Cohesity is piloting a DataHawk SaaS service and has set up a Data Security Alliance to buttress its anti-ransomware weapons to customers.
DataHawk combines threat protection via scans for attack indicators and malware, ML-based data classification to identify and detect sensitive or critical data, and its Fort Knox off-site virtually air-gapped secure vault service in the cloud.
Mohit Aron
Mohit Aron, founder and chief technology and product officer, said IT and security teams are “working night and day to stay ahead of bad actors. The key to keeping businesses running is minimizing the impact of a potential breach, bringing data security directly into the operational workflow, securing data at the source, and restoring critical workloads rapidly.“
Cohesity previewed a DataGovern service in October last year. It used AI/ML to automate discovery of sensitive data and detect anomalous access and usage patterns, which could indicate ransomware activity. However DataHawk uses classification technology from BigID to discover and classify large sets of data to help minimize risk and understand the impact of an attack.
Customers can save time chasing false positives and reach resolution faster with more than 200 built-in classifiers and ML-driven algorithms to analyze, tag, categorize, label, and classify data sets, Cohesity claimed. Predefined policies for data privacy and protection regulations like the General Data Protection Regulation (GDPR), Payment Card Industry (PCI), and Health Insurance Portability and Accountability Act (HIPAA) help customers identify and prioritize data sets.
Data Security Alliance
There are nine Data Security Alliance members along with Cohesity: BigID, Cisco, CrowdStrike, Mandiant, Palo Alto Networks, PwC UK, Securonix, Splunk, and Tenable. They can all integrate with Cohesity at an API level, and some are already doing that.
Tenable.io powers Cohesity’s CyberScan to assess data backup environments to help ensure that a recovery situation does not introduce vulnerable conditions into production. BigID technology is being used in DataHawk.
An integration with Palo Alto Networks’ Cortex XSOAR is designed to provide faster ransomware response. Pamela Cyr, VP technical partnerships at Palo Alto Networks, said there is a “bi-directional flow of data and commands to rapidly detect and respond to ransomware. Cohesity confirms infection, and the Cortex XSOAR automation platform manages the enrichment and initiates a safe restore of business-critical data.”
Sanjay Poonen
Cohesity CEO Sanjay Poonen said: “Today’s non-stop and increasingly sophisticated cyber threats require an all-hands-on-deck approach. It’s not the responsibility of one vendor to solve all cybersecurity challenges… We are partnering with these industry heavyweights so they can leverage our platform, the Cohesity Data Cloud, to help customers easily integrate data security and resilience into their overall security strategy.”
Cohesity formed a Security Advisory Council in September, with Mandiant CEO Kevin Mandia on its board of directors. Other members include:
Alex Stamos, director of Stanford Internet Observatory, partner at the Krebs Stamos Group, and former CSO at Facebook and Yahoo!
Jason Chan, former VP of Information Security at Netflix.
Marianne Bailey, partner at Guidehouse, and former senior cybersecurity executive at the NSA.
Laura Barrowman, Cohesity board advisor and CIO at Credit Suisse.
Sheila Jordan, Cohesity board advisor and chief digital technology officer at Honeywell, former CIO at Symantec.
It has now appointed Kelly Bissell, EVP of Microsoft Security Services, to this council too.
Cohesity has three security-focused initiatives: DataHawk SaaS; the Security Advisor Council; and the Data Security Alliance. It sees no need to offer ransomware guarantees given its approach.
We see Cohesity setting up a web of enterprise security connections, integrations, and influencers for its data protection services. This is a different marketing and product direction from other data protection suppliers.
DataHawk is currently available for early access preview from the Cohesity Data Cloud platform with general availability planned in the coming months.
WEKA, the developer of high-performance, parallel filesystem software, has taken in a second funding round this year, saying it’s needed to fuel sales growth and global expansion.
Liran Zvibel
The company was started in Israel in 2013 to develop its scale-out Matrix file system. This runs on commodity x86 hardware and in AWS, Azure, GCP, and Oracle clouds. It has notched up SPC and STAC benchmark wins that set it apart from competing fast-access parallel filesystem products such as Spectrum Scale and Lustre, and scale-out file systems such as Dell EMC’s Isilon/PowerScale and Qumulo. Matrix is used in high-performance computing, analytics, AI and machine learning, life sciences, finance, and engineering.
Liran Zvibel, co-founder and CEO, said of the funding round: “This funding round represents a tremendous milestone … and we are grateful to our new and existing investors for their support. The new capital will allow us to invest in our product and go to market, maintain hypergrowth, and gives us the runway we need to become a profitable, independent company.”
It raised $135 million, having already raked in $73 million in January, taking its total raised to $275 million and lifting its valuation to $750 million.
Funding history
2014 – $10 million A-round
2016 – $22.5 million B-round
2019 – $31.7 million C-round
2022 – $73 million C2-round
2022 – $135 million D-round
In September, privately-owned WEKA claimed it was surpassing internal business projections with a record third quarter that exceeded its entire fiscal 2021, boasting:
255 percent net dollar retention (NDR) while maintaining a zero-churn business
250 percent attainment against its Q3 financial plan
635 percent total contract value (TCV) growth
232 percent annualized run rate (ARR) growth
Significant expansion into the Asia Pacific region and beyond
The Matrix filesystem has evolved into the WEKA Data Platform and the cash will be used to expand its features and cloud integrations, support new use cases, and accelerate further development. WEKA said this funding will enable the company to reach profitability, fuel significant global expansion, and rapidly scale its cloud, sales and marketing operations and human resources teams.
