Wasabi, a cloud storage startup, has devised a pricing model whereby customers buy capacity up-front in return for lower charges.
Wasabi’s Reserved Capacity Storage (RCS) provides enterprises with price predictability and one-time billing. The deal sees storage costs cut by up to 27 per cent compared with pay-as-you-go when customers commit to fixed price terms of up to five years. Customers pay only for storage that is reserved; there are no fees for data egress, API requests or data retrieval.
David Friend, CEO of Wasabi, gave out a quote: “With most of the world’s data still stored in on-premises hardware, we want to make it easy for people to migrate their data to the cloud and start saving enormous amounts of money in the process.”
Cloud storage customers like predictability
Wasabi said a 100TB/five year RCS plan costs under $40,000 and includes technical support. By contrast, a five-year, on-premises 100TB hardware appliance deal from an unnamed vendor lists at $130,000. This rises to more than $240,000 with maintenance over five years and does not include data centre or people costs.
Wasabi said its research shows many customers prefer the predictability of a fixed price purchase order. The company thinks RCS will be useful for backups, second copies of data, and archives. Also there is a make-margin bonus for channel partners, who can resell the service as a ‘boxed cloud’ SKU.
Wasabi claims its service is faster and 80 per cent cheaper than Amazon S3. The company was founded in 2017 and it said it has 10,000 customers worldwide and has deployed more than 100 petabytes to date.
Cloudian and Veeam have introduced Amazon’s S3 cloud Object Lock on-premises to combat ransomware attacks.
In the event of an attack, ransomware-locked files can be replaced by Veeam backup data stored on the Cloudian HyperStore appliance, which sits in the data centre behind the firewall. These will be free of the ransomware infection, according to Cloudian, which said its storage has US government-certified data immutability.
Danny Allan, Veeam CTO, said in a prepared quote: “Preventing a ransomware attack has become increasingly difficult, so we’re focused on ensuring users can quickly and easily recover from such an attack – that’s the peace of mind that we’re delivering with on-premises S3 Object Lock in the new Veeam Availability Suite V10.”
S3 Object Lock
Launched yesterday, Veeam Availability Suite (VAS) V10 introduced S3 Object Lock for backup data, with backups to an S3 target made immutable through a software setting. Cloudian’s HyperStore is on-premises object storage which supports the Amazon S3 APIs. Cloudian supports Object Lock functionality and is now a VAS V10 Object Lock target device.
S3 Object Lock provides a virtual air-gap that logically disconnects stored objects from connected commuter systems. In Object Lock’s Compliance Mode, stored data can’t be overwritten or deleted by any user including the root account in AWS within the retention period. There is no way to change the retention mode or to shorten the retention period.
According to Cloudian, restores from HyperStore are faster than via the public cloud because an on-premises object store generally has a faster network link.
The company said the Veeam partnership means a HyperStore Object Lock system can be a platform for ransomware protection-as-a-service. Also in the case of VMware Cloud Providers, it can deliver single-point management via VMware vCloud Director.
Ransomware attacks are increasing
Cloudian quotes an US inter-agency report that reveals 4,000 ransomware attacks “have occurred daily since January 1, 2016. This is a 300 per cent increase over the approximately 1,000 attacks per day seen in 2015.”
Preventing users from disclosing account passwords to phishing emails and other preventative measures are useful hygiene but are not 100 per cent effective. According to the US federal agencies, ransomware attacks can be expected to surmount password and other barriers.
Organisations should prepare accordingly and take the following preventative measures:
Back up data regularly and verify the integrity of those backups and test the restoration process to ensure it is working.
Conduct an annual penetration test and vulnerability assessment.
Secure their backups – ensure backups are not connected permanently to the computers and networks they are backing up. Examples are securing backups in the cloud or physically storing backups offline. Some instances of ransomware have the capability to lock cloud-based backups when systems continuously back up in real time, also known as persistent synchronization. Backups are critical in ransomware recovery and response; if you are infected, a backup may be the best way to recover your critical data.
Western Digital is releasing a lower capacity embedded flash drive for smartphone makers. We infer this is a response to low demand for the company’s 512GB option, which launched last March.
