses around the globe are having their data held to ransom because Kaseya was the unwitting agent used to ship ransomware to their systems.

Future legal actions may try to get Kaseya to pay compensation to affected businesses. The amount of lost business by Sweden’s Coop could run into millions of kronor. Kaseya’s fast and constantly updated response to the attack will help mitigate reputational damage, but the consequential financial cost could be literally enormous.

HPE to acquire cloud disaster recovery startup Zerto

HPE is acquiring IT resilience specialist Zerto for $374 million in cash, and so gaining in-house disaster recovery and general backup products, including cloud-native app protection.

Update: 4 July 2021. HPE will offer employment to all Zerto employees.

Zerto was founded in 2009, by CEO Ziv Kedem and CTO Oded Kedem. It has recorded total funding of $162 million VC financing across five rounds, plus a $20M debt finance facility. It claims more than 9000 customers, including more than 350 Managed Service Providers.

HPE President and CEO Antonio Neri said in a prepared quote: “Zerto’s market-leading cloud data management and protection software expands HPE GreenLake cloud data services, allowing customers to protect their data and rapidly act on insights, from edge to cloud.“

Ziv Kedem wrote in a blog: “We see this as a strategic development for the storage industry — and true recognition of the importance of our cloud data management and protection technology … This acquisition expands HPE GreenLake and accelerates HPE Storage’s transformation to a cloud-native software-defined data services business.”

Zerto CEO Ziv Kedem.

That bears repeating: it accelerates HPE Storage’s transformation to a cloud-native software-defined data services business and, logically, away from a traditional external storage array hardware/software business.

Zerto had focused on disaster recovery, with its continuous data protection and journalling technology. It subsequently moved into general backup and protecting Kubernetes-orchestrated containerised applications, along with migrating them between clouds and on-premises.

Its software replicates and migrates data between vSphere and Hyper-V environments, and natively to Amazon Web Services and Azure. Its hybrid and multi-cloud mobility has support for VMware on public cloud with support for Azure VMware Solution (AVS), Google Cloud VMware Engine, and Oracle Cloud VMware Solution.

According to HPE, the data protection as-a-service market will grow from $7.7B in 2020 to $15.3B in 2024, representing a 19 per cent annual growth rate.

Zerto’s management team is joining HPE following the close of the transaction, expected in the fourth quarter of HPE’s fiscal year 2021. Altogether there are 500 Zerto employees, spread across the USA and Israel. Zerto will be organised under HPE Storage, reporting to HPE SVP and GM Tom Black. An HPE spokesperson said: “Currently we expect to make an employment offer to all Zerto employees. As part of the integration process, the HPE team will work with Zerto on the go-forward plan and structure.”

HPE said Zerto is expected to contribute more than $130 million of run-rate revenue at software gross margins, with approximately one-third of revenues incremental to HPE’s as-a-service annual recurring revenue. Zerto’s software helps customers recover in minutes from ransomware attacks and will soon be available as a service through HPE GreenLake, via the Data Services Cloud Console.

HPE’s Tom Black.

Zerto’s software should benefit from wider exposure to potential customers.

A blog by Black claims that, with Zerto, “we will be able to deliver the most robust disaster recovery as-a-service solution in the industry — for any storage device, any app, any cloud, and any recovery SLA. Zerto brings to HPE the fastest recovery objectives for always-on applications, ransomware protection from increasing cyber threats, and seamless hybrid cloud application and data mobility.”

Comment

This looks like a shrewd move by HPE to enter a growing and relatively early-stage market and build its own technology to compete with other data protection as-a-service vendors, such as Clumio, Cohesity, Commvault, Druva and Veeam’s Kasten. The acquisition price of $374 million is more than twice Zerto’s $162 million total funding.

DDN updates EXAScaler software with new management framework and Hot Nodes

DataDirect Networks (DDN) has released an updated version of its EXAScaler parallel file system software with configuration and management APIs to automate its operation, support for Nvidia’s GPU-Direct Storage, and Hot Node caching.

The company has also added a single SKU offering, combining its A3I AI storage system with Nvidia’s DGX SuperPOD, providing better management insights, and says it has extended its Tintri IntelliFlash portfolio.

DDN A3I system.

