Home Blog Page 134

Pliops and phoenixNAP team up to turbocharge Redis

By
Chris Mellor
-

Hardware-accelerated box-of-flash startup Pliops is partnering with IaaS provider phoenixNAP to offer petabyte-scale Redis clusters to users with in-memory-class performance.

Redis is a distributed open source, in-memory key:value database. Pliops’s XDP-Rocks hardware is a PCIe card that offloads low-level storage IO for a key:value store such as Redis from a host server. Distributed in-memory databases are getting larger and need more DRAM, which is expensive and limited in scale by a server processor’s socket count. Pliops XDP-Rocks makes flash drives an extension of the server’s memory with hardware acceleration making up for NAND’s slower speed versus DRAM. It provides sub-millisecond latency for large data sets.

Uri Beitler, Pliops
Uri Beitler

Uri Beitler, Pliops founder and CEO, said: “Terabytes of data sets per node and sub-millisecond SLAs mean that Redis users can achieve faster time to insights and drastically reduce their hardware infrastructure investment.”

Previously Redis users needing more DRAM than their servers could supply simply had to buy more servers. A Pliops setup with up to 128TB of flash capacity managed per card, soon to be 256TB, means petabyte scale Redis clusters can be deployed with fewer servers than before, meaning less power, cooling and rackspace is needed, all contributing to saved cost.

Bare Metal Cloud (BMC) servers with XDP-Rocks cards are offered by phoenixNAP. William Bell, EVP of Products at phoenixNAP, said: “We are excited to enhance our partnership with Pliops by integrating their groundbreaking Redis acceleration technology into our BMC services. This integration not only meets the demanding data processing needs of our clients but also helps reduce their infrastructure footprint and IT spending.” 

Pliops XDP
Pliops XDP card

Pliops and phoenixNAP reckon that the BMC server/XDP-Rocks combo can reduce the in-memory server footprint by up to 70 percent and enables 7x higher Redis throughput performance with 4x lower latency.

The XDP-Rocks card is equivalent to an all-flash array controller, but mounted in the host server’s chassis and dedicated to key:value store processing, supporting MySQL, MyRocks, Redis, KV-Rocks, TiDB, Kafka, Spark and others. As far as a host server is concerned, the XDP replaces storage engine software, such as InnoDB, RocksDB and WiredTiger, in its own storage IO stack.

Workload trials of the Pliops-powered Redis on Flash service are available through phoenixNAP.

Storage news ticker – May 26

By
Chris Mellor
-

Data protector ArcServe paid DCIG to review its UDP 9.0 backup product. The resulting study found that ArcServe UDP 9.0 offers organizations a clear edge in protecting against ransomware while tackling the persistent challenges of managing backup and recovery complexities. It hasn’t made the review public.

…

Disaster recovery specialist Continuity Center and Backblaze have announced Cloud Instant Business Recovery to help businesses of all sizes prepare for and recover from disaster – from ransomware to natural catastrophes. They say small and mid-sized businesses have historically been priced out of existing solutions designed for large enterprises with massive budgets and teams. Cloud Instant Business Recovery (Cloud IBR) instantly recovers Veeam backups from the Backblaze B2 Storage Cloud. The fully automated service deploys a recovery process through a simple web UI and in the background uses phoenixNAP’s Bare Metal Cloud servers to import Veeam backups stored in B2 and fully restores the customer’s server infrastructure.

…

DataOps.live – which enables companies to build, test and deploy applications on Snowflake – has raised $17.5 million in an A-round from Notion and Anthos, and is growing at speed. Its breakout fiscal 2023 saw 400 percent ARR growth.

…

Dell will be reselling Continuity’s StorageGuard, a security posture management solution which hardens storage and backup systems while guaranteeing compliance with relevant vendor, industry and regulatory best practices. StorageGuard scans customer environments to ensure they are secure and follow all relevant vendor, industry and regulatory best practices.

…

Google has announced GA of its C3 machine series VM with gen 4 Xeon SP (Sapphire Rapids) processors and made it available to GCE and GKS customers. C3 uses DDR5 memory and is currently available in the highcpu (2GB/vCPU) memory configuration. Standard (4GB/vCPU) and highmem (8GB/vCPU) configurations, as well as local SSD (up to 12TB), will be available in the coming months. C3 VMs use Google’s latest Infrastructure Processing Engine (IPU) to offload networking, deliver high performance block-storage with Hyperdisk, and speed up ML training and inference with Intel AMX.

…

Hitachi Vantara says the Arizona Department of Water Resources has chosen its Pentaho Data Catalog to mitigate its water crisis. The AZ State has collected centralized data from 330,000 distributed water resources that lead to better insights, improving water use for the state’s 7 million residents. To effectively manage Arizona’s water supply, the Department of Water Resources collects, stores and conducts analysis on reported water uses of nearly 6 trillion gallons per year across thousands of wells and surface water sources. By analyzing data against records of geolocation coordinates, depth to water and uses, the agency is better equipped to establish the adjudication of water rights with trusted, accurate data.

…

At its Innovative Data Infrastructure Forum (IDI Forum) 2023 in Munich, Huawei has added 3 entry-level arrays and an entry level protection array to its OceanStor line. This consists of 3000, 5000/6000 and 8000/18000 all-flash arrays (AFAs) and the 2000, 5000, 6000 and 18000 hybrid flash/disk arrays; the bigger the number the bigger and more powerful the product. The newcomers are the 2000 AFA and 2200 and 2600 hybrid arrays. It’s also added an entry-level X300 to its OceanProtect products, augmenting the existing X6000, X8000 and X9000 models.

Huawei believes emerging big data and AI applications pose higher requirements for the parallel processing of diversified data. Cloud-native applications are becoming more prevalent in enterprise data centers and reliable and high-performance container storage will be a necessity. The read/write bandwidth and I/O access efficiency of scale-out storage need to be significantly improved. An intelligent data fabric is required to implement a global data view and unified data scheduling across systems, regions, and clouds. 