The iNAND MC EU521 embedded flash drive stores up to 256GB and transfers data at up to 800GB/sec. This is slightly faster of its immediate predecessor, the iNAND MC EU511 – but only half the capacity. Presumably it is priced accordingly.
Itzik Gilboa, director of the mobile storage business, told us by email: “Customers’ preferences were leaning toward the 256GB capacity. This is about the practicality of having the right product at the right time.”
The EU521 delivers a six per cent speed improvement over the EU511. This may not be noticeable to the consumer, considering the EU511 can store a two hour movie in 3.6 seconds, according to WD.
Huibert Verhoeven, WD’s Automotive, Mobile and Emerging business unit SVP, issued a quote: “Smartphones now demand more performance and capacity as they often serve as the primary computing device for everything from streaming video, playing music, gaming and photography, to payments and mapping.”
The EU521 speed increase is achieved by using a newer UFS 3.1 Gear 4/2 lane specification. A WD spokesperson said: “This version of UFS (3.1), with Write Booster based on Western Digital’s 6th Generation SmartSLC, enables improvements in robustness in performance under load, in ‘save’ times for edited videos, for example, and download times are improved for movies and large files.”
Faster movie downloads
The iNAND MC EU511, announced just under a year ago, supports JEDEC’s Universal Flash Storage (UFS) 3.0 Gear 4/2 Lane specification and was for use in all smart mobile devices. It can transfer data at up to 750MB/sec and uses an SLC cache in front of its 96-layer 3D NAND store formatted with TLC flash.
EU521 with 256GB maximum capacity.
Verhoeven also said: “The SLC caching in the iNAND EU521 with Write Booster offers users several key performance improvements that, when coupled with 5G, are expected to bring faster movie downloads than ever before.”
The EU521 uses the same TLC flash and SLC cache design as the EU511 and supports a later UFS 3.1 Gear 4/2 lane specification. JEDEC’s UFS 3.1 standard has three extra features over UFS 3.0:
SLC cache write booster, which WD had already implemented in the EU511,
DeepSleep low power state,
Performance throttling notification so the device can tell its host its slowing storage speed due to a high temperature reading.
WD is eyeing 5G smartphones as target host devices for its EU521 chip and the chip will be available in March.
Egnyte has turned separate file sharing and protection services in the cloud into a unified content services offering with tiered storage and machine learning technology for better analytics.
The US venture-backed company offers on-premises and in-the-cloud file sharing and collaboration for enterprises, and claims 16,000 customers and a $100m revenue run rate.
Egnyte took in $75m of funding in 2018. It is using the money to expand, and will reach a total of 770 employees by the end of the year, with 225 added in the last six months. There are now more than 220 engineers in Poland and 35 employees in the UK.
Egnyte content services platform
CEO Vineet Jain told us in a briefing that the companyt’s cloud customers are increasingly buying separate services, Egnyte Connect and Egnyte Protect, in a single transaction.
He said: “Customers wanted one solution; a platform story converging content and governance. a content services platform category has been emerging.”
Egnyte customers accelerated moves towards the cloud from 2017 in response to growing volumes of unstructured data. File sharing was their primary need, according to Jain. “We thought they would move quite slowly, but under-estimated the speed. It was exponential, not linear.”
Now 45 to 50 per cent of Egnyte’s customers have a hybrid, on-premises and public cloud strategy. The others are all-in to the cloud. Jain thinks the IT world will stay hybrid, and not abandon on-premises installations.
As Egnyte’s customers moved to the cloud they wanted both file sharing and file security and governance. Egnyte saw “two colliding worlds going in one direction”.
The customers also wanted to control storage costs and Egnyte’s new content services offering has three-tiered cloud storage with fast access, medium access and slow access data tiers.
Jain said: “We’re actively working on data minimisation and use heuristics and policies to archive colder data. Generally, only three to five per cent of data is active.”
The software manages data egress and ingress charges to keep these low. It also uses machine learning technology to provide better governance, data classification and control for compliance and data privacy regulations. Egnyte’s software supports GDPR-type rules in 50 separate jurisdictions.