The EXAScaler product line delivers Lustre parallel file system arrays based on underlying SFA drive systems. The fifth generation product — EXA5 — was announced in June 2019 and the new EXA6 comes two years later.

There are ES200NVX and ES400NVX 2U 24-bay NVMe SSD appliances, SFA7990X 90x 3.5-inch bay appliances and the SFA18KX, billed as the world’s fastest appliance pumping out 3.2 million IOPS from its 4U box holding NVMe and SAS SSDs and HDDs.

The EXA6 software release adds:

  • EXAScaler Management Framework (EMF) with APIs for configuration and management;
  • Support for the latest Nvidia Magnum IO GPU-Direct Storage;
  • Online upgrades; 
  • Automatic tiering enhancements; 
  • Hot Nodes for client-side persistence.

The Hot Nodes feature automatically caches data on the local NVMe storage of Nvidia GPU systems, reducing IO latency and traffic by avoiding network round trips. DDN says Deep Learning (DL) models often require datasets to be re-read many times, and Hot Nodes cache data in the GPU server thereby avoiding repeated data reads from storage.

DDN Insight is a configuration and monitoring product. Version 4.0 collects data on individual jobs, IO metrics, workload analysis and GPU usage, to provide comprehensive job views and lightweight profiling. The idea is to help admin staff resolve workload and infrastructure problems more quickly.

The A3I + Nvidia DGX SuperPOD single SKU includes the 20 DGX A100 systems inside the SuperPod, InfiniBand infrastructure, all-flash storage and support in a converged infrastructure-style offering. It’s been produced in partnership with Arrow Electronics and is available for certified resellers.

Get a DDN Insight 4 data sheet here.

EXAScaler 6 and Insight 4.0 are targeted for release in Q3 2021. The expanded range of IntelliFlash systems is available now. The Nvidia DGX SuperPOD single SKU is available for DDN and Nvidia certified partners and their customers.

Fungible claims fastest storage node on the market, at 6.55 million IOPS

Dutch high-performance computing (HPC) researchers have reported a Fungible storage node delivering 6.55 million random read IOPS to a single server. Fungible claims this is the world’s highest performance between a single server reading data and a single storage target.

Fungible has developed DPU (Data Processing Unit) chips and used them to power its FS1600 storage node. This is a scale-out, block storage server in a 2U, 24-slot NVMe SSD box with two F1 DPU controllers and access over an NVMe-over Fabrics TCP link. In the Dutch test it was hooked up to a 64-core AMD-powered server across a Jupiter Networks set up.

Pradeep Sindhu, CEO and co-founder of Fungible, said in a prepared quote: “What we are achieving in the lab … can be deployed throughout the world. We believe customers, partners and research institutions can push innovation boundaries with the Fungible Storage Cluster to radically reduce processing time and advance experimental activities in accelerator-based particle physics and astro-particle physics. Ultimately, it is revolutionising the performance, economics, reliability and security of scale-out data centres.”

Fungible FS1600.

Fungible says its IOPS number is nearly 2x better performance than its nearest competitor (which it declined to identify).

The test was run by two organisations: Nikhef, a partnership between the Institutes Organisation of the Dutch Research Council and six universities; and SURF, the collaborative ICT association of Dutch educational and research institutions. These two are focussed on exploring high-performance storage for physics and related HPC science. 

Testing was carried out at Nikhef’s computing centre in Amsterdam. Nikhef is looking for fast and affordable data processing with a view to multiplying the data flows from experiments at CERN in Geneva when the intense high-luminosity LHC accelerator comes online there after 2026.

Tristan Suerink, IT Architect at Nikhef, said: “The Fungible FS1600 will be recognised as the start of a new era in storage, thanks to its unique DPU capabilities. We see a bright future for these devices and DPUs in general.”

Raj Yavatkar, CTO at Juniper Networks, said: “The results achieved are truly ground-breaking and will have far-reaching implications for data centres across the globe.”

Fungible claims that, as a result of this performance, data-centric workloads can be consolidated, leading to an increase in utilisation of storage media and cost per IOPS decrease compared to existing software-defined storage solutions. It claims its technology scales linearly to 300 million IOPS in a single 40 RU rack, and can extend further to many racks. The FS1600 enables data services such as data durability, data reduction, data security in-line and at line-rate — all selectable on a per-volume basis.

