Home Blog Page 366

Google launches instant access Cloud Archive

By
Chris Mellor
-

Google has delivered on its intent, revealed last April, to deliver an AWS Glacier-like Cloud Archive storage service.

This is pitched as a tape replacement, holding data that is kept for a minimum of one year without access. There is no delay on data retrieval, unlike direct rival AWS Glacier Deep Archive (GDA), which quotes up to 48 hours.

Comparing costs between AWS Glacier Deep Archive and Google’s Cloud Archive will require a lot of work. You will need to define an archive workload, including the various types of operations on the data within it, and run a calculation process. There are worked out examples on Google’s website. No doubt AWS and Google sales resources will be happy to help here.

Geoffrey Noer, Google’s cloud storage product manager, blogs: “At Google Cloud, we think that you should have a range of straightforward storage options that allow you to more securely and reliably access your data when and where you need it, without performance bottlenecks or delays to your users.”  

Google cloud storage options

Archive storage prices

AWS’s GDA costs $0.0099 per GB per month whereas Google’s Cloud archive is more expensive at $0.0012/GiB per month. (A GiB by the way is a Gibibyte; 1GiB is 2 to the power 30 bytes; 1,024MiB, whereas 1GB is 10 to the power 30; 1,000MB.)

Google’s Cloud Archive retrieves data in less than a second, the same as Google’s Nearline and Coldline storage.

AWS and Google charge for data retrieval but on a different basis. Amazon has a three-tier price scheme based on retrieval time; standard ($0.02/GB) is within one hour, expedited ($0.03/GB) is within 5 minutes and bulk ($0.0025/GB) is within 48 hours.

Google Cloud Archive tiers retrieval, known as egress, based on the data amount. Zero to 1TiB/month costs $0.12 (egress worldwide bar China) or $0.23 (China). One to 10TiB costs $0.11 or $0.022/TiB (China again) and more than 10TiB costs $0.08 or $0.20/TiB for China.

Additional charges

Network usage when reading data in the archive within the same region carries no cost. But network usage on access between regions in a continent is $0.01/GiB. General egress charges apply on read access between continents.

There are also operation costs, set in three tiers. Data retrieval costs $0.01/GiB (Nearline), $0.02/GiB (Coldline) and $0.05/GiB (Archive). Class B operations cost $0.01, $.05 and $0.50 per 10,000 for Nearline, Coldline and Archive respectively. Class A operations cost $0.10, $0.10 and $0.50 per 10,000 operations for the three storage classes.

Google has defined Class A and B operations in a table.

Certain operations are free:

  • JSON API storage.channels.stop
  • JSON API storage.buckets.delete
  • JSON API storage.objects.delete
  • JSON API storage.projects.hmacKeys.delete.
  • DELETE with XML API
  • Delete with Object Lifecycle Management


Insight Partners buys Veeam for $5bn

By
Chris Mellor
-

Private Equity firm Insight Partners is acquiring the data protection powerhouse Veeam Software for about $5bn.

Veeam, founded and owned by Russians based in Switzerland, will now become a US company, and eligible to bid for US Government contacts. It will have US-based executives, with William Largent, currently EVP for Operations, promoted to CEO, taking over from co-founder Andrei Baranov.

Largent said in a press statement today: “Veeam has enjoyed rapid global growth over the last decade and we see tremendous opportunity for future growth, particularly in the U.S. market. With the acquisition, we are excited that our current U.S. workforce of more than 1,200 will be expanded and strengthened to acquire and support more customers… we believe this acquisition will allow us to scale our team and technology at an unrivalled pace.”

Veeam has achieved a $1bn annual run rate and claims more than 365,000 customers worldwide. These include 81 per cent of the Fortune 500. The company says it is the leader by units and revenue in the software data protection market.

Mike Triplett, Insight Partners managing director and Veeam board member, said: “Veeam’s strong growth, coupled with high customer retention, unparalleled data management solutions and the opportunities to expand services into new markets, make Veeam one of the most exciting software companies in the world today.”

He would say that, though, having just invested big bucks in the company for a second time. Insight Partners invested $500m in Veeam in January 2019.