Google has announced the Anthos Ready Storage program for third-party on-premises storage suppliers.
Six vendors have qualified for the initial line-up: Dell EMC, HPE, NetApp, Portworx, Pure Storage and Robin.io.
Based on Kubernetes, Anthos is Google’s hybrid cloud. It enables containerised application workloads to be moved in any direction between Anthos environments on-premises and in the AWS, Azure and Google public clouds.
The overall Anthos idea is that customers can have a hybrid cloud Google Anthos-based environment with cloud-native applications moving between on-premises data centres and public clouds using a Kubernetes travelator, so to speak, and certified storage suppliers. These suppliers become Anthos supporters and sing the Anthos anthem.
Rayn Veerubhotla, Director, Partnerships, Google Cloud and Manu Batra, Product Manager, Google Cloud said in a blog today, that each company had:
Demonstrated core Kubernetes functionality including dynamic provisioning of volumes via open and portable Kubernetes-native storage APIs.
A proven ability to automatically manage storage across cluster scale-up and scale-down scenarios.
A simplified deployment experience following Kubernetes practices.
NetApp provides ONTAP unified file and block storage on all-flash and hybrid flash/disk arrays as well as object storage. Pure Storage supplies all-flash block (FlashArray) and object (FlashBlade) storage.
Portworx is a Kubernetes-based block storage provider while startup Robin.io has developed a a hyperconverged (compute+storage+networking) system, which extends Kubernetes with built-in storage, networking, and application management for Oracle databases and its ecosystem of applications such as WebLogic Server, Oracle RAC and EBS.
Dell EMC has announced a rack-in-a-box mini data centre, complete with new Streaming Data Platform software to store and analyse incoming data.
The Modular Data Centre Micro 415 puts a populated data centre rack inside an extreme temperature-resistant enclosure, with key lock doors, power, cooling, and remote management. There is an option for smoke detection and fire suppression.
The edge system can sit in a car park and act as a concentrator storage and processing box for locally generated data that needs analysing and acting upon faster than a remote on-premises or public cloud data centre can deliver.
Modular Data Centre Micro 415
Dave McCarthy, IDC research director for edge strategies, provided the obligatory analyst quote: “Edge computing will become a critical element for most every enterprise looking to gain business insights and competitive advantage.”
The Micro 415 box can be populated with new PowerEdge XE2420 servers. These two-socket systems can handle the incoming data streams and crunch the data and they have the flexibility to add up to four accelerators and 92TB of storage.
Management and monitoring is provided by the new Dell Remote Access Controller, iDRAC 9 Datacenter, which supports remote deployment and streaming telemetry.
Streaming Data Platform
Incoming data from devices feeding the 415 is handled by Dell EMC’s new Streaming Data Platform (SDP) software. This is cloud-native software requiring Kubernetes orchestration. It uses Apache Flink and Pravega components, and has elastic 2-tiered storage providing “instant access to data and infinite storage and access to historical data”.
Open source Pravega is a stream store built to handle large amounts of continuously arriving data. It handles ingestion and storage for all types of streams, including continuously streaming unbounded byte streams.
Flink is an open source software project and Dell EMC distributes the Ververica Flink enterprise version of Flink as its supported embedded analytics engine. Customers can develop Pravega software connectors or use community connectors that integrate with other engines.
The Streaming Data management platform is proprietary Dell EMC software. It integrates the other components and adds security, performance, configuration, and monitoring features. There is a web-based UI for administrators, application developers, and data analysts.
This management platform runs in a Kubernetes container environment that isolates projects, manages resources, and provides authentication and RBAC (role-based access control) services. The verified container environment is Pivotal Container Services (PKS), which is a Kubernetes service provider and deploys an instance of Kubernetes.
SDP runs on a reference hardware platform for processing and elastic scale-out storage capacity for tier 1 storage. Details of this were not available at announcement time but we’re thinking of NVMe SSDs inside a Dell server.