Comment

The Fungible FS1600 is basically a proprietary external SAN array and, as demonstrated here, with high read performance. How does it compare to computing products?

According to the Evaluator Group, Pure Storage’s FlashArray//X m70 model runs at 300,000 32K IOPs and 9GB/sec bandwidth. We don’t have a bandwidth number for the FS1600, but would guess it is substantially higher than 9GB/sec.

A single VAST Data enclosure provides 400,000 IOPS from its QLC flash drives shipping out file data.

Dell EMC’s high-end PowerMax systems are much faster than this, with the 4U, 2-brick PowerMax 2000 providing up to 2.7 million IOPS.

DDN’s SFA18K appliance is billed by DDN as the world’s fastest appliance, pumping out 3.2 million IOPS from its 4U box holding NVMe and SAS SSDs and HDDs.

But a 4U Pavilion Data HyperParallel array can deliver up to 20 million IOPS — which would cast doubt on Fungible’s claim that the FS1600 delivers the world’s highest performance between a single server reading data from a single storage target. Hopefully more details will be revealed about the Fungible test to provide a fuller picture.

We are also curious about how an FS1600 would perform using GPU-Direct and pumping data out to Nvidia GPU servers.

Micron earnings: thanks for the multi-million dollar memories

Micron had a great third fiscal 2021 quarter, with large increases in both revenues and profits driven mostly by DRAM — although the Storage business unit let the side down. It also announced its Lehi fabrication plant is being sold to Texas Instruments for $1.5 billion.

Revenues for the third fiscal 2021 quarter were $7.42B, a 36.5 per cent year-on-year increase. It earned profits of $1.74B which compares to the year-ago $803M — a 116.7 per cent jump. The profits represented 23.5 per cent of Micron’s revenues, implying a supplier-friendly pricing environment, with demand higher than supply.

Sanjay Mehrotra.

President and CEO Sanjay Mehrotra issued a quote: “Micron set multiple market and product revenue records in our third quarter and achieved the largest sequential earnings improvement in our history.”

Financial summary:

  • Gross margin — 43 per cent vs 33 per cent a year ago;
  • Cash and cash equivalents — $9.8 billion vs $9.3 billion a year ago;
  • Operating cash flow — $3.6 billion, 48 per cent of revenue;
  • Positive free cash flow — $1.5 billion;
  • Diluted earnings per share — $1.88 ($0.82 a year ago).

Storage lets the side down

Summary of results for each business unit:

  • Compute and networking (DRAM products) — $3.3 billion, up 49 per cent year-on-year;
  • Mobile (Phone DRAM and NAND) — $1.9 billion, up 31 per cent year-on-year;
  • Storage (NAND) — $1.1 billion, zero growth year-on-year;
  • Embedded (DRAM and NAND) — $1.1 billion, up 64 per cent year-on-year.

The Storage BU stands out, with zero per cent growth compared to the 31 to 64 per cent growth rate from the three other BUs. The Storage BU sells SSD and NAND products into the cloud, enterprise and client computing markets.

Mehrotra’s prepared remarks included this point: ”NAND hit record revenue, propelled by record mobile MCP, consumer SSD, and client SSD revenues.” In other words, cloud and data centre NAND revenues did not hit a record number, although there was sequential growth over the quarter, and these look to be the weak points in the storage BU markets.

He said: “Propelled by the transition to DDR5, strong capabilities in graphics memory and the introduction of HBM and NVMe SSD product offerings, Micron’s strong product roadmap across DRAM and NAND positions us for success in the data centre.”

Micron said it will introduce new data centre NVMe SSD products in the next few months, with internally-developed controllers.

Wells Fargo senior analyst Aaron Rakers told subscribers ”We continue to consider MU’s improved positioning in enterprise SSDs via next-gen NVMe drives (with MU internal controllers) as being a 2H2022/2023 story.” That’s a year away then, as the drives will have to be qualified before being adopted and sold by OEMs.

DRAM and NAND revenue details

DRAM represented 73 per cent of Micron’s revenues in the quarter and was up 52 per cent year-on-year with average selling prices rising 20 per cent. NAND provided 24 per cent of its revenues and increased nine per cent year-on-year. Average selling prices for NAND rose in the high single-digit per cent range — less than half of the ASP increase in DRAM.