The global economic situation is threatening to turn into a full-scale recession. Several tech startups have run out of money because fundraising in this climate is challenging. Competitor Qumulo laid off around 80 people in June and Pavilion Data closed down in October.
Yet stronger players have scored big funding wins, such as analytics supplier Imply with $100 million in May, SaaS backup vendor HYCU’s $53 million in June, and Lightbits, Scale Computing, and Pliops, which raised $100 million in August. Now WEKA has joined their ranks.
The latest round was led by Generation Investment Management. New and existing investors also contributed including 10D, Atreides Management, Celesta Capital, Gemini Ventures, Hewlett Packard Enterprise, Hitachi Ventures, Key1 Capital, Lumir Ventures, Micron Ventures, Mirae Asset Capital, MoreTech Ventures, Norwest Venture Partners, NVIDIA, Qualcomm Ventures, and Samsung Catalyst Fund.
We asked WEKA some questions about the latest round.
Blocks & Files: Having raised $74 million in January, why does WEKA need another $135 million now?
WEKA: As we shared in our recent Q3 momentum announcement in September, WEKA is seeing significant success and customer demand around GPU-accelerated projects on-premises and in the cloud.
We are investing heavily in enhancing the WEKA platform, expanding its applicable use cases, and expanding our global footprint and go-to-market. The new capital will enable us to continue doing this at scale. It will also allow us to reach profitability and become an independent company, which we endeavor to achieve in the next 12-18 months.
Blocks & Files: Is WEKA in danger of becoming over-capitalized, restricting its IPO opportunity and preventing an acquisition?
WEKA: No, WEKA is not approaching over-capitalization. We have taken great care to maintain healthy funding levels to fuel growth without over-leveraging the company. Given current market conditions, we will not be rushing to pursue an IPO anytime soon.
We have a large ecosystem of go-to-market partners and enterprise customers. We can better serve them as a profitable, independent company.
Blocks & Files: Will raising more money improve WEKA’s standing with customers, helping it to grow sales faster?
WEKA: Absolutely. The new capital will enable us to invest ahead of the growth and provide the runway we need to become profitable. This is an important consideration for our major enterprise customers, several of whom are Fortune 100 companies that rely on WEKA to support their cloud and AI initiatives. They want to ensure they are partnering with a company that will be around for the long haul, so their future tech stack is secure.
Blocks & Files: Is WEKA planning to make an acquisition with the new cash?
WEKA: We have no immediate plan to make an acquisition.
CopprLink – The PCI-SIG announced in November 2023 that the new naming scheme for PCIe Internal and External Cables will be CopprLink. The PCIe 5.0 and PCIe 6.0 Internal and External Cable Specifications are currently in development and are targeted for release in 2024.
ExaGrid has deepened its integration with Veritas and Veeam backup software to make backups complete in less time and take up less space.
Update: Bill Andrews notes on deduplicaion and competition added below, 18 November 2022.
Bill Andrews
It makes scale-out, adaptive (see below) deduplicating disk backup appliances with restores from a native format ingest buffer, aka landing zone, and tiers data to send to the public cloud for longer-term retention and disaster recovery. It has more than 3,600 customers and competes with Dell’s PowerProtect (Data Domain), Quantum’s DXi, and HPE StoreOnce products. v6.2 of its software adds closer intergration with Veritas and Veeam, improved ransomware detection, and extended public cloud support.
CEO Bill Andrews said the company “continues to focus on improving backup and restore performance, scalability, cost savings, and security.”
v6.2 adds:
Veritas NetBackup Single Target Storage Pool for Veritas NetBackup: ExaGrid presents a single virtualized target pool of storage at each site with automated job management sending backup data to available ExaGrid appliances for front-end performance. ExaGrid automatically load balances all data across all non-network facing repositories for full storage utilization and deduplicates data across all repositories to save space.
Veritas IT Analytics integration: Enables reporting on Exagrid appliance statistics to Veritas’s unified insights for backup, storage, and virtual infrastructure. ExaGrid has a total of nine Veritas NetBackup integrations, including support of NetBackup Accelerator, OST Integration, integration with NetBackup media server hardware, AIR (Auto Image Replication), GRT (granular level restore), optimized deduplication, and instant recovery.
ExaGrid Cloud Tier: Improved performance and scalability of Cloud Tier for AWS and added support for Azure. Data sent to AWS and Azure is deduplicated and uses about a 50th of the bandwidth versus undeduplicated data. ExaGrid encrypts the data in transit, and the cloud providers encrypt the data at rest.
Faster deduplication: ExaGrid has accelerated Adaptive Deduplication performance, increasing the Repository Tier’s overall performance, scalability and recovery points, and doing the same for disaster recovery copies in the public cloud.
Faster Veeam backup: ExaGrid has greatly improved the performance of Veeam synthetic fulls (Veeam v12) by adding support for Veeam’s Fast Clone technology, meaning synthetics fulls are at least 30x faster than previous versions.
Improved Security: ExaGrid has added two early warning features. An alert on large delete is generated as an early detection warning if an intruder is trying to delete a large amount of recently ingested data (above a set operational threshold) on the ExaGrid front-end Landing Zone. An alert on data deduplication change is generated when the deduplication ratio changes below a set threshold signaling that a large amount of encrypted data is being sent to the ExaGrid front-end Landing Zone or that the data on the Landing Zone is being encrypted.