Persistent tier 2 scale-out storage is provided by an Isilon cluster. Tier 2 storage may be a file system for test and dev, or for use cases where only temporary storage is needed. This can be a local mount or an NFS mount point.
Pravega event data streams are durable, ordered, consistent, and transactional. The software coordinates a two-tiered storage system for each incoming stream.
Tier 1 stores the recently ingested tail of a stream temporarily.
Tier 2 places ingested data into the Isilon long-term storage.
Users can configure data retention periods.
Availability
The PowerEdge XE2420 begins initial availability in the second quarter of 2020 and the Modular Data Center Micro is slated to make its public debut in the second half of 2020.
The iDRAC9 Datacenter and the Streaming Data Platform are available globally today.
Comment
In concept the Modular Data Centre Micro 415 is a scaled down version of Sun’s original 2006-era containerised data centre. That box was copied by Dell in 2008. It now appears that rack-in–box units might be more popular. We expect other server vendors, such as Cisco, HPE, Lenovo and Supermicro, to follow in Dell EMC’s footsteps. Also, Dell’s “infinite storage” claim is logically impossible in a bounded enclosure. Therefore we anticipate the company will build public connectivity to a third tier of storage, making it easier to justify this claim.
Veeam’s tenth major release of its software adds single-click recovery of multiple VMs at a time, anti-ransomware backups with AWS’s S3 Object Lock, a data integration API for third-party software, and many more new features.
Altogether there are 150 enhancements to make Veeam’s Availability Suite work faster, cover more use cases, and integrate with downstream software products for analytics and other purposes,
Veeam CTO Danny Allan said in a prepared quote today: “V10 is the biggest release in the history of Veeam, and the enhancements we’ve made take data protection to the next level, creating the simplest, most flexible, and most reliable solution for hybrid-cloud environments.”
In more detail the release offers:
Improved and simplified protection of large file shares and file servers with NAS (Network-Attached Storage) backup,
Next-generation Instant Recovery engine, and Multi-VM Instant Recovery for automated disaster recovery from massive data centre outages,
Faster, automated copies to on-premises and cloud object storage using S3,
Increased security via immutable backups with S3 Object Lock (to ensure that data is protected against threats such as ransomware, insider threats and malicious admins.)
More platform support including new capabilities for Linux, Nutanix AHV, PostgreSQL, and MySQL.
Veeam Data Integration API to make backup data available for third-party data analysis software integration
The recovery of multiple VMS with a single click, together with faster recovery, makes disaster recovery processes simpler and faster. It also means that DR verification testing becomes easier to do and so ensure DR processes are accurate.
More powerful NAS backup
Veeam added the capability to backup NAS servers to tape in Veeam Availability Suite v9.5 Update 4 via NDMP, the standard market protocol for NAS backup. NDMP has scalability limitations and entire NFS volumes needed to be backed or restored, not files within them. Also there was no support for NAS backup to disk-based target systems.
V10 adds file-level restores from NAS backups and support for disk targets. It also improves the UI experience with registering a CIFS share as the data source for file to tape jobs, which was quite clumsy before.
Performance optimisations have been added to the v10 release to speed file to tape and NDMP to tape jobs. File-level restore from an image-level backup will be faster with v10.
There’s no need to run a Mount operation when restoring NAS backups, as they are not image-level. This makes the restore process almost instant and faster than file-level restore from an image backup.
The NAS backup file format differs from the image-based backup format in being object-based with added metadata. To restore files from an image backup requires a so-called crawl through the image backup to find the desired file and folder. With the object format NAS backup, wanted files and folders are found and located via a metadata lookup, which is faster than image crawling.
Users need a Veeam Universal License for every 250GB backed up to disk with a NAS CIFS backup. No additional licenses are needed for NAS CIFS backup to tape. Veeam has a 1PB NAS specific license to enable customers to protect PB-scale NAS environments.
Faster copies with S3
A native S3 interface for Veeam Backup & Replication, called Cloud Tier, was part of the Release 9.5 update 4, generally available since January 2019. Users can push backups to an S3-compatible service to increase overall backup capacity. That included AWS S3, Azure Blob Storage, and IBM Cloud Object Storage.