The Veeam acquisition is expected to close by the end of March this year.

Co-Founders Andrei Baronov and Ratmir Timashev will step down from the Board. Timashev is currently EVP worldwide Sales and Marketing and here is no information on how that role will change. Blocks & Files expects him to step down in due course.

Veeam background

Veeam co-founders Andrei Baranov (left) and Ratmir Timashev (right).

Baranov and Timashev founded Veeam in 2006 and took in no outside funding until the Insight $500m injection in January 2019. They have had an amazing run of growth, capitalising on the VMware-led server virtualization boom. 

Reading between the lines of the acquisition announcement there is an indication that US sales are lower than anticipated, and that this can now change, driving Veeam revenues further upwards.

Veeam bought US firm N2WS in 2018 a year ago for its cloud-native enterprise backup and disaster recovery for Amazon Web Services. However N2WS was divested in August 2019 because it was unable to bid for US federal contracts.

Intel adds PCIe 4.0 support to Optane SSDs

By
Chris Mellor
-

Intel is adding PCIe 4.0 support to its gen 2 Optane SSDs.

This was revealed in a tweet by Frank Ober, a data centre solutions architect at Intel, and first reported by Storage Newsletter.

Optane is Intel’s 3D XPoint storage-class or persistent memory. It is bit-addressable, like DRAM, near DRAM in speed, and more expensive than slower, block-addressable NAND.

Current Intel Optane SSD 905p and M15 products have a PCIe v3.0 interface and use first generation Optane, Intel’s 3D XPoint memory media. Intel announced second generation Optane technology in September last year. This new version is some 50 per cent faster than gen 1.

The PCIe 4.0 interface specifies a 16Gbit/s data link speed with up to 16 links or lanes, delivering 64GB/sec, double PCIe gen 3’s 32GB/sec maximum.

Combining gen 2 Optane with PCIe 4.0 will increase data access speed nicely. It should enable Optane SSDs to at match and probably exceed the speed of Gigabyte’s PCIe 4 Aorus SSD ,which streams data at up to 5GB/sec.

Intel’s gen 1 Optane M15 gumstick drive tops out at  just 2GB/sec. A gen 2 Optane M15 with PCIe 4.0 support may well beat the Gigabyte drive.

Commvault tackles endemic under-performance with sales exec refresh

By
Chris Mellor
-

Commvault has appointed APJ and EMEA territory heads, as chief revenue officer Ricardo Di Blasio puts his own hires in the top sales spots. The enterprise data protection vendor wants to put an end to endemic sales under-performance.

The new recruits are Callum Eade, VP Asia Pacific Japan, who joins from VMware where he was APJ VP, software and defined data centre; and Marco Fanizzi, VP EMEA, previously head of Dell Italy.

Di Blasio said in a canned quote: “The industry’s most successful people are recognising the opportunity ahead of us. We’re focused on the needs of our partners and customers and our objectives are clear – simplify, innovate and execute. This is a new Commvault.”

Callum Eade (left) and Mario Fanizzi

CEO Sanjay Mirchandani, hired from Puppet Labs to replace retiring CEO Bob Hammer in March last year, presided over two disappointing quarters. The most recent quarter ended September 30, 2019 suggests things could be getting better.

Di Blasio joined the firm in May 2019 and his career highlights include global head of sales for VMware Cloud Platform Services at DXC Technology, CEO at Globetouch, COO at Cohesity, and SVP sales and marketing at VMware and EMC.

He wants Commvault’s sales to make a greater impact everywhere in its sales territories, its global accounts, and through its channels. He has initiated a wave of new exec appointments, including Anthony Faustini, VP of global accounts; Mercer Rowe, global head of channels; David Boyle, VP Sales for the Americas; David Wigglesworth, head of Hedvig global sales.

That’s seven new sales heads including Di Blasio. The team now has to deliver.