The company believes in the better mousetrap theory. Purpose-built backup appliances are just backup software targets which need power, cooling, and datacenter space. They have to avoid limiting backup windows with slow ingest rates and support fast restores. If an ExaGrid box can hold more data in less disk and rack space for less power, and restore it faster, then it’s a pretty easy job to replace an existing system that’s reaching a limit point.
Bill Andrews notes
Inline does the dedupe on the way to the disk and this slows down backups. Also, the data is all deduplicated and has to be rehydrated for each request so the restores are slow as well.
Post process, which I don’t think anyone does anymore, waits until all the backups fully land (i.e. are complete) and then does the deduplication. The good part of this is fast backups and fast restores but the offsite DR RPO (recovery point time is not good).
We did this in the early years before we switched to Adaptive deduplication and Sepaton and Diligent (IBM Protectier) both did this but both were discontinued years ago (no longer in the market).
What ExaGrid does is Adaptive Deduplication. As the data hits the Landing Zone, we immediately start deduplicating the data into the repository tier and replicating to the DR site. So we are deduplicating and replicating all night as the backups are coming into us and therefore we have a much better RPO. In fact, we are done with backups and replication faster than any of our competitors with this approach.
Also, remember Dell, HPE, etc. have a front end controller and then disk shelves, we have compute and memory in every appliance which allows us to apply a lot more compute. ExaGrid is the fastest for backups, as fast as disk with non-deduplicated data for restores, and have the best RPO. We have no problem doing side by side tests to prove this out. We have tested 3 times against Pure SSD and won on performance in all 3 as they are not optimized for backup.
Lastly, we just don’t see Quantum in the marketplace and have not for years. I believe they sell but don’t develop that product any more. In fact, when was the last time you saw new features for HPE StoreOnce or Dell Data Domain? [Federated management in June this year.]
We replace in order of volume the following:
Primary storage disk from Dell, HPE, NTAP, Cisco, IBM, Hitachi, etc. – 50 percent of our business (we see more disk behind backup apps than anything else)
Dell Data Domain – 30 percent of our business
HPE StoreOnce -10 percent of our business
Veritas storage appliances and everything else – 10 percent of our business
Primary storage disk dominates followed by Dell Data Domain who is the king of deduplication appliances.
Nvidia has given VAST Data an unexpected lift by publishing a report showing customers using its Bluefield server-offloading DPU in large enough numbers need fewer servers. VAST’s Ceres storage box has a Bluefield front end and fits right in with Bluefield server adopters.
The Nvidia DPU Power Efficiency report says that reducing the number of servers in a datacenter can help the operator use less electricity to power and cool their server estate.
The report says: “In typical servers, virtualization, networking, storage, security, management, and provisioning are all handled by VMs, containers, or agents running on the server’s main CPUs. Not only does this consume up to 30 percent of the processor cycles, but a CPU is not efficient in running these types of infrastructure workloads.”
A DPU is a domain-specific accelerator, like a GPU, that “frees up server CPU cores to run the types of applications they do best.” The Bluefield DPU (Data Processing Unit) uses specialized hardware to run infrastructure-specific operations, such as data encryption/decryption, key management, etc. to do with storage and networking, more efficiently than a server’s general purpose x86 CPU. Also “DPU cores can perform SDN, telemetry, deep packet inspection, or other networking tasks more efficiently than the server CPU.”
The report has various spreadsheet type cost examples supporting its point, with a final total cost of ownership (TCO) calculation based on having a 10,000 server datacenter offload IPsec encryption/decryption to BlueField DPUs. The three-year TCO for these 10,000 servers without BlueField are $148 million. Adding BlueField to the mix enables the operator to get rid of 1,800 servers and reduce the three-year TCO to $121.7 million – a $26.6 million, 17.8 percent saving, as a table in the report shows:
The report concludes that: “In a world facing rising energy costs and increasing demand for green IT infrastructure, the use of DPUs will become increasingly popular to reduce TCO by decreasing both CapEx and OpEx in the datacenter.”
Nvidia obviously has a vested interest here: it sells BlueField DPUs. But then so does Intel, the x86 server CPU king, calling them IPUs (Infrastructure Processing Units), as does x86 pretender AMD, with its Pensando acquisition.
VAST Data sells its scale-out all-flash storage in Ceres-branded enclosures. These are 1RU in size and hold up to 22x E1.L 9.5mm Solidigm D5-P5316 drives, using 144-layer 3D NAND and a PCIe 4.0 interface, in 15.36 or 30.72TB capacities. VAST says Ceres makes it “possible to build highly available flash storage enclosures without the need for power-hungry x86 processors.” That’s right on message as far as Nvidia is concerned.
We envisage VAST taking this Nvidia report and working up its own BlueField-enhanced storage TCO saving spreadsheets.
Filers that use x86 controllers are at a disadvantage here. But flash JBODs, such as Kioxia’s KumoScale, can more easily add front-end BlueField access and play in the same general server-offload arena as VAST Data.
France-based Kalrayhas signed a major contract with an unnamed NASDAQ-listed tech organization to develop and supply acceleration cards and support the customer’s development teams for their next-generation product.