…

IDrive e2 announced the release of its on-premises object storage appliance at VeeamON 2023. It will enable users to radically simplify the ability to manage, store, and protect data while allowing S3/HTTP access to any application, device or end user. A spokesperson told us: “You recently wrote about Object First, so we will compete with them, our pricing is available upon request but we will be more aggressive than Object First, and considering competitors like Wasabi or Backblaze, we are the first to launch such a device.” 

…

Intel is volume shipping the Agilex 7 FPGA with the R-Tile chiplet. This FPGA is the first with PCIe 5.0 and CXL capabilities and the only FPGA with hard intellectual property supporting these interfaces. The “hard intellectual property” term signifies that chip designers can’t significantly modify the digital IP cores involved. Intel says by using Agilex 7 with R-Tile, customers can seamlessly connect their FPGAs with processors, such as 4th Gen Intel Xeon Scalable ones, with the highest bandwidth processor interfaces to accelerate targeted datacenter and high-performance computing (HPC) workloads. 

Intel Agilex FPGA

…

Kasten by Veeam has released its Kasten K10 v6 Kubernetes data protection software. It includes features to help customers scale cloud-native data protection more efficiently, better protect applications and data against ransomware attacks, and increase accessibility with new cloud native integrations. It now supports Kubernetes 1.26, Red Hat OpenShift 4.12, and a new built-in blueprint for Amazon RDS. The platform also has hybrid cloud support on GCP, cross-platform restore targets for VMware Tanzu environments, and new Cisco Hybrid Cloud CVD with Red Hat OpenShift and Kasten K10. There are new storage options for NetApp ONTAP S3 and Dell-EMC ECS S3.

…

Lenovo says it is the number 5v global player in storage, behind, we understand, Dell, HPE, NetApp and Pure Storage. It was number 8 in its previous fiscal year. Its storage revenues, inside its Infrastructure Solutions Group, were up 208 percent Y/Y in its Q4 fy22/23 results. Servers grew 29 percent while its SW unit grew 25 percent ; storage strongly outgrew servers and SW. ISG made $2.2 billion in the quarter and $9.68 billion in the full year, up 37 percent , with the component server, storage and SW contributions not revealed.

…

Liqid has announced the successful delivery and acceptance of the largest composable supercomputers by the US Department of Defense (DoD). Liqid served as the prime contractor for the $32 million contract, delivering two systems to the Army Research Laboratory DoD Supercomputing Resource Center (ARL DSRC). These systems boast the largest composable GPU pools worldwide, with 360 Nvidia A100 GPUs. 

…

The LTO organization reports that a record 148.3EB of total compressed LTO tape capacity shipped in 2022, up a barely noticeable 0.5 percent over 2021. LTO says it’s a strong result. LTO-9 technology (45TB compressed) capacity shipments are demonstrating the most rapid capacity adoption rate since LTO-5. Tape media unit ships dropped, continuing a long-term trend.

Tape storage capacity
Tape storage shipments

There are no numbers for tape media shipments but it looks like a rough 22 percent drop from 9 million cartridges in 2021 to some 7 million cartridges in 2022. On that basis, a 22 percent unit shipment drop and mere 0.5 percent capacity increase, we can say the tape market almost declined in 2022 compared to 2021.

…

TechCrunch reports that Microsoft has launched Fabric, an end-to-end data and analytics service (not to be confused with Azure Service Fabric). It centers around Microsoft’s OneLake data lake, but can also pull in data from Amazon S3 and (soon) Google Cloud Platform, and includes everything from integration tools, a Spark-based data engineering platform, a real-time analytics platform, and, thanks to the newly improved Power BI, an easy-to-use visualization and AI-based analytics tool.

…

MSP data protector N-able is collaborating with the US Joint Cyber Defense Collaborative (JCDC) to help create a more secure global ecosystem and work towards helping reduce security risk for MSPs and their customers.

…

Resilio will soon be launching a private beta for a new scale-out file transfer technology that will dramatically improve the speed of file transfers for data-intensive companies. It says internal Resilio Connect tests, using AWS and Azure respectively, are regularly exceeding 100Gbps. Cloud scenarios tested include transferring a 1TB data set between Microsoft Azure regions in 90 seconds and transferring a 500GB data set in Google Cloud from London to Australia in 50 seconds.

Its automated scale-out technology features horizontal scaling with linear performance acceleration across multiple nodes in a cluster, collectively pooling Resilio agents to transfer and sync unstructured data in parallel. If an organization has 10 Agents with a 10Gbps bandwidth connection, files will transfer at 100Gbps. And through custom WAN optimization, Resilio overcomes latency due to distance.

Companies interested in participating in Resilio’s closed beta may sign up here. A technical blog post with additional details is available here.

…

Data protector Rubrik has appointed Toby Keech as VP for UK and Ireland. He joins from Zscaler, where he headed up UK&I and MEA business units.

…

Secure Data Recovery purchased 100 discarded drives (for less than $100 total) to see what they could recover. Its engineers recovered data from 35 hard drives without repair (they found 31 hard drives in the sample were either damaged or encrypted). In total, the 35 unsanitized hard drives contained 5,786,417 files. All devices were listed as refurbished, used, or for parts. The release date of the hard drives ranged from 1994 to 2022.

Sensitive material left on discarded storage

…

Snowflake revenue for Q1 fiscal 2024 was $623.6 million, representing 48 percent year-over-year growth. Product revenue for the quarter was $590.1 million, representing 50 percent year-over-year growth. The company now has 373 customers with trailing 12-month product revenue greater than $1 million and 590 Forbes Global 2000 customers. Net revenue retention rate was 151 percent as of April 30, 2023. Snowflake is acquiring Neeva, a search company that leverages generative AI and other innovations to allow users to query and discover data in new ways. 

Snowflake revenues

…

Swissbit Device Manager (SBDM) is the latest edition of Swissbit’s software tool for storage devices. It allows customers to initiate firmware updates in addition to insight into the lifecycle status of Swissbit storage devices. It supports customers who are in the design-in stages, giving them flexibility for extensive testing and customization. A user-friendly app presents all the captured data. It serves as a main interface for all Swissbit storage products. The Swissbit website offers a free download of the tool, which is available for both Windows and Linux.