Sales execution issue history

Commvault has struggled with sales execution for several years. Each attempt to solve the problem has turned into a short-lived Band-Aid needing a fresh fix. Here are some examples of sales issues from the past:

  • Q1 2015 – Commvault reported a depressing set of results as sales headcount and mid-market product and pricing strategy faux pas cause a stumble – another stumble, as first quarter results were poor too
  • Q4 fy2017 – After three quarters of losses, Commvault made a $3.2m profit in its final fiscal 2017 quarter, just enough to tip the full year into profit.
  • Q4 fy2018 – CEO Bob Hammer said: “We know we can and must do more in order to return the company to sustainable, profitable growth.”
  • Q1 Fy2019 – analyst quote: “It is clear that the company continues to have execution/competitive risks.”

Kioxia fab fire update – two week hit to NAND flash production

By
Chris Mellor
-

Yesterday’s fire at Kioxia’s Fab 6 foundry in Japan could cause a two-week halt in production of NAND flash memory at the plant.

The fire broke out in a single machine in a clean room and was extinguished before the Fire Brigade arrived. Overall production in most of the plant is expected to be resumed within two weeks.

Aaron Rakers, a Wells Fargo senior analyst, estimates there will be a “less than a one per cent impact (or 0.8 exabytes) to total global NAND supply in 1Q20 if a two-week resumption is achieved.”

The Chinese language TechNews website reports (thank you, Google Translate) that most of the NAND wafers on the production line can be recovered through a re-work process.

Kioxia operates a joint venture with Western Digital in the flash foundry business.

MinIO earns Cisco validation for Data Intelligence Platform

By
Chris Mellor
-

Cisco has validated Minio as an object store for its scale-out Data Intelligence Platform. Minio joins Cloudian, Scality, Ceph and SwiftStack as approved suppliers.

Cisco launched the Data Intelligence Platform for business analytics, big data and AI workloads in June last year. The system concept includes two storage tiers, a Hadoop data lake layer for hot data and an HDFS/object store for warm data. 

Diagram from Cisco Minio Solution Overview document.

Each of the three elements in the diagram above can scale out independently.

Cisco has published a Solution Overview document describing how Minio fits in as a high-performance object store.

Cisco and MinIO

Cisco states that the MinIO system is fast, with a benchmark result of 12.8GB/sec aggregate read throughput quoted/ The company sais there is “a clear case for the Hadoop stack to operationalize data tiering from the Hadoop Distributed File System (HDFS) to MinIO, offering an attractive alternative for today’s critical workloads. The current design offers a proven deployment model for enterprise Hadoop while enabling a second, highly economical second tier of warm storage.”

For Cisco, the MinIO tier complements the Hadoop tier. That’s the same role played by the Ceph, Cloudian, Scality and SwiftStack object studs when the platform was launched in June. Yet MinIO has demonstrated that its software is faster than Hadoop.

Data Intelligence Platform buyers that use a single MinIO storage tier t could avoid buying the Data Lake/Hadoop part of the system and save a lot of rackspace and money too.

Cisco’s adoption of MinIO marks another validation for the object storage startup which has picked up support from VMware and Nutanix plus a certification from Qumulo.

Hardware scheme

For Cisco the hardware is based around three types of its UCS server. The AI/compute engine box uses UCS C240 M5 (2-socket, 2U, Xeon SP, Optane support, 26 x 2.5-inch drives)  and C480 M5 ML (8 x NVIDIA GPUs, 4U, Xeon SP, Optane support, 24 x 2.5-inch drives) servers. 

The apps running in the compute engine can access a Hadoop data lake tier, again using C240 M5 servers or an object storage tier using S3260 servers. These come in a 4U box holding up to 56 x 3.5-inch drives with dual Xeon SP or E5-2600 v4 controllers.

A hardware picture from a Cisco reference architecture document makes things clearer:

According to Cisco, the Data Intelligence Platform provides:

  • Extremely fast ingestion and engineering of data performed at the data lake
  • An AI computing farm, allowing different types of AI frameworks and computing resources (GPU,
  • CPU, and FPGA) to work on this data for additional analytics processing
  • A storage tier, allowing the gradual retirement of data that has been worked on to a dense storage system with a lower cost per terabyte, reducing TCO.