Micron revenue and profit history to Q3 fy2021.

Micron’s earnings presentation noted that NAND hit record revenue, propelled by record mobile MCP, consumer SSD, and client SSD revenues. Its embedded business exceeded $1 billion for the first time, with record revenue across automotive and industrial markets. Both its 1α DRAM and 176-layer NAND reached a meaningful portion of its overall bit production, and QLC NAND accounted for a majority of its client SSD bit shipments.

Although there has been a drought in Taiwan, where it has manufacturing facilities, it worked around this with no reduction in production output.

It expects DRAM and NAND supply to remain tight into CY22, meaning prices should be high. 

Both memory and SSD demand is high in data centre servers and client PCs, graphics workstations and games consoles. Micron achieved record multi-chip package quarterly revenues in the mobile phone market. Mobile unit sales are expected to grow this year, with an an expected doubling of 5G units in calendar year 2021 to more than 500 million units.

Micron said it had a third consecutive record quarter in the automobile market, driven by continued manufacturing recovery and increased LPDDR4 memory and embedded multi-media card (eMMC) content for in-vehicle-infotainment and driver assistance systems.

Lehi fab

The Lehi fabrication plant, which made 3D XPoint chips, is being sold to Texas Instruments for $900 million in cash and about $600 million from select tools and other assets. Some of the assets have been sold, and Micron will retain the remainder to redeploy to its other manufacturing sites or sell to other buyers.

Micron’s Lehi fab.

TI will offer jobs to all the Lehi employees upon the closing of the sale, and intends to deploy its own technologies at the site. The sale should close later this year.

Back in March, Micron said it would end 3D XPoint chip production and pursue CXL-based memory technology development. Mehrotra said: “We are developing exciting CXL-enabled products and will have more to share on our roadmap in the future.”

Outlook

CFO Dave Zinsner said: “Both DRAM and NAND markets are tight, and we expect pricing increases for both markets in Q4 … Despite cost headwinds, we expect strong improvement in our financial performance in Q4. Our growth opportunity is healthy, and market momentum heading into FY22 is strong.”

Micron is guiding $8.2 billion +/- $200 million revenues for the next quarter — a 35.3 per cent increase on the year-ago $6.1 billion revenues at the mid-point. This would mean $27.63 billion revenues for the full fiscal 2021 year — a 28.9 per cent year-on-year increase, and near Micron’s previous highest-ever annual revenues of $30.4 billion in its fiscal 2018.

Comment

If Micron’s Storage BU had exhibited the same growth as its Mobile BU, then its quarterly revenues would have been approximately $310 million higher. The weak sales of SSDs and NAND products into the cloud (hyperscaler) and enterprise markets is costing Micron dearly. We can hope that Micron will introduce excellent NVMe SSDs as it tries to grow enterprise SSD sales.

We may also expect Micron to become visibly active in the zoned namespace SSD area, where Western Digital is making the running. Samsung recently introduced its zoned namespace SSD, the PM1731a, for enterprises and hyperscaler customers. Zoning can improve both an SSD’s endurance and performance, and Micron needs to get aboard that train.

We love ya! Weka hugs Nvidia’s HGX to its fast filesystem

Weka
Weka video

Weka’s fast filesystem — WekaIO — has added support for Nvidia’s latest offering, the HGX AI-focussed GPU server, shipping file data to its AI-processing GPUs.

This so-called AI supercomputer features an A100 80GB GPU, NDR4 400G InfiniBand, and GPU-Direct Storage software. Weka’s filesystem software can run in an attached Xeon server and ship data over to the HGX using GPU-Direct and avoiding any involvement in data handling by its host Xeon CPU or data staging in the host server’s DRAM.

Liran Zvibel, co-founder and CEO at Weka, issued a quote mentioning the close relationship between Weka and Nvidia: “Nvidia  is not only a technology partner to Weka but also an investor. As such, we work closely to ensure that our Limitless Data Platform supports and incorporates the latest improvements from Nvidia as they become available.”

Weka says it recently delivered among the greatest aggregate NVIDIA Magnum IO GPU-Direct Storage (GDS) throughput numbers of all storage systems tested at Microsoft Research. Tests were run on a system that had WekaFS deployed in conjunction with multiple Nvidia DGX-2 servers in a staging environment. Weka says the engineers achieved the highest throughput of any storage solution that had been tested to date.