This card can deliver two million IOPS, with 12GB/sec bandwidth for both RoCE and NVMe/TCP, Kalray said. It adds latency of between 30μs for RoCE and 50μs for NVMe/TCP storage IO accesses. Kalray’s competition includes AMD-Xilinx (Pensando), Nvidia with BlueField, Fungible, Marvell, Intel, and others.
Eric Baissus
Eric Baissus, Kalray CEO, said in a statement: ”The signing of this contract is a major milestone for Kalray. It is the result of nearly 18 months of in-depth evaluation by our customer of our technology, which has demonstrated its relevance compared to the most advanced alternatives in the industry, the quality of our teams … and of our ability, as a strategic partner company to offer the guarantees required to sign a contract of this magnitude with a major international player.”
Internal testing showed that Kalray’s Massively Parallel Processing Array (MPPA) DPU was three to five times better than the competition, representing potential savings for the customer of several hundred million dollars over the life of the project. The customer is said to be a major high-tech player and generated several tens of billions of dollars in revenue in 2021, when it had a market capitalization of over $100 billion. Kalray says the contract has a commercial potential of several tens of millions of euros per year.
Kalray K200 storage acceleration card
The customer has already committed $1 million to Kalray to cover the various phases of its next-generation product, development of which is expected to be completed in the second half of 2023.
The second step will be for the customer to launch pre-series production of its new product. The third step will be volume production.
Kalray MPPA diagram
In July we learned that Kalray and Sanmina’s Viking Enterprise BU were jointly developing a FlashBox array, a JBOF disaggregated storage system using Viking’s VDS2249R chassis and Kalray’s MPPA-based K200-LP storage acceleration card. NASDAQ-listed Sanmina has a $3.91 billion market capitalization and its 2021 revenues were $6.76 billion; it’s not Kalray’s latest customer.
Kalray was founded in 2008 as a spin-off of CEA and two of its labs, Leti and List. CEA is France’s Alternative Energies and Atomic Energy Commission (Commissariat à l’énergie atomique). Kalray investors include Alliance Venture (Renault-Nissan-Mitsubishi), Safran, NXP Semiconductors, CEA, and Bpifrance.
ATTO Technology and Nyriad have teamed to integrate ATTO FastFrame 3 SmartNICs and ThunderLink Thunderbolt adapters with the Nyriad UltraIO storage system, powered by a GPU-CPU combination. The combination quadruples write throughput, enabling the scrubbing and editing of raw 8K and above video footage inline without the need to transcode, create proxies or copy files across the network to local editing stations. Nyriad’s UltraIO system is claimed to withstand up to 20 drives failing simultaneously while maintaining 95 percent of its maximum throughput. Dual-port ATTO interfaces can survive the loss of entire switch fabrics without any outages, it said.
…
AWS SaaS data protector Clumio has announced capabilities it claimed allows customers to protect their AWS data at any scale and recover from disruptions instantly. Highlights include:
Continuous [virtual] air-gapped protection for any volume of data: Customers can now protect up to 30 billion objects per Amazon S3 bucket, totaling exabytes of data, with a granular RPO of 15 minutes.
Instant recovery for Amazon S3: Clumio has achieved zero RTO (recovery time objective) for Amazon S3. This feature is now available for early access.
Intelligent data protection recommendations to optimize cost efficiency: Customers can get a full view of their data environment on AWS, consolidate backups across multiple services under a unified [virtual] air gap, and receive recommendations on how to optimize data protection costs.
…
IBM has launched a carbon offsetting program. For every Flash System sold with its business partners in Q4, IBM will plant 100 trees as part of its Flash4Good initiative. IBM’s product carbon footprint (PCF) data is now available for the IBM FlashSystem 5200, IBM FlashSystem 7300, and IBM FlashSystem 9500.
…
Intel has quietly launched a 2.5-inch Optane 3D XPoint datacenter SSD, the DC P5810X, with a slightly shorter latency and slightly worse IOPS and throughput performance than its existing P5801X drive – the PCIe 4 x 4 NVMe interface drive. A table shows the similarities and differences:
No more Optane chips are being made, Intel is working through its inventory, and the product line is being closed down.
…
MSP remote monitor and data protector N-able reported revenues of $93.5 million in the third 2022 quarter, ended September 30, up 6 percent year over year. Subscription revenues were $91.2 million, compared to $861 million a year ago. Net profit of $300,000 was down from the year-ago $1.9 million. At quarter end, total cash and cash equivalents were $87.7 million and total debt was $337.5 million. N-able president and CEO John Pagliuca said: “We are proud of our third quarter results, which came in above the high end of our outlook for both revenue and profit.”
…
Scale-out filesystem supplier Qumulo has announced a “Simply Secure” initiative with a suite of security features that continue to harden over time, all-inclusive with their Qumulo subscription, without additional cost for future releases. Its most recent software release added five new layers to storage security for greater data protection including:
Multi-tenancy VLAN Isolation: Organizations can now use virtual local area networks (VLANs) to isolate administrative interfaces from their filesystem clients, such that the general network population cannot reach the interfaces.
Single sign-on & Access Tokens: Cluster administrators can now eliminate the need for sensitive user passwords when logging into the Qumulo administrator UI or API since user credentials are prime targets for theft by cyber attackers.
NFSv4.1 Kerberos Authentication & Encryption: All data is encrypted before transmitting across networks.
Federal Information Processing Standards (FIPS) 140-2 certification of Qumulo encryption: Customers can maintain compliance and independently verify that Qumulo’s data-at-rest encryption meets the standards set by the National Institute of Standards and Technology (NIST).
