Server offload card startup Pensando has its product available as a factory-installed option on HPE servers and is partnering with Elastic, Guardicore, ServiceNow, and Splunk to extend its product’s usability.
The news was announced at HPE’s virtual Discover event. Pensando’s Distributed Services Card is a PCIe-attached device bearing its proprietary chip that carries out security, networking and storage functions on behalf of its host server, freeing the host to run more application code. Infrastructure services such as security, encryption, flow-based packet telemetry, and fabric storage are deployed on the DSC with Pensando’s Policy and Services Manager (PSM) software collecting events, logs, and metrics from the installed DSCs to help with troubleshooting.
Prem Jain, CEO of Pensando, provided a canned quote: “While the last year has been challenging on many fronts, we’re very excited with the progress we’ve made, in particular what we’ve done with our ecosystem partners and HPE.”
The DSC can be factory-installed on HPE’s ProLiant and Apollo servers and Edgeline Converged Edge systems with single-call HPE hardware and software support.
Pensando’s new partnerships provide ecosystem support for security and infrastructure troubleshooting:
Integration with Elastic and Splunk aids network and security operations with identifying issues before they grow into major problems. Operations can uncover workload and workload connection issues and begin remediation in real time.
Guardicore integration offloads and separates security functions from the environment, operating system or hypervisor, automating application and flow discovery, delivering the advantages of a dedicated security appliance at the server edge.
ServiceNow integration means customers can track hardware associated with virtual machines/containers and troubleshoot application/network issues within or between servers. It also provides an automated and searchable inventory of all Pensando-enabled nodes within any infrastructure deployment.
Pensando asserts that, as data centre traffic moves to East/West, the traditional North/South firewall is no longer sufficient; the Pensando and Guardicore partnership is a response to this problem.
Pensando’s DSC is not supported by HPE’s cloud-based InfoSight predictive analytics system health service, based on source device and software telemetry. The HPE-Pensando relationship is roughly similar to the HPE Nebulon partnership, although that lacks a factory-installation option.
Nebulon’s smartCore and smartEdge products are intelligent cards, managed through the cloud, that offload various infrastructure tasks from their host CPU. HPE appears to be using the Pensando and Nebulon cards to answer customer requirements for its servers to be equipped with so-called SmartNICs.
We could not imagine an HPE server being fitted with both cards so, to this extent, the two suppliers are competing for HPE’s favours.
Seagate has announced its Exos CORVAULT, a 1.9PB SAS JBOD taking up just four rack units, providing 19PB in a full rack — if your floor can take the weight — and no-touch drive service for up to five years.
The CORVAULT provides up to 36GB/sec of bandwidth, Seagate characterising this as SAN-like performance, and is based on Seagate’s Exos 4U106 enclosure, with 106 x 3.5-inch drive bays supporting 18TB drives.
Seagate’s announcement says CORVAULT is designed to streamline data management and reduce human intervention for macro edge and data centre environments.
There are two drive controllers, each based on a sixth generation VelosCT-branded Seagate ASIC which provides ADAPT-branded erasure coding for data protection and ADR (Autonomous Drive Regeneration). Each controller has 4 x 12Gbit/sec SAS connectors and is located inside the enclosure. There is an Ethernet interface for management access, not data. The controllers do not use RISC-V cores.
Blocks & Files diagram.
ADAPT (Autonomic Distributed Allocation Protection Technology) distributes data across every drive and enables faster drive rebuilds by having all the existing drives contribute. The VelosCT chip is said to optimise all drive actuators in parallel, which helps speed both general and failed drive rebuild performance.
A Seagate spokesperson told us “a single drive failure in a CORVAULT ADAPT configuration would rebuild in approximately 20 per cent of the time required for an equivalent 16 drive RAID 6 array” and 5.3 hours is noted on an ADAPT data sheet.
All drives in an ADAPT disk group must be the same type, and in the same tier, but can have different capacities. The system automatically defaults to a target spare capacity that is equal to the sum of the two largest disks in the group. The target capacity is large enough to fully recover fault tolerance after loss of any two disks.
The ADR functionality comes into play if a drive fails. It can bypass (depopulate) errant disk drive heads so that the rest of the drive can continue operating. This does not use any of the IP that was developed by Xiotech for use in its ISE devices — sealed canisters of disk drives back in the 2009 era.