…

According to new data in the Veeam 2023 Ransomware Trends Report, one in seven organizations will see almost all (>80 percent) data affected as a result of a ransomware attack – pointing to a significant gap in protection. It says attackers almost always (93+ percent) target backups during cyber attacks and are successful in debilitating their victims’ ability to recover in 75 percent of those events, reinforcing the criticality of immutability and air gapping to ensure backup repositories are protected.

ChatGPT, LLMs, and storage

By
Chris Mellor
-

Analysis: The IT analytics world is being dominated by a storm of interest in Large Language Model (LLM) machine learning and generative AI (GAI). The release of the ChatGPT chatbot at the end of November last year generated a huge amount of interest, with 1 million users in a week and an upsurge in similar foundation model applications such as Google’s Bard and Amazon’s Titan.

Such GAIs, with their ability to understand text requests and output competent answers, have the potential to be applied across the entire enterprise and public sector IT landscape enabling fundamentally better search and analytics. Their use threatens/promises to replace and/or improve the productivity of knowledge workers of all sorts, ranging from call center and inside sales staff to accountants, lawyers and financial analysts over the next decade.

The fast and high rising interest in GAIs will have an effect on the IT industry and such effects are being researched and predicted by analyst and research houses like Forrester, Gartner and others. Parts of the data storage industry faces strong sales benefits from GAI adoption and we have attempted to catalog them.

A 126-page William Blair document, “Generative AII: The New Frontier of Automation”, provided much information for this effort.

Hardware

DRAM – More, more, more will be needed for CPU/GPU servers running LLMs for training and inference and that includes high High Bandwidth Memory HBM for GPUs.

PCIe – PCIe 4 and 5.0 component suppliers should get ready for a surge in demand.

CXL – CXL 2.0 memory pooling should receive a gigantic shot in the arm from LLMs and that incudes CXL hardware suppliers, meaning DRAM expanders, CXL switches and components Companies such as Micron, Samsung, SK hynix and others should all benefit.

NAND and SSDs – More, more and more will be needed, with an emphasis on NVMe access, PCIe 4.0 and 5.0 connectivity plus a mix of performance and capacity. This suggests QLC and high layer-count TLC NAND will benefit. All the NAND fabs and SSD vendors should be focussed on this market.

Storage arrays – The need will be for high capacity and high speed IO. AI/ML clusters will need petabyte levels of capacity. LLM training runs will require high speed dataset reading and checkpoint writing. This will need parallel access facilitated by hardware and software. LLM inferencing runs will need high read access rates with parallel data delivery paths to the processors.

We think this will primarily benefit file access all-flash arrays using NVMe protocols and with GPUDirect support for Nvidia GPU servers. Suppliers such as Dell, DDN, NetApp, Pure Storage (AIRI) and VAST Data are well-positioned to capitalize on this. Panasas sees an opportunity in edge AI. Object storage and block storage suppliers are not so well-positioned.

Vendors lacking GPUDirect support should, we think, pursue this with urgency.

Software

CXL-focused software – Suppliers such as MemVerge and Unifabrix should expect to see a large and sustained rise in interest for their products.

Data analytics – suppliers need to investigate adopting LLMs front ends as a matter of urgency.

Database, warehouses and lakehouses – They need to support the vector embeddings needed by LLM models. Vector databases that have the support will become more important. The need to investigate and trial chatbot front ends for their users is intense. This will enable non-data scientist and unskilled SQL users to run sophisticated analytics. They also have an opportunity to find ETL (Extract, Transform and Load) processes to pump selected data quickly out to LLMs for training and inference runs. See SingleStore and Databricks as examples.

Data managers – They can benefit by applying LLM technology to anayse their data sets and by feeding LLM proceses with data. See Cohesity as an example.

High speed arrays – Suppliers may well find it worthwhile to port their software to the public clouds which will run GAI models. That way they can support their customers who adopt a hybrid on-premises/public cloud approach to running LLMs.

Scale-out parallel filesystem – Suppliers such as IBM (Storage Scale) and WEKA are well placed as their existing customers adopt GAI technology and new customers look to them for fast and high-capacity file access software. These suppliers could be big winners.

Indirect beneficiaries and the unaffected

Cloud file services suppliers – They can use datasets they store in the cloud to feed LLMs but the data will need moving from their underlying object vaults to faster access stores; some form of ETL in other words. That is unless the CSPs like AWS, Azure and Google find some GPUDirect-like way of pumping data from S3 and Azure Blobs to their GPU instances. 

Data orchestrators – They may get an indirect benefit if they can orchestrate data needed to feed LLMs.

Data protection and security – Vendors need to check out chatbot interfaces to their management facilities to better protect and secure datasets and identify vulnerabilities. Domain-specific chatbots could inspect an organisation’s attack surface and identify actions to protect it. Data Protection backup data sets could feed LLMs given ETL processes.

Disk drive arrays – Your products are too slow and can only be used as a second tier behind flash storage primary stores.

Lifecycle managers – Vendors need to research how chatbot interfaces could make their users more productive.

Security suppliers – Chatbots and other AI tech could make them more effecting in detecting and responding to malware and handling user interactions.

Software-defined storage suppliers face being left behind by the GAI tsunami unless they find some means of satisfying the high performance access needed.

Storage admins – Chatbots could make them more productive or be used to enable less skilled staff to do more skilled work.

Tape systems – Archival systems are just too slow to feed data to LLMs, but still have their place.

Web3 – Such storage is going nowhere in an LLM world. It’s too slow.

Nutanix sales up but customer acquisition rate drops

By
Chris Mellor
-

Hybrid cloud and hyperconverged software and cloud vendor Nutanix beat Wall Street with its third fiscal 2023 quarter results, but its customer acquisition rate is slowing.

Revenues in the quarter ended April 30 were $448.6 million, up 11 percent year-on-year and beating last quarter’s outlook of $430-440 million. There was a loss of $81.2 million, lower than the year-ago $112 million loss. Annual Contract Value billings rose 17 percent to $239.8 million and annual recurring revenues went up 32 percent to $1.47 billion.

CEO Rajiv Ramaswami summarized the results: “Our business performed well in the third quarter against an uncertain macro backdrop as the value proposition of our cloud platform continued to resonate with customers.”