DDN and VAST Data are also offering HGX support and shipping GDS-supporting software. Excelero, IBM, Micron, Pavilion and ScaleFlux are due to announce availability of their offerings in the next few weeks or months, with Dell EMC, Hitachi, HPE, Kioxia, Liqid, NetApp, Samsung, System Fabric Works, and Western Digital all possibly (and probably) following them.

Nvidia is leading a great herd of storage suppliers to the enterprise AI promised land, where stored data feeds analytic routines for ever and ever.

StorMagic virtual SAN meets Hivecell stackable edge HCI

Virtual SAN supplier StorMagic has launched Hivecell HCI with StorMagic SvSAN, claiming it delivers edge computing on an enterprise scale for remote sites with no IT tech expertise.

Brian Grainger, President of StorMagic, Chief Revenue Officer and board member, issued a quote: “Together with Hivecell and Intel, we’re proud to deliver a solution that eliminates the need for a large technical staff to install or maintain, while still delivering 100 per cent uptime for the mission-critical edge site. Customers can easily implement, run and scale edge sites from a few to thousands without the heavy IT infrastructure often needed to maintain the hardware.”

Hivecell stack.

Hivecell, or The Ricker Lyman Robotic Co., is based in Beacon, New York and was started up in 2008 by CEO Jeffrey Ricker and President Paul Lyman with $415K in funding, according to Crunchbase. The company produces stackable edge computing boxes that are supplied on a managed service basis and don’t need a rack framework. 

Some Hivecell box details.

These are hyper-converged infrastructure (HCI) appliance boxes, either Arm-powered (Hivecell Horizon) or using an Intel Atom CPU (Hivecell Perimeter). They process information from a set of edge devices, and can send processed data summaries to the public cloud. 

Think of them as occupying the same near-embedded edge HCI device market space as Scale Computing’s HE150 product.

Scale Computing HE150 edge computing box.

Partnering with Intel

The StorMagic Hivecell deal involves the Intel-powered Perimeter system. The two say centralised monitoring, management, updates and upgrades are provided as a service, eliminating capital expenditures and reducing operating expenses. Hivecell HCI has no need for technical staff to install or maintain hardware, and no special requirements for power, cooling or networking. The combined system can be deployed and running with just a click, the two suppliers claim.

HiveCell box specs.

There are three different modes for customer software deployment: cloud-based, enterprise and disconnected.

Ricker said: “It’s becoming clear that edge computing is the future for data processing, and we are thrilled to partner with StorMagic and Intel to offer a fully complete, edge-as-a-service platform with a high-availability storage solution that doesn’t require buying or maintaining software or hardware and can be installed with no training.”

To find out more, a joint company solution paper should be live on the StorMagic site later today.

Forget passwords — try fingerprint drive access

No need to bother with passwords — Verbatim has built external drives with secure fingerprint access to prevent data loss or unauthorised access.

There are 512GB and 1TB handheld recorder-sized SSDs with an M.2 format card inside them and a fingerprint reader at the top of the device. The disk version uses a 2.5-inch drive with either 1TB or 2TB capacity and the same fingerprint reader. Both the SSD and HDD connect to a host using USB 3.2 gen-1 across a USB-C connection with up to 5Gbit/sec bandwidth.

Drive contents are encrypted using AES-256 hardware encryption to add extra security. Up to eight users’ fingerprints can be registered to provide for secure file sharing.

Executive Fingerprint Secure drives; SSD (left) and HDD (right).

These drives can also be connected to a TV once a fingerprint is registered — an unusual feature.

Verbatim provides Nero Backup software for Windows with the drives to back up all files, folders, and drives from laptops or PCs as well as schedule auto back up to a set time and day to ensure that it gets done regularly. 

There is no Mac support for Nero, which is a pity, and the drive capacities are too low for anything over 2TB of data on your laptop or desktop PC. We would hope Verbatim will eventually bring out 3TB and 4TB models.

Recommended retail prices: 

  • 1TB (PN 53652) Executive Fingerprint Secure HDD — £79.99;
  • 2TB (PN 53653) — £99.99;
  • 512GB (PN 53656) Executive Fingerprint Secure SSD — £87.99;
  • 1TB (PN 53657) — £134.99.