Seagate suggests a CORVAULT can continue operating for up to five years with no human physical access required. Even so, the disk drives, power supply units, cooling modules, side-plane expanders and IO modules are all hot-swappable.
CORVAULT offers “five nines” (99.999 per cent) reliability to help provide consistent high availability. A CORVAULT cannot scale past a single enclosure, and there are no plans to support shingled disk drives, NVMe drives or NVMe-oF access.
Systems will be available globally via qualified Seagate distributors in July.
Kioxia has added up-to-date OpenStack access control and open source integrations to its NVMe-oF KumoScale flash array software and increased its network access availability and bandwidth with a preview of multi-path networking support — an industry first for NVMe-oF storage over TCP/IP networks.
KumoScale software disaggregates NVMe SSD array storage from compute. A KumoScale system can scale to hundreds of storage nodes, which are certified servers running KumoScale software and fitted with Kioxia NVMe SSDs. Kioxia does not sell its own branded SSD arrays running KumoScale software. Kioxia has added NVMe-oF functions to the latest OpenStack Wallaby release and also integrated BGP (Border Gateway Protocol) support for multi-path networking to its v3.18 KumoScale release.
Joel Dedrick, VP and GM for networked storage software at Kioxia America, issued a quote: “We’re happy to contribute code and our considerable NVMe-oF technology expertise to the OpenStack community so that customers can benefit from the flexibility and scalability that NVMe-oF storage enables.”
Kioxia has:
Refactored the OpenStack os-brick NVMe-oF connector (nvmeof.py) and upgraded it to support more recent NVMe-oF protocols;
Contributed an enhancement to the OpenStack infrastructure Wallaby release connector that supports client-side replication via md-raid;
Contributed a KumoScale software Cinder driver that seamlessly integrates KumoScale storage backends into OpenStack environment.
This provides for replication across multiple failure domains with parallel writes, and so protection against single points of failure.
On the other side of the OpenStack relationship, Brian Rosmaita, OpenStack Cinder Project Team Lead (PTL) and a principal software engineer at Red Hat, said “These new features will enable the OpenStack community to take advantage of the fast-evolving NVMe-oF protocol. Other driver maintainers are already looking at leveraging the updated os-brick connector to offer NVMe-oF with TCP, so the Kioxia contribution, in addition to bringing KumoScale into the OpenStack family, is a real benefit to the entire OpenStack community of users.”
Networking
Storage arrays often use port channels — layer 2 in the OSI model — to connect to an IP network. Kioxia is moving KumoScale up to layer 3 and aiming to use IP routing instead. This means KumoScale systems can participate in so-called Clos networks. These have a multi-stage, circuit-switched design, with ingress points, mid-way switches and egress points.
In modern spine and leaf networking the spines are the mid-way switches and the leaves are both the ingress and egress points. In such a network all end hosts accessing the network can reach each and every other end host via a leaf-to-spine-to-leaf connection.
Blocks & Files diagram.
KumoScale will support BGP via an integration of the Free Range Routing (FFR) software and so provide reliable and dynamically reroutable L3 level multi-path network connectivity between client initiators and KumoScale storage targets. Overloaded or failed points or links in the network can have a new path routed around them to preserve connectivity and bandwidth.
The v3.18 KumoScale release also has improved install and upgrade processes, end-to-end security and reporting telemetry, and a sample reporting dashboard built on Prometheus logging, Loki telemetry and Grafana dashboard support.
Comment
Kioxia’s KumoScale competes with other suppliers of NVMe-oF-accessed flash arrays, including SW-only Excelero, and SW+HW suppliers Dell EMC, HPE, IBM, Infinidat, NetApp, Pure Storage, StorONE and VAST Data. The company is targeting cloud service suppliers and enterprise data centres, saying KumoScale offers native NVMe-oF performance at data centre scale, whatever that means.
A Hardware Compatibility List is said to exist on the “Deploying Storage Nodes” KumoScale webpage, but Kioxia’s link to it fails. We’re told customers can request it through Kioxia sales channels. Kioxia documentation refers to 2U x 24 slot configurations. These could hold 368.6TB of PCIe Gen 4-connected Kioxia CD6-R NVMe SSDs, enabling a 7.4PB rack — or 184PB capacity using 500 such servers.