Gross margin was 81.4 percent, up 80.2 percent annually, and free cash flow was  $42.5 million compared to -$20.1 million a year ago. The operating cash flow came in at $64.3 million.

CFO Rukmini Sivaraman said: “Our third quarter results continued to demonstrate a good balance of growth and profitability, resulting in year-to-date ACV Billings growth exceeding 20 percent, combined with strong year-to-date free cash flow generation. We continue to execute on our growing base of subscription renewals and remain focused on sustainable, profitable growth.”

The average contract term remained at three years.

William Blair analyst Jason Ader said: “The third quarter outperformance was driven by strength in renewals, with customers renewing at higher prices and some customers co-terming multiple renewals … We expect continued progress on profitability and cash flow.” 

The prior quarter’s results were marred by a discovery that third-party software, obtained for evaluation, had been used to gain revenue. This prompted an audit committee inspection and delayed quarterly report filing to the SEC. Ramaswami said: “We are … pleased to have completed the Audit Committee investigation regarding third-party software usage and to have filed our Form 10-Q for our second quarter of fiscal 2023.”

Evaluation software from two software providers was used in a non-compliant manner since 2014 and certain employees concealed this in violation of Nutanix’ policies. They have been fired.

Inadequate financial reporting controls are being fixed. This all resulted in cumulative estimated expenses of $11 million and there will be an ongoing annual impact in the low single-digit millions.

US law firms like Kuznicki Law, Glancy Prongay & Murray, and others are looking to assault Nutanix with class action law suits for investor losses over the issue.

Nutanix gained 430 new customers in the quarter, down from 480 in the prior quarter and 530 in the quarter before that.

Although the new customer acquisition rate is trending lower (see chart above), and for the third quarter in a row, customers are spending more money with Nutanix. 

Life time bookings:

  • 133 customers with >$10million – up 30 percent
  • 209 customers with $5 – $10 million – up 20 percent
  • 301 customers with $3 – $5 million – up 19 percent
  • 1,442 customers with $1 – $3 million – up 18 percent

Land-and-expand is still working in Nutanix’ favour.

Ramaswami was asked in the earnings call how the Broadcom-VMware acquisition was affecting Nutanix. He said customer engagement had risen as a result but, since sales cycles for VMware to Nutanix migration were nine to 12 months, there was no visible revenue impact yet.

The Q4 outlook is for $470-480 million in revenues, a somewhat surprising 23 percent higher than a year ago at the mid-point. The full fy2023 revenue outlook is $1.84-1.85 billion, up 14.3 percent on fiscal 2022. Sivaraman said the good outlook was due to Nutanix “seeing continued new and expansion opportunities for our solution despite the uncertain macro environment,” and also due to better renewals income with improved execution by the renewals team.

Seagate waits for HAMR time amid debt paydown

By
Chris Mellor
-
Seagate HAMR technology
Seagate HAMR technology

Seagate wants to raise $1 billion in a private bond placement to pay off debt as it begins the process of laying off staff following dreadful Q3 results.

Revenue for the three months ended 31 March fell 33.6 percent year-on-year to $1.86 billion – the third consecutive quarter of shrinking sales. A loss of $433 million contrasted sharply with the year-ago $346 million profit. It still paid dividends, however – $145 million – and had a $174 million free cash flow versus $363 million. Seagate said it needed to lower expenses and decided to make a round of layoffs and slash the most senior exec base salaries to zero for six months. 

More than 100 job losses were recently reported at Seagate’s Springtown disk drive plant in Northern Ireland, for example. Unconfirmed reports in thelayoff.com claimed a 20-plus percent cut across the board. A Seagate spokesperson said up to 12 percent of global staff were affected.

The spokesperson told us today: “As noted in our FY23 Q3 Earnings press release on April 20, 2023, Seagate announced a restructuring program to reduce our cost structure in response to changes in macroeconomic and business conditions.

“These actions include a global workforce reduction, which is one of the most difficult decisions a leadership team undertakes. Our goal is to take these steps thoughtfully, treating all employees in the affected roles with the respect they deserve for their contributions to Seagate.  We expect most of these global actions to be complete before the close of our fiscal year on June 30, 2023.”

Proceeds of the bond sale will be used to repay a senior notes (bond) issue due in 2024 and $450 million in outstanding loans under a credit agreement, with the remainder used for general corporate purposes, such as other debt repayments, capital expenditures and investment in the business. Seagate’s total debt at the end of its Q3 was $5.96 billion, compared to $5.64 billion a year earlier.

A Seagate 8K SEC filing, dated May 19, highlights a credit agreement amendment during a covenant relief period running from May 19 to June 27, 2025 (a covenant is an agreement between lender and borrower laying out loan policy conditions). The amendment permits a maximum net leverage ratio of 6.75 in the current quarter (ending June 30) shifting to a maximum total leverage ratio of 4 to 1 for any fiscal quarter ending outside the covenant relief period.

A leverage ratio refers to the amount of debt carried by a business compared to assets such as equity, with the notion of ensuring that a business has or generates enough cash to cover its debts.

Seagate needs to increase its allowable leverage ratio because it is operating in a severe downturn and bringing in fewer dollars. It is relying on its HAMR technology disk drives, with 30TB+ capacities, to arrive this quarter and give it a lead over Western Digital and Toshiba. That may generate increased sales and provide a revenue and cash boost. That’s the plan anyway.

How to master multicloud from the ground to cloud (and back again)

By
Sharon Maher, Dell Technologies
-
Three clouds
Three clouds

Commissioned: We live in a rich world of cloud services. Inside many companies, innovation engines within different lines of businesses utilize these services to gain instant access to cutting-edge tools, enabling them to work with rapid scalability and high availability. This has unlocked an unprecedented rate of digital transformation in the past several years.

But while this new era of cloud services has brought with it many advantages, it has also presented certain challenges. IT organizations must now grapple with how to manage, govern and rationalize a complex ecosystem that spans across public clouds, on premises, colocation facilities and edge locations. As a result, they may discover the innovation sought by consuming multiple cloud services is stymied by the toil of managing it all. This is what’s known as multicloud by default.