Kioxia suffers flash fab fire – foundry makes 3-4% of world’s NAND memory

By
Chris Mellor
-

A fire at a Kioxia flash foundry in Japan on January 7 could affect output for the company and Western Digital, Kioxia’s NAND chip partner.

According to a Google translated report in TechNews, a Chinese language website based in Taiwan, “it is believed that this event will have a significant impact on the supply of NAND Flash in the OEM or channel market in the first quarter of 2020, making the price increase trend more clear.”

Even before the fire NAND flash prices were expected to rise 40 per cent in 2020, according to memory chip maker sources polled by Digitimes last week.

In a letter to customers dated Jan 7, 2020, Kioxia said the impact to production was “under review”. The company is investigating the cause of the fire which damaged unspecified manufacturing equipment at Yokkaichi Plant’s Fab 6. The fire has been extinguished and there were no casualties.

Fab 6 makes 64-layer and 96-layer 3D NAND for Kioxia, formerly called Toshiba Memory, and Western Digital and pumps out 3-4 per cent of total world NAND production.

Kioxia letter to customers

In a note to subscribers, Wells Fargo senior analyst Aaron Rakers said the fire took place in a clean room and the facility has been shut down for inspection.

A 13-minute power outage at the entire Yokkaichi foundry last year caused a significant loss of worldwide NAND production capacity.

There’s an update to the Kioxia flash foundry fire here.

Hitachi Vantara targets solution sales with new army of consultants

By
Chris Mellor
-

Hitachi has completed the merger of Hitachi Vantara and Hitachi Consulting, first announced in September.

The company has grand plans for the enlarged Hitachi Vantara, which will focus on digital transformation and vertical solution sales. For example, it cites Lumada Manufacturing Insights, which “integrates silos of manufacturing data and applies AI and machine learning to evaluate and enhance overall equipment effectiveness”.

“A barrage of data and technology is disrupting enterprises and industries the world over,” said Toshiaki Tokunaga, chief executive officer and chairman of the board, Hitachi Vantara. “Through the integration of Hitachi Consulting, the new Hitachi Vantara will be uniquely equipped with the capabilities our customers need to guide them on their digital journeys. We’re going to be the company that helps customers navigate from what’s now to what’s next.”

Hitachi Vantara role and positioning

We can see how Hitachi Vantara’s IT products and consulting services can fit in with these activities, providing IT underpinnings across the group’s business units.

In other words Hitachi Vantara will look to sell its storage products as part of a larger deal, set in a market context relevant to one or more of its business units. Competitors, such as HPE or NetApp, could bid for the storage systems element but would not be able to match the rest of Hitachi’s offering, putting them at a disadvantage. This is Hitachi Vantara’s hope.

Competitors may suggest that as part of a larger business Hitachi Vantara may lose specialist storage technology expertise, putting it at a disadvantage.

Social Innovation Business

In a press release announcing the merger completion, Hitachi said the new Hitachi Vantara will “play a key role in advancing Hitachi’s 2021 Mid-term Management Plan, which aims to make the company a global leader through “Social Innovation Business.”

This strategy is outlined in Hitachi’s 2021 Mid-term Management Plan. This is based on three trends; increasing urbanisation, ageing workforces, and climate change-led  resource shortages. 

Hitachi 2021 Mid-term Management Plan slide. Lumada is the group’s digital transformation Internet of Things brand

Hitachi aims to use its product, IT and operational Ttechnology capabilities across its mobility, smart life, industry, energy and IT business units. Hitachi is focusing them on providing good economic results while meeting, environmental and social values.

The company wants make its products better suited to urban populations. This means providing safe, comfortable and environmentally-friendly railway transportation systems and services. It means designing smart cities to be more convenient and environmentally-friendly. This encompasses smart health therapy kit, city water supplies, energy generation, transmission and management, connected cars and connected home electronic appliances.

Seagate touts Lyve Drive Mobile System for business

By
Chris Mellor
-

Seagate unveiled the Lyve Drive Mobile System series of integrated and modular data storage drives, carriers and receivers for multi-stage workflow processes at CES 2020.