OpenMetrics API provides telemetry data to third-party monitoring and alerting systems to help proactively detect and respond to anomalies such as an attack in progress.
…
Samsung Semiconductor US has said that Oracle “will be collaborating with AMD and Samsung on a project that will utilize AMD’s Epyc4 CPUs and Samsung’s Memory Semantic SSD to provide high performance for IMDB and Exadata machines.” The Memory Semantic SSD is a hybrid DRAM and NAND drive with a PCIe 5.0 interface.
…
Hybrid cloud manager and monitor Virtana is integrating its flagship Infrastructure Performance Management (IPM) solution, VirtualWisdom, into Virtana Multi-Coud Insights Platform. Virtana MCIP is a unified Software-as-a-Service (SaaS) Platform that can consolidate several products into a single tool, helping customers overcome tool sprawl challenges. It has four main products:
Infrastructure Performance Management: Ensures maximum performance of the entire infrastructure across multi-cloud environments.
Capacity Planning: Forecasts capacity demands, projects how capacity matches to future business objectives, and leverages actionable insights on potential issues.
Disk drive maker Seagate, hit by a market slowdown, is looking to sell then lease back property in California and Colorado.
The company’s revenues were down 35 percent on an annual basis to $2.04 billion in its latest quarter ended September 30, with a $29 million profit, 94 percent lower than the year-ago $526 million. It said it would make fewer disk drives, cut annual capital expenditure plans, and layoff staff to deliver cost savings of up to $95 million.
Dave Mosley
CEO Dave Mosley told analysts during the results earnings call: “In addition to adjusting our production output, to drive supply discipline and pricing stability, we are implementing a restructuring plan to sustainably lower costs, including a reduction in our global workforce.”
The San Francisco Business Times reports that Seagate wishes to sell its 31-acre, 575,000 sq ft research facility at Kato Road in Fremont, California, for $300 million and then lease the buildings for a seven-year term.
Seagate bought the property for $90.3 million in 2013.
It is aiming to operate a similar sale-and-lease-back deal for its 40.5-acre, 400,000 sq ft facility at 389 Disc Drive, Longmont, Colorado. It runs a design center at the Longmont facility, developing a Lyve Reference Architecture, for example, and bought the facility in 1999.
As Mosley said, Seagate’s restructuring plan also includes reducing its headcount by 8 percent, around 3,000 jobs.
Workers at Seagate’s plant in Derry, Northern Ireland, are concerned that they will be targeted in the layoff exercise. The plant makes disk drive read/write heads. If fewer disk drives are being made, fewer read/wrote heads will be needed.
Mosley said in the earnings call: “We currently expect customer inventory drawdowns will remain a factor through at least the December quarter. We reacted quickly to adjust our production levels to the current demand environment and our gross margin performance reflects the resulting factory underutilization costs that increased markedly through the month of September.
“We continued to respond to the changing market conditions and further reduced production output across all product lines with the exception of our 20-plus terabyte products, where demand has held firm and pricing relatively stable.”
As reported in the Belfast Telegraph, Foyle constituency councillor Shaun Harkin said: “Seagate has made tremendous profits in recent years. This is reflected in its shareholder payouts and big pay increases for top management. It still remains profitable but has seen a slowdown. However, this slowdown was expected and the company’s own financial projections are very upbeat. We need to do everything we can to ensure workers in Derry and the North West are protected.”
Several hundred Seagate staff in Derry were disappointed with a recent pay award. According to Derry Now, Harkin said: “Workers were also disappointed at their pay award this year, which led to 400 members of staff signing a petition calling for a decent pay increase as we face into a ‘cost of living’ crisis.”
AMD has released its Genoa 4th generation EPYC 9004 series CPUs with 96 cores and support for PCIe 5.0. and CXL 1.1 memory expansion. …
Astera Labs said its Leo Memory Connectivity Platform provides memory expansion when combined with AMD’s Genoa CPUs. Astera and AMD will demo CXL memory expansion at SuperCompute ‘22 in Dallas next week.
…
Dell announced the next generation of PowerEdge servers with 4th Generation AMD EPYC processors. They are available in one- and two-socket configurations, with support for as much as 50% more processor cores compared to the previous generation.
…
Marvell has announced a new Compute Express Link (CXL) Development Platform for cloud data center operators and server OEMs. It supports memory expansion and pooling and pairs Marvell’s CXL technology with AMD’s Genoa and Intel’s coming Sapphire Rapids CPU.
…
Big memory software supplier MemVerge has developed software-defined Compute Express Link (CXL) memory management products that run on 4th Gen AMD EPYC processors featuring support for CXL 1.1+ specifications.
…
Micron announced availability of DDR5 DRAM memory and DDR5 SRAM for the datacenter that is validated for the new AMD EPYC 9004 Series processors. Its DDR5 SDRAM delivers a 1.5 times increase in data rates at 4800 MTps when compared to DDR4 3200, and enables features like on-die ECC, Error Check and Scrub (ECS) and fault bounding for increased reliability in the data center.
…
SK hynix announced its DRAM – 1anm 16Gb DDR5 and 1anm 24Gb DDR5 – and CXL products have been validated with AMD’s 4th Gen Epyc processors. The DRAM supports 4800Mbps. SK hynix also provides a CXL memory device with 96GB capacity composed of 24Gbit DDR5 DRAMs based on 1anm.