Shifting from multicloud-by-default to multicloud-by-design

There are two key reasons organizations rethink their multicloud approach. The first is to increase security, compliance, and governance. The second reason is to manage costs more effectively.

Many organizations find cloud invoices get complicated. And when you finally add up the cost of your multicloud-by-default estate, you might find the bill is bigger than you expected. In some cases, this has led organizations to repatriate workloads from the public cloud on premises, or to colocation facilities.

For example, 37signals has estimated they will save seven million over five years repatriating workloads onto their own infrastructure. Of course, your mileage will vary but for 37signals, it’s material for their business.

Naturally, there are instances where it doesn’t make sense to move workloads out of public clouds. For example, some organizations use cloud-based AI/ML capabilities that can’t be replicated on premises or simply wouldn’t run there as well. There are also complexities around data gravity and often costs associated with moving data from one location to another.

Every organization is unique; the right solution for you will be based on your unique needs. My advice is to spend time thinking through your workload placement – what should be in the public cloud and what might benefit from being brought back in-house. This is known as a multicloud by design approach and it allows you to unlock all multicloud has to offer without being limited by its complexities.

Embracing multicloud by design through ground-to-cloud and cloud-to-ground strategies

So how can you adopt a multicloud by design strategy? There are two ways to address this problem. The first is to take technologies your organization already uses and run them in the public cloud in a ground-to-cloud approach. This lets your teams use familiar tools in public cloud environments, so you leverage existing skillsets, spend less time managing data and infrastructure and even spend down existing public cloud commitments.

Another way is to bring public cloud stacks to dedicated IT environments in a cloud-to-ground approach. Developers use a lot of cloud-based tools and tool chains. But few organizations are running 100 percent of their workloads in the cloud – most have at least some on premises infrastructure. This means they need solutions that help bridge the divide. Utilizing multicloud appliances that can replicate cloud stacks on premises offers the best of both worlds. You get access to cloud services within your data center or colocation facility and you can seamlessly move data back and forth from the cloud using the same storage technology.

How to bring multicloud by design to your organization

At Dell we make these strategies possible through our Dell APEX portfolio. APEX was designed to simplify the complexities organizations face when it comes to multicloud. This means everything from applying management consistency across multicloud environments, to bringing cloud stacks to dedicated IT environments, to adopting cloud experiences on premises. This helps you move faster, react quicker to challenges and opportunities and ultimately, innovate at scale. Here’s where you can learn more about Dell APEX.

Sponsored by Dell.

Infinidat upgrades to hybrid cloud storage

By
Chris Mellor
-

High-end array supplier Infinidat says it is now a hybrid on-premises and public cloud storage supplier with strengthened cyber-resilience and new deep scanning and indexing detection features.

Update. AWS implementation details added. 26 May 2023.

Infinidat provides high-end primary block and file storage arrays competing with Dell’s PowerMax and similar class storage gear from Hitachi Vantara, HPE, IBM and NetApp. Infinidat’s selling points for its hybrid and all-flash InfiniBox arrays is based on its unique memory caching, active-active-active architecture, a 100 percent availability guarantee, and strong data and cyber-protection features. Now it has ported the InfuzeOS to the AWS cloud to support disaster recovery, business continuity, backup, test/dev ops, app integration, and proof-of-concept work plus burst storage and storage tiering.

Eric Herzog, Infinidat
Eric Herzog

CMO Eric Herzog said: “With Infinidat’s operating system now in the public cloud, we can provide a consistent core-to-cloud experience through our full support for hybrid cloud deployments.

“Simultaneously, we’re addressing concerns about cyberattacks by adding cyber storage detection … with the introduction of InfiniSafe Cyber Detection.”

InfuzeOS uses EC2 and EBS instances in AWS. Herzog told us: “We use an EC2 instance for the compute portion and the storage that it connects to is AWS EBS.  A recommended EC2 instance gets deployed and then attaches to EBS that is provisioned to the EC2 instance for storage capacity. InfuzeOS Cloud Edition then provisions storage from EBS that is connected to the EC2 instance.”

Customers can deploy a full version of InfuzeOS residing in their AWS environment in a single node InfiniBox configuration. That is in contrast to three-controller architecture of InfiniBox and InfiniBox SSA. That means performance and availability is limited to SLAs derived from AWS and not the same as an on-premises Infinidat system

We understand other cloud environments will follow on from AWS. On-premises InfiniBox systems see a public cloud InfuzeO Cloud Edition instance as if it is just another InfiniBox platform.

The move gives Infinidat a stronger public cloud presence in the block and file space, and follows closely behind Dell’s cloud file and block storage news. Dell’s PowerMax does not have a cloud presence, though.

Infinidat also supplies InfiniGuard data protection arrays, and has InfiniSafe software technology, protecting snapshots that can be used with InfiniBox and InfiniGuard. This InfiniSafe technology now includes added detection features to find and intercept malware attacks faster. It scans and indexes immutable snapshots of all files, applications, core storage infrastructure (such as volumes), and databases for signs of cyber threats, helping to ensure the integrity of all the data in the system.

Cyber Detection uses machine learning models. Over 200 points of determination are included, using content-based analytics that inspect inside files for subtle signs of attack. A post-attack dashboard, with forensic reports, details the last known good copy of the data for rapid recovery. Infinidat claims a 99.5 percent confidence level in detecting cyber threats, which helps dealing with false positive/negatives and reduces the effort in any additional forensics work.

Get an InfuzeOS datasheet here. InfuzeOS Cloud Edition is available from the end of May and directly through the AWS Marketplace later in 2023. InfiniSafe Cyber Detection will ship in 2H 2023 and works on the InfiniBox and InfiniBox SSA platforms.

Panasas shrinks HPC powerhouse into edge-sized box

By
Chris Mellor
-

Panasas has halved the size of its ActiveStor Ultra parallel file system node to make a lower-cost edge datacenter box.

The ActiveStor range runs scale-out PanFS, with NFS and SMB/CIFS protocol support, and optimized performance to match workflow needs. The OS protects against drive, node, and site failures. PanView and PanMove services provide data visibility and movement capabilities.