The aim is to capture data as it is generated and then transfer it to data centre facilities. Target applications include the Internet of Things, autonomous vehicle development and video production. Physical data transfer is needed where networking is too slow or costly. 

Jeff Fochtman, Seagate’s marketing VP, issued a quote: “Today’s data management tools are too costly and inefficient for businesses… [Lyve Drive] is Seagate’s first step toward a unified data experience, which will turn data’s possibilities into tangible growth for the world’s most critical industries.”

The range comprises ten products;

Lyve Drive product range.

Lyve Drive Cards – removable CFexpress format flash cards with up to 1TB capacity and an NVMe PCIe 3.0 interface. These are intended for video cameras and other endpoint data generating sources.

Lyve Drive Card Reader – portable reader for the CFexpress cards.

Lyve Drive Cartridge – portable single drive with U.2 interface.

Lyve Drive Cartridge Shipper – for shipping the Cartridge.

Lyve Drive Shuttle – a file copy and transfer device based around a 16TB disk or lower capacity SSD. It has a touchscreen e-ink interface and can be used without a host PC/notebook to capture data. There is also a network interface.

Lyve Drive Shuttle.

Lyve Drive Modular Array – a 4-bay array which can receive up to four disk drives or SSDs and transported in a carrier device.

Lyve Drive Mobile Array – a sealed 6-bay array the same physical size as the Mobile array. A demo version contains six 18TB Exos HAMR drives, which are not yet available.

Lyve Drive Array Shipper – container for shipping Lyve Drive arrays.

Lyve Drive Cartridge and Array Mounts for deploying Cartridges or Arrays in data generating locations.

Lyve Drive Rackmount Receiver – a 4U data ingestion hub which can receive two Lyve Drive arrays so their data can be ingested into data centre processes.

The Lyve Drive brochure does a good job of explaining how Seagate sees the various modular products interacting in use cases like video production.

As yet there is no pricing or availability information for the Lyve Drive products. Blocks & Files expects these products to roll out over the next few months. The publication of a Lyve Drive Shuttle data sheet suggests the range will roll out quite soon.

Micron samples faster DRAM with DDR5 DIMMs

By
Chris Mellor
-

Micron today became the first supplier to sample DDR5 memory modules. They use 1z nm technology and should speed memory access for multi-core CPUs.

DDR5 memory is expected to become the standard memory for most computing devices. Micron claims it will help servers cope with greater workloads by delivering 85 per cent-plus better memory performance than DDR4.

Tom Eby, SVP of the compute and networking business unit at Micron, issued a quote: “The key to enabling these workloads is higher-performance, denser, higher-quality memory. Micron’s sampling of DDR5 RDIMMs represents a significant milestone, bringing the industry one step closer to unlocking the value in next-generation data-centric applications.”

Double data rate

The JEDEC standards authority has specified DDR5 (Double Data Rate 5), the fifth DRAM generation, to have double the bandwidth and capacity of DDR4.

The technology is called double data rate because it transfers two messages per clock tick (MT/sec). Double data rate DRAM that runs at 1066.6MHz can transfer 2133 MT/sec.

Measured this way, DDR4 data rates range from 1600 to 3200 MT/sec. DDR5 increases this to 3200-6400 MT/sec.

At 3200 MT/sec, DDR4 memory provides 134.3 GB/sec effective bandwidth whereas DDR5 3200 will deliver 182.5 GB/sec effective bandwidth. This is a 1.36x speed up. The claimed >1.85x speed increase comes with a 4800 MT/sec rate product, as the chart below shows:

Micron’s one zee

Micron’s 1x memory nodes have 17nm cell sizes; 1y is unspecified but smaller and 1z is smaller again. Its three node shrinks after 1z are known as 1α, 1β and 1γ. The smaller the node the higher the chip density.

Micron sample chips are RDIMM format which are more reliable than ordinary DIMMs because the control signals are buffered in a register.

SK hynix will begin mass-production of its 1Znm DRAM in 2020. ITs 1Z measurement may well be different from that of Micron’s 1z.

Check out a Micron White Paper to dig deeper into its DDR5 DRAM tech.