Other storage news
Real-time data platform supplier Aerospike announced early availability of Aerospike Cloud running on AWS. The new Database-as-a-Service offering gives developers and development teams the capability of the multi-model Aerospike Database 6 delivered as an on-demand, pay-as-you-go, fully elastic service.
…
Craig Nelson
Open-source data orchestrator Alluxio has appointed Craig Nelson as SVP of Global Sales; along with the company’s first dedicated SVP, Business Development and Marketing role led by John Mracek, who formerly headed sales. Previously, Craig was CRO for FirstWave, a leader in network monitoring and cybersecurity solutions HQ in Sydney, Australia. He was the CEO of Opmantek Software, an open source software company which was acquired by FirstWave in Jan 2022. Craig joined Opmantek Software from Ixia, which was acquired by Keysight Systems in 2017 for $1.6 billion. At Ixia, Craig was VP of Worldwide Enterprise Sales.
…
Cloud and backup storage supplier Backblaze has promoted Dr Brian Beach to CTO. Brian Wilson, who has served as CTO since 2007, will remain an employee at Backblaze and will become Inveterate Software Engineer, as well as serve in an advisory CTO capacity until the end of the year.
…
ChaosSearch today announced new platform integrations designed to support growing demand for less complex, less expensive cloud data services. The combination of Cribl Stream and the ChaosSearch Cloud Data Platform will optimize customers’ SIEM, APM, and observability systems, improving incident investigation, threat hunting, and more. Datadog users can use ChaosSearch to get a centralized, 360-degree view of log data and enhance the benefits of Datadog for metrics and tracing, while reducing costs.
…
Supercomputer, HPC and enterprise storage supplier DDN is growing its services product set with embedded onsite professionals and managed services to support in-house technology teams. Several public sector organizations, including federal and state government as well as private businesses, are using DDN’s technologists as Storage Account Managers, Technical Account Managers and Storage Analysts. DDN technologists with secret level clearances and above can support national security organizations.
…
Struggling survivor and archive storage supplier Falconstor has reported its second consecutive growth quarter, with Q3 revenues of $2.4 million in the prior quarter. It eked out a $220K profit. Todd Brooks, FalconStor CEO, said: “The hybrid cloud reseller relationship we formed with IBM in the second quarter of 2022 continued to expand during the quarter with new customers taking advantage of our joint offerings and existing customers expanding their monthly usage.”
…
SaaS backup supplier Keepit has won auto distributor Porsche Holding Salzburg as a customer, because it had European-based datacenters. Keepit is a vendor-neutral SaaS data protection cloud with a blockchain-verified offering. It follows the 3-2-1 principle of backup, where 2 copies of data are stored at separate locations, and one copy is backed up in an off-site, private cloud.
…
MASV (massive.io), declaring it provides the fastest large file transfer solution for media professionals, has integrated its SW with Object Matrix. It says this MASV integration is the fastest way to upload media assets to Object Matrix MatrixStore. Users don’t require account-level contributor access to a MatrixStore account. MASV supports transfer workflows with a variety of cloud providers, including AWS, Frame.io, Google Cloud, Dropbox, Microsoft Azure, Sharepoint, Microsoft OneDrive, Backblaze, Box, DigitalOcean, MinIO, Postlab Drive, Wasabi Cloud Storage, Docker, iconik, Adobe Premiere Pro Panel, Slack, Reach Engine and more.
…
Micron is shipping qualification samples of its 1β (1-beta) DRAM technology to select smartphone manufacturers and chipset partners and has achieved mass production readiness. It is using it in low-power double data rate 5X (LPDDR5X) mobile memory, delivering top speed grades of 8.5 gigabits (Gb) per second.
…
Cloud file system supplier Nasuni announced strong customer growth in the energy industry with data under management increasing by more than 165 percent since the end of Q1 2021. This was driven by a greater amount of cloud migrations, strong demand for file access and sharing capabilities from remote locations and the persistent need for business resiliency from threats such as ransomware.
…
Seagate said Robert (Bob) Bruggeworth has been appointed to its Board. Bruggeworth has over 20 years experience as CEO of two public companies, RF Micro Devices and currently Qorvo. Prior to these roles, he spent 16 years in various senior leadership positions at AMP, now part of TE Connectivity, residing in Asia and the U.S.
…
Cloud database accelerator Silk has appointed Tom Murphy as its CMO. Murphy has worked at Turbonomic, a resource-simulation software company acquired by IBM. Before that, he held the CMO role at Bradford Networks, Relicore, and Bit9.
…
Speedb which supplies the Speedb data engine, a drop-in replacement for RocksDB, is going open source. Open sourced by Facebook as a fork of Google’s LevelDB in 2012, RocksDB is the de facto standard Embedded key value store (KVS). As RocksDB adoption expands to cover a broader range of use cases and deployment scenarios, the Speedb community fills a gap in the KVS space by providing developers with innovation, support and services to overcome emerging challenges. The Speedb enterprise edition enables users to routinely generate up to double throughput, with half the latency, and double the capacity as compared to the same deployment with the RocksDB storage engine.
Casey George
…
Data integrator and manager Talend has appointed Casey George as its CRO. Casey was most recently SVP and GM at Verint, a global customer engagement supplier, and spent over 20 years at IBM in many executive roles.