Tom Shea, Panasas CEO, said: “Our customers require storage infrastructures that can dynamically adapt to a changing mix of both AI/ML and traditional HPC workflows, but also have the capacity to support workflows across core data centers, remote satellite locations, and into the cloud.“ 

Panasas ActiveStor Ultra Edge 100
ActiveStor Ultra Edge 100 enclosure with 12 drive slots

“The reduced footprint of ActiveStor Ultra Edge delivers the required performance, reliability, and ease of management that Panasas is known for, making it ideal for smaller and remote HPC datacenters.”

The Ultra 100 Edge system comes in a 2RU box and our table shows how it stacks up against the other ActiveStor systems:

ActiveStor performance

There is one physical enclosure that contains four nodes and two redundant 2200W titanium-level power supplies. There are two chassis versions. A Smart enclosure has one director node and three storage nodes. An Expansion enclosure has four storage nodes. The director node processes file system metadata, coordinates the activities of the storage nodes and Direct-Flow software drivers for file access, manages membership and status within the storage cluster, and controls all failure recovery and data reliability operations. Both node types contain servers with one M.2 NVMe SSD, and run the PanFS file system. The storage nodes use 8TB or 16TB disk drives.

The system then comes in two standard configs, Minimum and Base, varying in storage capacity:

Panasas ActiveStor Ultra Edge capacities

Panasas claims the system is simple to operate, needing part-time attention from a single person, and costs less than its other products. View it as an HPC storage appliance for smaller-scale workloads in the genomics, cryo-FM, and decentralized instrumentation areas.

We view Panasas, at least in the enterprise AI/ML market, as being positioned with smaller systems than DDN’s Lustre-using ExaScaler systems, although there is a fair degree of overlap. Panasas has indicated it did not favor QLC flash in the past, preferring disk. DDN has recently adopted QLC flash drives and Panasas may be evaluating them too. Generative AI workloads may be so valuable to customers that they’ll pay for QLC flash over disk.

You can access an Ultra Edge 100 datasheet here and a solution brief here.

Dell unveils Databricks deal, plus storage hits Apex cloud

By
Chris Mellor
-
Dell logo
Dell logo

Analysis: A deeper look into Dell’s APEX announcements in Vegas this week reveals a deal with Databricks for Dell storage to feed its lakehouse, the use of a stopgap block service, and both PowerScale file and PowerProtect being migrated to AWS.

Making Dell’s file and block storage available in the public cloud is the goal of Project Alpine, announced a year ago. This was part of its APEX Multi-Cloud Data Services initiative. Dell wants to provide a consistent public cloud-to-on-premises experience with PowerStore (file and block), PowerFlex (HCI block), PowerScale and/or ObjectScale (file and object).

At Dell Technologies World in May last year it had put its PowerProtect software into the AWS and Azure clouds in the form of PowerProtect Cyber Recovery and a preview showed Dell storage could feed data into Snowflake’s data warehouse. A demo showed PowerScale’s Isilon SW operating in AWS.

Later, in November that year Dell said its PowerFlex hyperconverged infrastructure system was available on the AWS marketplace as part of the APEX program. PowerFlex is like VxRail but doesn’t use the vSphere hypervisor. It scales out to thousands of nodes and originated as a ScaleIO upgrade and rebrand. PowerFlex provided block storage in AWS using Elastic Block Storage (EBS) or the EC2 Instance Store. 

Now, a year later, Dell has announced that APEX Storage for Public Cloud, with block and file storage,  is available in AWS and will debut in Azure later this year.

The announced Databricks partnership follows on from the earlier Snowflake deal and we have APEX Protection Storage for Public Cloud in existence as well.

APEX File Storage for AWS is based on PowerScale’s OneFS, the Isilon scale-out filesystem software, but APEX Block Storage for AWS and Azure is not based on block storage software from its PowerStore or PowerMax arrays. PowerStore’s operating system provides unified file and block storage and, like the PowerMax OS, is not a scale-out system.

Instead the scale-out PowerFlex system software is used to provide block storage. The public cloud scales out by adding instances and PowerFlex matches that capability. This means that PowerFlex customers will see a consistent environment across the on-premises,  AWS and Azure public clouds, as will PowerScale customers across their on-premises and AWS cloud, but PowerStore block and file storage customers will not, nor will PowerMax customers. PowerFlex is being used as a kind of stopgap. We will have to wait longer for Dell’s mainstream PowerStore software to be available in the public cloud. NetApp still has its Data Fabric lead in this respect. Pure Storage also has its Cloud Block Store. HPE with its GreenLake storage has yet to match these moves.

Cloudy object, PowerProtect and Databricks

APEX Protection Storage for Public Cloud is PowerProtect DDVE (DataDomain Virtual Edition), building on what Dell showed a year ago. This has been ported to four public clouds, supporting AWS, Azure, Google Cloud and Alibaba Cloud. Dell says more than 17 exabytes of data is protected by its software in public clouds to date.

Will there be an APEX Object Storage for public cloud? A Dell spokesperson told us: “Yes. We plan to offer object storage for the public cloud in the future. Stay tuned.” We may then see Dell object storage SW in AWS and Azure built on top of S3 and Blobs respectively.

With the Databricks deal, a Databricks Lakehouse in the public cloud can use Dell ECS object storage, on-premises or in a co-lo, to analyze data in-place, store results and share it with third parties using Databricks’ Delta Sharing capabilities. Dell says it is collaborating with Databricks to jointly engineer additional integrations that will deliver a seamless experience for Dell object storage within the Databricks Lakehouse Platform. We dare say ChatPT-like large language models, like Databricks’ Dolly chatbot, may be considered as part of this.

Dell gets edgy with chatbots, Zero Trust, backup

By
Chris Mellor
-

On the second day of Dell Technologies World, the Texan tech titan has announced projects looking at chatbots, Zero Trust security, edge kit and more.

Dell wants enterprise customers to extend their use of its hardware, software, and services as they adopt AI chatbots, move deeper into zero trust security, and develop their edge site facilities at scale. 