Samsung’s Stellus Technologies emerges from stealth. Soon…

By
Chris Mellor
-

Samsung’s Stellus Technologies is preparing to step out of stealth. Its LinkedIn entry states: “We are officially one month away from unveiling our revolutionary Native File System for unstructured data to the world.”

We understand that to mean providing reliably consistent access latencies to enterprise unstructured data. But beyond that, little is known about its activities.

The company is running job ads for account executives in Media and Entertainment sectors and, secondly, the Life Sciences and Industrial Internet of Things (IIoT) sectors. Both mention working with Stellus partners. The Media and Entertainment post requires “Experience selling enterprise storage unstructured (File) data.”

Set up in 2016, Stellus is headquartered in Silicon Valley and has a development facility in Pune, India. LinkedIn records 52 employees. We can deduce more details about the company’s technology from its job ads, in which it describes itself as a “stealth company focused on solid state storage system technology for next-generation software-defined datacenters”.

One job opening exists for a senior software engineer skilled in file systems and object storage. It mentions the implementation of highly scalable, high-performance, distributed system software. Another ad mentions building the next generation object storage system for data centres. A further opening is for a staff software engineer in the NVMe area. These job requirements typically also mention file access protocols such as NFS, SMB (CIFS), and also LDAP. 

The ads also mention experience working with open source server-side software such as ZeroMQ, RocksDB, Ceph, Elasticsearch, and Key/Value (KV) stores is highly desired. RocksDB is a key:value-based store. Keep this point in mind.

Guessing game

Samsung manufactures DRAM, NAND chips and SSDs. We don’t know if Stellus is developing intelligent NVMe SSDs that are file access, object storage drives or storage arrays using its SSDs, and providing file access to object storage; unstructured data.

In a sense, customers can use any enterprise hard disk drive or SSD for storing unstructured data. The drives are simply incorporated into file and object storage systems. Blocks & Files believes Samsung’s notion is to implement the file and object storage in the drive by adding a CPU + DRAM + software.

Seagate first tried this in 2014 with its so-called Kinetic disk drives – which failed to take off. Samsung is trying again with SSDs.

In September last year Samsung revealed a Key:Value store SSD development. This has an object-like storage scheme on the drive and the company showed a demo in action using RocksDB. Pair this point with the RocksDB and Key:Value items above and Blocks & Files is shooting for a KV SSD featuring file access to an underlying object store.

Seagate lights up gamer SSDs

By
Chris Mellor
-

Seagate has launched a brace of external SSDs for gamers at CES. The accent is on speed and lighting effects.

The FireCuda Gaming SSD is based on Seagate’s FireCuda 510 NVMe SSD, an internal M.2 format drive, and sacrifices much of its speed in the transformation to an external drive. The 510 speeds along at up to 3,450MB/sec while the FireCuda Gaming manages up to 2,000MB/sec across  its USB 3.2 Gen 2 x 2 interface.

If your host system has a USB 3.1 port the speed falls to a maximum of 1,250 MB/sec.

While you are waiting for the IO to complete you can customise the drive’s LED lighting. This can be synchronised with a FireCuda Gaming Dock, “creating an immersive setup.” 

Seagate’s second new gaming SSD, the BarraCuda Fast SSD, has a SATA connection that can manage up to 540MB/sec. This is pedestrian compared to the FireCuda but better than a disk drive. It has a green LED light effect.

Both drives come in 500GB, 1TB and 2TB capacity points and have a five-year (FireCuda) or three-year (BarraCuda) warranty. The 510 uses Toshiba 64-layer 3D NAND set up in TLC (3bits/cell) form. We think the Gaming SSD version uses the same NAND while the BarraCuda Fast SSD uses 96-layer NAND.

The FireCuda Gaming SSD retails for $189.99 (500GB), $259.99 (1TB), and $499.99 (2TB) and is available in March. The BarraCuda Fast SSD is out in February,and retails for $94.99 (500GB), $169.99 (1TB), and $299.99 (2TB). Slower speed gets you lower prices than the FireCuda Gaming SSD.

1...365366367...435Page 366 of 435

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