Michael Cade, senior global technologist at Veeam, told us via email: “In V10, this feature is enhanced with the ability to immediately copy backup files to object storage. This enables our customers to follow the Veeam 3-2-1 rule (create three copies of the most important files, use at least two different media and make sure that one copy is always stored off-premise) because as soon as those backup files land on the performance tier of the scale-out, backup repository they are immediately copied to object storage.”
Additionally, there will also be the ability to make those backups immutable by leveraging the object lock API within AWS S3. Veeam’s S3-compatible object storage vendors will also implement these API capabilities.
S3 object lock
Amazon S3 Object Lock is a WORM (Write Once Read Many) technology which prevents stored objects from being deleted or overwritten for a specified amount of time (retention period).
It has two modes;
Governance mode – there is no specified retention period and object locking can be removed by any user with s3:PutObjectLegalHold permission.
Compliance mode – data stored with this retention can’t be overwritten or deleted by any user including the root account in AWS within the retention period. There is no way to change the retention mode or to shorten the retention period.
Additional platform support
Veeam is introducing Linux proxies in V10 to enable its customers to use either Windows proxies or Linux in their environment.
There have been further enhancements around NFS as a Veeam Backup Repository. Previously this was supported but needed a Linux server to write data to the NFS share – a middleman, in essence. This requirement removed in V10.
The Veeam Backup Repository is already able to use the XFS file system to store Veeam backup files, as well as Windows RFS and other file systems. V10 supports XFS and a block cloning technology which together provide similar space savings to those seen with Windows REFS. This means faster merge operations and space-less synthetic full backups.
Nutanix AHV gets a tighter integration with the Veeam Availability Suite, centralised management and AHV proxy deployment from Veeam Backup & Replication, and closer backup and restore integration with the Nutanix AHV snapshot engine.
Performance and scalability enhancements have also been added to PostgreSQL and MySQL as well as further integrations into Application Aware processing for the database types.
The tech storage industry consensus is that enterprise hard drives sales will fall at the expense of solid state drives (SSDs) in all areas except for hyperscale data centres.
This migration seems well underway – however… 2019’s fourth quarter saw the SSD share of all enterprise shipped storage capacity shrinking. Total SSD capacity shipped was 10.4 per cent of total enterprise storage capacity shipped, down from 12 per cent a year ago.
This unlikely turn of events was recorded by the market research firm TrendForce and revealed in a note sent by Wells Fargo analyst Aaron Rakers to his subscribers. Total shipped enterprise storage capacity across disk drives and SSDs was 169.79 EB, with disk drives up 83 per cent year on year to account for 144.1 EB,
Within this figure, mission-critical HDD capacity was 5.93 EB, up four per cent. This is the high-performance, 2.5-inch format sector – which is most at risk from SSD cannibalisation. Yet it is still growing in capacity terms. Nearline enterprise drive shipped capacity was 138.2 EB, up a stonking 89 per cent.
Total enterprise SSD capacity shipped was 16.69 EB, up 60 per cent y/y, and lagging behind nearline enterprise disk drive capacity growth. Some 10.325 million enterprise SSDS were shipped, a rise of 43 per cent. There has been a flash supply glut but this has not yet prompted enterprises to buy a lot more flash.
Samsung accounted for 44.4 per cent of the enterprise SSD capacity in the 2019 fourth quarter, Intel accounted for 21.9 per cent, Micron 12.7 per cent, sk Hynix 6.1 per cent, Kioxia 5.7 per cent, and WD 4.8 per cent.
N.B. Blocks & Files has covered the topic of HDDs vs SSDs in enterprise data centres in several articles, including these four below.
Chinese DRAM and NAND foundries have so far escaped the impact of coronavirus, which has disrupted the supply chains of many global industries.
Trendforce, a market research firm, notes that China-based DRAM and NAND flash production, such as Samsung’s Xi’an fab, SK Hynix’s Wuxi fab, YMTC, CXMT, and JHICC, will not be affected by the deadly outbreak, certainly in the short term.