Project Helix

Project Helix is a joint offering with Nvidia that involves validated blueprints to build on-premises generative AI systems. These will use Dell and Nvidia hardware and software with the aim of improving enterprise search, market intelligence, customer services, and other activities with chatbot front ends.

Dell COO and vice chairman Jeff Clarke said: “Project Helix gives enterprises purpose-built AI models to more quickly and securely gain value from the immense amounts of data underused today.” Enterprise customers can, he said, use them to reinvent their industries whilst maintaining data privacy. 

Dell and Nvidia product components include the PowerEdge XE9680 and R760xa servers, Nvidia H100 Tensor Core GPUs, networking and AI Enterprise software, Dell PowerScale arrays and ECS Enterprise Object Storage plus CloudIQ observability. 

AI Enterprise comprises around 100 frameworks, pre-trained models and development tools like the NeMo large language model framework and NeMo Guardrails software for building secure chatbots. 

Project Helix will support a generative AI project lifecycle, we’re told, from infrastructure provisioning, modeling, training, fine-tuning, application development and deployment, to deploying inference and streamlining results. The on-premises presence, Dell claims, reduces inherent risk and helps companies meet regulatory requirements. 

The validated designs developed through Project Helix will be available in July through traditional channels and APEX consumption options.

Project Fort Zero

Rubrik and others have promoted Zero Trust tech for over a year. Fort Zero is Dell’s equivalent product and it is planning to deliver a validated solution within the next 12 months, building on its existing Zero Trust Center of Excellence. 

Herb Kelsey, Federal CTO at Dell, said: “Zero Trust is designed for decentralized environments, but integrating it across hundreds of point products from dozens of vendors is complex – making it out of reach for most organizations. We’re helping global organizations solve today’s security challenges by easing integration and accelerating adoption of Zero Trust.” 

Dell will use a private cloud approach and handle the technology integration and orchestration that it says typically falls to customers and involves several vendors. Dell expects Project Fort Zero to be used to secure datacenters, remote and branch office sites, mobile edge sites with transitory connectivity, road vehicles, planes and more.

A US government assessment team will evaluate the Project Fort Zero end-to-end offering for accreditation and certify compliance against the Department of Defense Zero Trust reference architecture. Customers can then build their own Zero Trust defenses using a validated Dell blueprint.

Upping the edge ante

Dell has announced five edge site offerings with a generic NativeEdge software framework and specific retail site warehouse automation and networking.

COO Clarke claimed: “As our customers look to fuel new workloads and AI at the edge, they are turning to Dell to find simpler and more effective ways to manage and secure their ecosystem of edge technologies and applications.” 

NativeEdge is described as an edge operations software platform helping customers deploy and manage edge site hardware and software at scale, with zero touch deployment. The software is said to cover any enterprise edge use case and includes Dell products plus third-party ones as well in an open systems approach.

Dell NativeEdge

Dell claims a 25-site manufacturing organization could get a 130 percent ROI with a three-year NativeEdge deployment. It says customers across various industries could get similar returns.

The other four offerings are:

  • Dell Validated Design for Retail Edge with inVia Robotics intelligent automation. This design uses software and automation to help retail employees with last-mile picking, packing, shipping and delivery by converting existing warehouse and retail space into micro-fulfillment centers. 
  • Dell Private Wireless with Airspan and Druid is a validated private wireless setup providing wireless connectivity for thousands of remote location edge technologies, like devices and sensors.
  • Enterprise SONiC Distribution by Dell Technologies 4.1 is a scalable, open source-based networking operating system on Dell switches for edge deployments, including User Container Support (UCS) and streaming telemetry.
  • Dell ProDeploy Flex is an edge-focused modular deployment service .

NativeEdge software will be available to customers, OEMs, and partners in 50 countries beginning August 2023. Dell Validated Design for Retail Edge will be available globally in August 2023. Dell Private Wireless with Airspan and Druid will be available globally in June 2023. Enterprise Sonic Distribution by Dell Technologies 4.1 is available globally from today. Dell ProDeploy Flex will be available globally in August 2023. 

Don’t forget services

A Project Success Accelerator (PSX) for Backup service comes after the recent PSX for Cyber Recovery, which implements and operationalizes an isolated cyber recovery vault. Customers can use the PSX for Backup service to help them protect and recover data in the event of disruption. 

It has three components:

  • Ready – includes planning workshops, configuration of a validated backup or vault environment, a success plan, a runbook and outcome-based skills training. 
  • Optimize – adds quarterly assessments, improvement recommendations and assisted restore test simulations. 
  • Operate – adds ongoing operational assistance to meet the solution’s performance objectives. Dell experts monitor and investigate alerts, initiate corrective actions and help with restore tasks at the customer’s direction. 

Availability 

Project Fort Zero’s offering will be available in the next 12 months. Dell Product Success Accelerator for Backup is now available in locations across North America, Europe, and Asia Pacific. Availability of Dell Product Success Accelerator for Cyber Recovery has expanded to locations in Europe, Latin America, and Asia Pacific in addition to North America. 

Spectra Logic and Arcitecta take on NAS market

By
Chris Mellor
-

SpectraLogic and Arcitecta are partnering to enter the high-performance scale-out NAS market with the former’s all-flash BlackPearl storage system and the latter’s MediaFlux software.

BlackPearl is SpectraLogic’s hybrid flash/disk, front-end cache and gateway that can be used to store files as objects on back-end tape devices. It also comes in all-flash BP-X form for faster response to file and object I/O requests. Spectra and Arcitecta have developed their partnership and produced two products: Arcitecta Mediaflux + Spectra BlackPearl NAS, and Arcitecta Mediaflux  + Spectra BlackPearl Object Storage system.

SpectraLogic CTO Matt Starr said: “The combined solutions offer unprecedented high performance, scalability, security, efficiency and significant cost savings … Customers can cost-effectively manage massive data volumes in powerful new ways – a game-changer in the data storage industry.”

Arcitecta partnering with SpectraLogic

Jason Lohrey, Arcitecta CEO and founder, said: “We developed our joint solutions’ NFS and SMB protocols in-house to achieve the level of computing performance organizations require to process enormous amounts of data. Mediaflux plus BlackPearl NAS can support hundreds of billions of files in a single, easily managed global namespace. Multiple storage nodes work in parallel, allowing the system to handle vast I/O request volumes, increase throughput and reduce latency.”