…
ThinkParQ, which supplies the parallel file system BeeGFS SW, announced a new hierarchical index tool, BeeGFS Hive Index. It’s a hierarchical index that stores file system metadata to enable users to run queries and searches on a file system with millions to billions of files and receive results in seconds without affecting the performance and efficiency of the file system itself. It initially scans the file system and then incrementally updates the index by reacting to file system modification events it receives from the BeeGFS metadata modification event logger.
…
Data pipeline builder Upsolver announced GA of SQLake at a new low price of $99 per TB ingested with no charge for transformation processing and no minimum commitment required. The new service provides a SQL-based, self-orchestrating data pipeline platform that ingests and combines real-time events with batch data sources for up-to-the-minute analytics. It says companies can achieve a quantum leap in data freshness for use cases like ML model training, anomaly detection and real-time BI and data science. It also allows them to make data easily available to any SQL user – not just data engineers but data scientists, analysts, product managers and other data consumers.
…
Veeam announced that UK supermarket chain Asda has selected Veeam Availability Suite backup and restoration. Following its divestiture from Walmart in early 2021, Asda had to take responsibility for protecting its more than 2,000 hosts and 3,000 VMware virtual machines. Veeam software currently backs up and replicates data between the company’s primary and secondary data centres, its head office and one of its depots.
…
Weebit Nano announced the first silicon wafers integrating Weebit’s embedded Resistive Random-Access Memory (ReRAM or RRAM) module have been delivered to Weebit from SkyWater’s U.S. production fab. This is the first time silicon wafers of Weebit ReRAM have been received from a production fab, a major milestone toward commercialization of the technology at SkyWater.
The good ship Backblaze, loaded with cloud and backup storage, grew revenues 27 percent year on year in the third 2023 quarter, beating expectations, but more than doubled its loss, as customer demand stayed strong in the current unfavorable economic climate.
Backblaze builds its own storage enclosures, called Pods, and over the years has expanded from storing customers’ backups to providing general cloud storage in its datacenters. It also provides a popular service to disk drive buyers, its quarterly disk drive reliability statistics, detailing per-model failure rates as drives age.
A couple of weeks ago it said its third quarter revenues would likely be larger than the second quarter guidance of $21.4 million to $21.8 million. And so it has turned out.
CEO and co-founder Glen Budman’s results statement said: “As highlighted in our October 24, 2022 preliminary results, we had a strong Q3. We had 48 percent revenue growth in B2 Cloud Storage, and total company revenue grew 27 percent. We also made significant progress in our channel program, including the launch of our channel partner portal and signing of several new national resellers. We are pleased to continue to scale in an uneven macro-economic environment.”
Revenues in the quarter ended September 30 were $22.1 million, with a loss of $12.8 million, compared to the year-ago loss of $5.99 million. However the loss rate has been stable over the last three quarters, hovering between $11.6 million and $12.8 million.
Backblaze had $27.5 million in cash and restricted cash at the end of the quarter. Balance sheet cash, investments and restricted cash is much higher: $80 million. The adjusted gross margin was 76 percent.
The computer backup business segment pulled in $13.1 million, up 17 percent year-on-year, while cloud storage grew much faster, increasing 48 percent to $8.8 million. A chart shows the cloud storage trend of catching up with backup storage:
Cloud storage now represents 40 percent of Backblaze revenues. Third quarter annual recurring revenue (ARR) was $88.0 million, up 24 percent. The net revenue retention rate (NRR) was 114 percent, as opposed to 110 percent a year ago.
William Blair analyst Jason Ader commented: “Though B2 is positioned well as a low-cost cloud storage alternative (priced at 1/5 the cost of AWS’s S3), Backblaze’s high exposure to SMBs means it could see demand headwinds in an increasingly tougher macro environment. That said, we continue to like the Backblaze story for its large opportunity in providing cloud storage to the underserved midmarket, with the company intently focused on driving greater selling leverage through channel partners.”
The outlook for the fourth quarter is for revenues between $22.5 million and $22.9 million, $22.7 million at the mid-point, which would represent a 21.4 percent on the year-ago $18.7 million, and make for an $84.99 million revenue total for the full year. That would be a 26 percent year-on-year increase. It’s all steady steady growth with no drama and no upsets. Somebody could snap this business up; it’s a great little earner.
Panasas is known for shipping PanFS ActiveStor storage systems for high-performance computing shops. Now it’s branching out by partnering with Atempo and its Miria data-moving product.
Panasas started getting a deeper involvement with machine learning this summer, and has now decided to pursue the data movement and analytics market. This is a broadening move to bring in AI and machine learning and also high performance data analytics to its core HPC storage activity.
Panasas’ VP of Marketing and Product at Panasas, Jeff Whitaker, put out a statement saying: “These new offerings are key additions to our PanFS software suite, and they underscore our continued transformation into a multiplatform, software-first company.”
Atempo’s Miria Data Management software has modules for analytics, migration, backup, archiving and synchronization, and can support billions of files. It happens to support Lustre and S3. Panasas has created two new pieces of software from this: PanView Analytics by Atempo and PanMove Advanced by Atempo.
Tom Shea, Panasas CEO and president, provided more context: “Volumes of unstructured data are growing rapidly today, driven by the increasing convergence of traditional HPC and AI workloads. … “The PanView and PanMove suite continues the Panasas new product trajectory and gives our customers the control they need over their data – across platforms and in the cloud.”
PanMove extends Panasas’ reach beyond a cluster. It enables users to move large volumes of data between different ActiveStor systems, locally and across geographic distances. Data can also be moved between PanFS storage environments and S3 object stores, whether in public or private clouds, or on-premises systems.