Semiconductor fabs are highly automated, have relatively low demands for manpower and operators stocked up with raw materials before the Chinese Lunar New Year. These factors will prevent output shortages in the near term.
This will remain the case – so long as additional materials that need to be imported can pass through customs as usual. Foundry output can be delivered to customers in China because semiconductor fabs hold national special licenses. These allow them to ship their products throughout domestic China, even with cities under quarantine.
TrendForce forecasts DRAM prices in the first 2020 quarter will continue an uptrend in spite of the outbreak. It said client-side inventories are still showing shortages and purchasing momentum of memory products will continue despite any problems of labor and material shortages downstream .
Your occasional storage digest this week features boatloads of data protection news from Atempo, Clumio, Cobalt Iron, Datrium, Druva, Infrascale and Overland-Tandberg.
Also we cover mainframe data extraction and HAMR disk platters.
Model9 gets data off mainframes
Startup Model9 is developing cloud data management for mainframes and has raised $9m in A-round funding led by Intel Capital.
It has a zOS agent with a virtual tape library interface that takes in mainframe data and can pump it out to AWS, Azure, GCP and the IBM Cloud, or Dell EMC ECS, Hitachi Vantara HCP, Infinidat and NetApp StorageGRID storage. Amazon’s EBS, EFS and S3 interfaces are supported.
Model9 diagram.
Anthony Lin, a VP at Intel Capital, said: “We identified Model9 as having a first-mover edge with a software-only solution to extract and load mainframe data from the mainframe to any cloud or to any on-prem’ storage through a standard TCP/IP protocol. This ability opens up a variety of DR, Archive and Storage use cases, scaling to thousands of companies from the financial and public sectors, among others.
“Furthermore, Model9’s solution enables conversion of mainframe-compatible file formats to object storage, thus allowing the company’s customers to utilize the latest business analytics, big data and AI solutions to gain insights from previously untapped data sets.”
SDK HAMRs aluminium
Japanese storage media manufacturer Showa Denko (SDK) has developed Heat-Assisted Magnetic Recording (HAMR) disk platters, using Aluminium and not glass. HAMR uses heat to overcome the room-temperature high resistance (coercivity) of specific magnetic media formulations to changing their magnetic polarity.
A competing MAMR (Microwave-Assisted Magnetic Recording) uses microwaves to do the same job. Western Digital is introducing MAMR aluminium platter drives in the relatively near future while Seagate intends to produce HAMR glass-based platters. Toshiba has a MAMR inclination, as far as we know.
Western Digital and Seagate produce their own platters while Toshiba buys them in, with SDK as a supplier.
SDK’s HAMR platter uses thin films of Fe-Pt magnetic alloy with new magnetic layers and temperature control in its production. Its claimed to have magnetic coercivity several times as high as the existing most-advanced HD media, while achieving low noise due to very small crystal grain size and optimized grain size distribution control.
This announcement opens up the possibility of Toshiba introducing HAMR technology, were it to develop or obtain HAMR read/write heads.
Shorts
Data protector/archiver and migrator Atempo has joined the Active Archive Alliance. An active archive is said to be that provides online access to archived data and that includes data stored on SSDs, HDDs, tape or in the public cloud.
Atempo CEO Luc d’Urso provided a canned quote: “Our vendor-agnostic Miria platform for archiving [needs] moves data intelligently and rapidly between different locations. Our current and future machine learning and artificial intelligence innovations will serve to boost Miria’s power and scope.”
Active Archive Alliance members and sponsors include Atempo, Fujifilm, Iron Mountain, Harmony Healthcare IT, Komprise, MediQuant, Quantum, Qumulo, QStar Technologies, Spectra Logic, StrongBox Data Solutions and Western Digital.
SaaS backup startup Clumio said its Enterprise Backup as a Service has gained “VMware Partner Ready for VMware Cloud on AWS” validation. It has tested and verified interoperability for Clumio Backup as a Service with VMware Cloud on AWS.