The multiple nodes provide high availability with node failover. A distributed file system, with data stored across multiple nodes, allows data to be accessed from any node in the cluster, increasing data availability.

SpectraLogic, Arcitecta diagram
For the archiving config, replace the NFS and SMB tags with S3 ones

The combined offerings provide automatic tiering with data lifecycle management, easy administration, and on-premises S3 object storage with infinite scalability and easy search and access control. By storing data on-premises, the joint product reduces the latency associated with accessing data stored in the cloud. Infrequently accessed data can be stored on lower-cost media such as tape or object storage – while leaving active data on more expensive, high-performance flash storage. That should lower cost.

The joint products have ransomware resiliency and security with multiple layers of protection, such as intrusion detection and prevention, and real-time monitoring and response capabilities to detect and prevent unauthorized access or activity.

A SpectraLogic spokesperson tells us that the Arcitecta/BlackPearl combo is an “order of magnitude cheaper than Isilon (PowerScale)” and mentioned “about $30,000/PB.”

They added: “For BlackPearl a small scale NAS disk unit will be about $120–150 per terabyte with five years of service, or about $0.004 per gig per month. At scale, 30-plus PB, with a large portion going off to tape archive, [it will be] under $50 a terabyte with five years of service or $0.00071 per gig per month.”

The archiving is claimed to be more affordable than Amazon Glacier. 

Comment

This will be a natural extension for existing Spectra and Arcitecta customers. How it will fare in a straight fight against Dell PowerScale and Qumulo remains to be seen. They have substantial scale-out NAS market presence and their existing customers are unlikely to look at an alternative unless there is a compelling reason – such as a substantially lower price or need for specific Arcitecta and/or Spectra features.

VAST Data also has skin in the game, but is probably aiming at a higher-scale customer. HPE may not be so picky, though. It’s going to be an interesting contest.

Arcitecta Mediaflux + Spectra BlackPearl NAS is available from channel partners as a turnkey offering. Get a NAS backgrounder for the combo here and an archive (object) backgrounder here.

MemVerge, SK hynix say memory is endless now

By
Chris Mellor
-

DRAM virtualizer MemVerge has teamed up with SK hynix on what they’re calling Project Endless Memory to run applications needing more memory than is actually attached to the server by using external memory resources accessed by CXL.

This is the Computer Express Link, the extension of the PCIe bus outside a server’s chassis, based on the PCIe 5.0 standard. CXL v2.0 provides memory pooling with cache coherency between a CXL memory device and several accessing server or other host systems. SK hynix has built CXL-connected DRAM products and MemVerge provides a software bridge to use them in the so-called Endless Memory project.

Charles Fan, CEO and co-founder of MemVerge, said in a statement: “The ability to download RAM automatically and intelligently will lead to improved productivity for users of data-intensive applications.” 

Endless Memory combines MemVerge elastic memory service software and CXL-based Niagara Pooled Memory System hardware from SK hynix enabling host servers to dynamically allocate memory as needed to avoid running out of DRAM. Basically, if an app needs more DRAM than is physically present, the server can use MemVerge’s software to request more to be made available across CXL.

Hoshik Kim, VP/Fellow for Memory Forest R&D at SK hynix, said: ”Testing shows that just 20 percent extra CXL memory from our Niagara Pooled Memory System can improve application performance by 3x compared with the existing swap memory approach.” The swap memory approach is basically paging between physical memory and some kind of storage.

SK hynix's Jonghoon Oh explains Memory Forest
SK hynix‘s Jonghoon Oh explains Memory Forest

Memory Forest is an SK hynix concept. “Mem4EST” means memory for environment, society, and technology. Jonghoon Oh, EVP and CMO at SK hynix, described it like this: “Memory Forest represents how different layers of memory form a computing architecture, just like a forest, where everything from the bottom like the ground, grass to dense trees, fulfill their respective roles to form a huge ecosystem. Forests are generally, from bottom to top, divided into forest floor, understory, canopy layer, and emergent layer. These layers are naturally comparable to those of the computing architecture such as SSD (Solid State Drive), SCM (Storage-Class Memory), high-capacity memory (Capacity Memory), main memory and IPM (In Package Memory).”

Basically, SK hynix’s CXL expander memory hardware is included in this MemVerge initiative.

Project Gismo

Gismo stands for Global IO-free Shared Memory Objects and is a CXL-based multi-server shared memory architecture for data access and collaboration in distributed environments. It enables real-time data sharing, using memory, across multiple servers, eliminating the need for network IO and so reducing data transfer delays.

MemVerge Endless Memory and Gismo diagram
B&F diagram

Fan said: “Network and storage IO are performance bottlenecks for distributed, data-intensive applications. Project Gismo leverages the emerging CXL platform and will effectively remove this bottleneck known as the IO Wall. We envision a memory-centric future where memory is the network and memory is the storage.”

He suggests example use cases could be AI/ML applications, next-generation databases, and financial trading platforms.

One of the early adopters of Project Gismo is Timeplus, which is developing a real-time streaming database. Timeplus CEO Ting Wang said: “With Gismo’s revolutionary CXL-based multi-server shared memory architecture, we have experienced a remarkable improvement in the fault-tolerance of our database system. The speed of streaming query fail-over has been accelerated by an impressive 20x, allowing us to provide unparalleled user experience for the continuation of data streaming processing.”

MemVerge did not provide a timescale for Gismo completion.

Demonstrations of Endless Memory are being presented at the International Supercomputing Conference (ISC) in MemVerge’s booth, D401.

1...133134135...442Page 134 of 442

ABOUT US

Blocks & Files is a storage news, information and analysis site covering storage media, devices from drives through arrays to server-based storage, cloud storage, networking and protocols, data management, suppliers and standards. We publish news, views, opinions and analysis of storage industry events and developments, aiming to keep storage users, developers, consultants, marketeers, salespeople, support staff, analysts and everyone else involved in IT storage up to date with what’s going on.

FOLLOW US

© Situation Publishing, 2018-2024