Panasas customers’ end-users can extend PanMove’s protection, synchronization, and replication functionality even further by pairing it with Atempo’s Miria products. They can then support backup and archive targets such as tape and tape robotics as well as Azure and Google cloud object storage.
PanView is a data management and analytics tool that provides end-users with a simple, consolidated global view of their data across Panasas storage systems. It also delivers comprehensive storage activity reporting from a dashboard.
Atempo has been steadily building out partnership activities, making deals with Quantum and Qumulo. Now it has a third US-based storage partner.
Panasas will be demonstrating its software products at booth #713 at SuperComputing ‘22 in Dallas, Texas on November 14th – 17th, 2022. The PanView and PanMove products will be available to customers in the first 2023 quarter.
Hammerspace Global Data Environment (GDE) users can access files from servers, across datacenters, and global Internet links, and Hammerspace has upped the speed of all three stages, with parallel NFS, RDMA, and metadata and data path separation.
GDE 5.0 adds higher performance at server level, with near maximum IO subsystem speed, across interconnects in data centers, saturating Ethernet and InfiniBand links, and onwards to to the cloud, saturating links at this level too. The server-local IO acceleration applies to bare metal, virtual machines and containerized applications. It says this allows organizations to take full advantage of the performance capabilities of any server, storage system and network anywhere in the world.
David Flynn
Hammerspace CEO and founder David Flynn said in a supplied quote: “Technology typically follows a continuum of incremental advancements over previous generations. But every once in a while, a quantum leap forward is taken with innovation that changes paradigms. … Another paradigm shift is upon us to create high-performance global data architectures incorporating instruments and sensors, edge sites, data centers, and diverse cloud regions.”
The company claims GDE 5.0 changes previously held notions of how unstructured data architectures can work, delivering the performance needed to free workloads from data silos, eliminate copy proliferation, and provide direct data access to applications and users, no matter where the data is stored.
Hammerspace GDE v5 data orchestration visualization.
Server level
Hammerspace CTO Trond Myklebust, maintainer for the Linux Kernel NFS Client, said: “Hammerspace helped drive the IETF process and wrote enterprise quality code based on the standard, making NFS4.2 enterprise-grade parallel performance NAS a reality.”
The company’s Parallel Global File System architecture separates the file metadata control plane from the data path and can use embedded parallel file system clients with NFS v4.2 in Linux, resulting in minimal overhead in the data path. It takes full advantage of the underlying infrastructure, delivering 73.12 Gbits/sec performance from a single NVMe-based server, providing nearly the same performance through the file system that would be achieved on the same server hardware with direct-to-kernel access.
When servers are running at the edge they can become disconnected. With GDE, file metadata is global across all sites, and on disconnect local read/write continues until the site reconnects, at which time the metadata re-synchronizes with the rest of the global data environment.
Data center level
File system parallelism and separation of metadata traffic applies at this level as well. GDE 5.0 can saturate the fastest storage and network infrastructures, orchestrating direct I/O and scaling linearly across different platforms to maximize aggregate throughput and IOPS. Hammerspace supports a wide range of high-performance storage platforms that organizations have in place today.
GDE 5.0 adds a 20 percent increase in metadata performance to accelerate file creation in primary storage use cases, and accelerated collaboration on shared files in high client count environments. It also adds RDMA (Remote Direct Memory Access) support for global data over NFS v4.2, providing high-performance access to all data in the global data environment, no matter where it is located.
Hammerspace GDE architecture graphic.
There are data services nodes, called DSX (Data Service eXtensions) In the GDE. These provide the ability for Hammerspace to parallelize front-side and store-side I/O across the network. DSX nodes handle all I/O operations, replication, data movement, etc., and are designed to scale out to over 60 nodes in a cluster to accommodate any level of performance requirements.
A test showed that with a cluster of 16 DSX servers, the Hammerspace file system hit 1.17 Tbits/second; the maximum throughput the NVMe storage could handle. It did this with 32Kb file sizes and low CPU utilization. Performance scaled linearly and to even higher levels when more storage and networking were added.
Metro, continental and global level
GDE supports AWS, Azure, GCP and Seagate’s Lyve Cloud. V5.0 adds:
Continual system-wide optimization to increase scalability, improve back-end performance, and improve resilience in very large, distributed environments
New management GUI, with user-customizable tiles, better administrator experience, and increased observability of activity within shares
Increased scale, increasing the number of Hammerspace clusters supported in a single global data environment from 8 to 16 locations.
Hammerspace says its Parallel Global File System orchestrates data automatically and by policy in advance to make data present locally without users wasting time waiting for data placement. The GDE can send data across dual 100GbitE networks at up to 22.5GB/sec.
It says this level of performance enables workflow automation to orchestrate data in the background on a file-granular basis directly, by policy. This makes it possible to start working with the data as soon as the first file is transferred and without needing to wait for the entire data set to be moved locally. It means data can flow to on-premises servers or in and into the public clouds.
Hammerspace claims it can saturate the network within the cloud when needed to connect the compute environment with applications. An EDA (Electronic Design Automation) test in Azure showed GDE performance scales linearly, making it suitable for compute-intensive use cases. For example, processing genomics data, rendering visual effects, training machine learning models and general high-performance computing.
Download a Hammerspace Technology white paper here to find out more.