Data protector Cobalt Iron has extended its Compass enterprise SaaS network-attached storage (NAS) backup with a proxy Compass NAS Agent. It’s claimed to be better than the existing and supported NDMP (Network Data Management Protocol) for backing up NAS filers. The new agent has a parallel workflow – faster than NDMP’s sequential process.
Hyperconverged storage supplier Datrium is focusing on cloud-based disaster recovery and has extended its channel program to new regions and added a new Datrium Select tier. The other tiers are Preferred and Authorised. Invitation-only Select Tier partners receive the highest level of training and certification in DR with VMware Cloud on AWS.
LucidLink is partnering video management software supplier Milestone Systems to offer immediate and concurrent access to video surveillance in the cloud. LucidLink’s Filespaces product provides instant access to large data sets over long distances, streaming it into on-site caches as it is needed. The cloud supplier is Wasabi.
Data protector Overland-Tandberg is partnering with the U.S. Black Chambers (USBC) to provide data protection, storage, and archive technology solutions to USBC members. They say the partnership “will help Black-owned businesses survive challenging data loss incidents including system failures, weather disasters, or cybersecurity.”
The University of Florida is using Qumulo’s scale out filers to store data for researchers in bioinformatics, genomics, DNA sequencing and cancer. The University of Florida IT Department was using an Isilon system in 2014.
Seagate has opened one of its Lyve Labs in Tel Aviv. Local companies can use the lab to find out how they can use Seagate drives to store their data. One offering is a Lyve Drive Shuttle which can move data physically in shuttles instead of uploading across a network. This can be more cost-effective and faster at scale. Another Lyve Labs center is located in Longmont, Colorado, United States. Additional Lyve Labs centers are being planned globally
People
Data protector Druva has appointed Bernd Wachtler, ex-Veritas, as partner director for its DACH (Germany, Austria, Switzerland) region. There is also a new channel program called Druva Compass with expanded enablement resources, an accreditation curriculum, streamlined sales process, and opportunities for recurring margins.
Cloud-based backup and DR service supplier Infrascale has appointed Russell Reeder as its CEO. He led successful exits at content-on-demand and digital publishing platform LibreDigital (which sold to RR Donnelly), and cloud hosting business MediaTemple, (which sold to GoDaddy).
It has also appointed Brian Kuhn as COO, Robert Peterson as CFO, Carolyn Kress as Chief People Officer, Lindsay Haun as VP Global Customer Success, and Adam Berger as VP Global IT and Cloud Operations. Infrascale founder and former CEO Ken Shaw will continue as an advisor to the company.
BackBlaze has reported “a higher-than-typical failure rate among some of our 12TB Seagate drives” at its cloud storage facility.
In a recent hard drive stats report, Yev Pusin, BackBlaze’s director of marketing, notes Seagate’s ST12000NM0007 12TB drive has a 3.32 per cent annualised failure rate (AFR).
This is worse than the previous least-reliable drive, a 4TB Seagate model with two per cent AFR. Backblaze has two HGST 12TB drive models and their AFRs are 0.4 per cent and 0.56 per cent, as the table below shows.
A chart of the drive model AFRs indicates how Seagate’s 12TB drive stands out.
In 2017 and 2018, the 12TB Seagate drive’s AFR was 2.01 per cent and 1.39 per cent.
“It’s worth noting that situations like this are not uncommon in our industry and often go unnoticed by the end-users of the services,” Pusin wrote, “as most cloud providers do not inform customers or the public when they experience issues like what we’re describing.”
Testing new drive platforms
BackBlaze is working with Seagate to understand what’s going on. This work involves “things like testing new drive platforms in real workload environments, providing telemetry tools to predict failures, performing ongoing custom adjustments, and employing firmware development and replacement units (RMAs). … we’re also working together on a migration effort to replace these particular drives in our data centres.”
As a fix, the company is migrating to a different Seagate 12TB drive model, the ST12000MN0008, and is also looking at 14TB and 16TB drives.
Pusin said: “In the near term, we expect to see moderately increased failure rates for this specific subset of 12TB drives, but as we complete the drive migration, we project our fleet’s failure rates will restore to historical norms.”
Backblaze customers should be unconcerned because the company’s processes mask drive failures and recovery.
NetApp shares fell 13 per cent last night on the back of disappointing third quarter results and lower guidance for the next quarter.
The storage giant recorded its fourth quarter of declining revenues in a row as large enterprise customers slowed on-premises product purchases, public cloud revenue growth was anaemic, and sales force growth has yet to bear fruit.
Q3 fy20 revenues of $1.4bn were 10.4 per cent lower than a year ago. Net income was up 11.3 per cent to $277m – equivalent to 19.8 per cent of revenues. The Q4 revenue outlook is $1.53bn at mid-point – a 3.8 per cent decline. The full fy2020 revenue outlook is a 10 per cent decline on fy2019’s $6.15bn to $5.53bn.
CEO George Kurian said in statement: “Our third quarter results, highlighted by strong gross margins, cash flow and operating leverage, reflect our continued operational discipline.”
The company may be run more efficiently. But its customers are hesitant about moving to the cloud and the all-flash array boom looks to be over.
NetApp has a Data Fabric strategy of helping customers move to a multiple public cloud, hybrid on-premises model. But cloud data services annualised recurring revenues of $83m are low. Wells Fargo MD Aaron Rakers estimates this translates to $15m to $18m in the quarter. As we said, NetApp’s enterprise customers are cautious about moving to the public cloud.
Revenue trend
NetApp has a strongly seasonal revenue pattern, as this chart shows:
The current fiscal year’s quarterly revenues are tracking lower than last year, indicated by the yellow arrows on the chart. By charting the quarterly revenues against fiscal years we can see the revenue decline more clearly.
Kurian said: “We see significant opportunity ahead and are focused on replicating the areas where we have proven success. Our strong business model enables us to navigate the market dynamics, while making the strategic investments necessary to position the company for long-term growth.”
Large account problem
The earnings call revealed a large account issue. Outgoing CFO Ron Pasek said: “We had zero ELA (Enterprise License Agreement) revenue in the quarter, although we had expected approximately $50 million. Product revenue of $787 million, decreased approximately 19 per cent year-over-year.” He pointed out the all-flash array run rate was $2.3bn. Wells Fargo MD Aaron Rakers said this is down 1 per cent y/y.
NetApp CEO George Kurian
Kurian said: “It is in our largest accounts, which have the greatest exposure to the macro that the demand environment is the least predictable.” The macro” refers to the uncertain overall economic environment and NetApp’s customers are buying to mostly satisfy short term needs only as a result. Kurian pointed out: “Buying cycles are longer and the amount of spend per transaction is smaller.”
He added: “Our largest customers who were most impacted by the macro were heavily all-flash customers… our biggest accounts underperformed or bought less impacted our all-flash business more substantially than it impacted pretty much every other product in the Company. the ELAs were also heavily all-flash-oriented. So, both of those have been contributing factors to the year-on-year declines in the all-flash category.”
Sales fix
Kurian said NetApp is “focusing our compensation plans and our sales objectives on returning to growth in the all-flash category. … we have every confidence that we should be able to meet or beat the market next year.”
Rakers told subscribers that will coincide with a “more aggressively competitive landscape with the long-awaited Dell EMC midrange refresh cycle now expected in 1HC2020,” referring to Midrange.next.
NetApp’s prime fix is to increase the sales headcount, mainly in America, to increase sales outside large accounts and get a larger share of the spend in large accounts. Kurian said: “We are on track to increase sales capacity by approximately 200 primary sales resources by the end of Q1 fiscal year ’21, without adding to the total operating expenses for the Company.
“The majority of the sales headcount will be deployed in our Americas geography. They will be focused on acquiring new accounts and engaging new buyers like cloud architects in existing accounts.”
It takes three to four quarters for new sales reps to become productive and NetApp sees the results starting to appear in the financial year. Therefore the next quarter’s results are likely to be down as well as this one.
However, Pasek noted: “We are seeing early signs of success from our strategic investments in sales coverage, which provides confidence in our ability to return the Company to long-term growth.” Also Kurian said there have been good early results in gaining new customer accounts.