KumoScale SW is sold via a perpetual licence and we have no information about KumoScale sales. We do know Kioxia has steadily extended KumoScale functionality since the software was first revealed in late 2017.
KumoScale history
Jun 2021 — v3.18 SW supports OpenStack Wallaby and BGP Routing.
Nov 2020 — PCIe Gen 4 added.
Oct 2020 — KumoScale faster than Ceph on block IO.
May 2020 — Thin provisioning support.
Apr 2020 — CSI-compliant Snapshot and clone functions added.
Sep 2019 — Supports Graphite and Prometheus technology.
Mar 2019 — Supports NVMe-TCP.
Mar 2018 — Toshiba announces KumoScale all-flash array SAN software supporting RoCE.
Aug 2017 — KumoScale software development revealed, running in a server and virtualizing a pool of disaggregated physical NVMe SSDs.
Newbie containerised app backup supplier Catalogic is adding a slew of new features, including virtual air-gapped ransomware protection in AWS and Azure to its CloudCasa BaaS offering.
CloudCasa, launched in November last year, is Catalogic’s main focus following the sale of its ECX copy data management software to IBM in May. It provides scalable data protection, with a free service tier, and disaster recovery service for cloud-native apps, and supports all leading Kubernetes distributions and managed services and cloud database services.
Catalogic COO Sathya Sankaran said in a statement: “We are pleased to introduce new features to CloudCasa, including the choice of public clouds and regions, to provide air-gapped backup storage for ransomware protection, and to provide a promotion for early registrants to protect up to 5TB for free.”
The added features are:
Support for Azure cloud storage along with Amazon S3 to store backups in [virtual] air-gapped storage for ransomware protection;
Choice of cloud storage regions to support data sovereignty and optimise data transfer costs;
Kubernetes Persistent Volume (PV) backups with 5TB free promotion for early access registrants;
Support for Amazon EBS persistent volumes for Amazon EKS in addition to CSI volumes;
Other service improvements such as tag-based selection for Amazon RDS backups and security and operations improvements.
Supplied diagram
The free service tier that lets users backup their K8s metadata and resource data to CloudCasa’s storage, and orchestrate Container Storage Interface (CSI) PV snapshots on their Kubernetes clusters. It also allows creation and management of AWS RDS database snapshots. A CloudCasa console enables users to scan their cloud-native application environment for recovery points and always be aware of how protected they are.
Comment
CloudCasa is complementary to Catalogic’s open-source KubeDR software, catering to some overlapping use cases, but delivering additional protection for CSI persistent volumes (PVs).
Catalogic is developing premium CloudCasa service offerings that it says will be coming soon, with additional features such as being able to move backup data from a user’s cloud off onto its cloud. In the third 2021 quarter we should see a premium tier available for PV backups, along with some other object storage and marketplaces that will be supported.
Road map items for the future include additional application hooks, more relational database service support, and a “bring your own keys model” for encryption.
The suppliers competing for cloud-native app data backup include Clumio, Commvault, Dell EMC PowerProtect, Druva, HYCU, NetApp Astra, Pure’s Portworx, Robin.IO, Rook, StorageOS, Veeam’s Kasten and others. Several provision storage and migrate cloud-native apps as well as protect data. It’s likely, we think, that CloudCasa will also provide cloud-native app migration, and hence a form of disaster recovery, but not necessarily storage provisioning.
Having a free storage tier should help CloudCasa gain new users.
HPE has added on-premises Azure Stack HCI and Microsoft SQL Server availability to its GreenLake offer, complementing existing SimpliVity and Nimble dHCI hyperconverged systems support in GreenLake, and Nutanix as well.
GreenLake is HPE’s subscription-based arrangement for supplying hardware and software as a cloud-like service. Azure Stack HCI is Microsoft’s Azure cloud service embodied in on-premises hardware, supplied by Microsoft partners such as HPE, with a backend link to the Azure public cloud. Azure Stack HCI is delivered as an Azure service on GreenLake to run in data centres, remote office offices and other edge locations.
Keith White, HPE GreenLake Cloud Services GM, offered this statement: “The world is becoming hybrid and that’s why we are so excited about this collaboration with Microsoft, especially as we see significant growth in the Hyperconverged Infrastructure area. By combining Microsoft Azure Stack HCI with offerings like the HPE GreenLake … customers benefit from a unified, automated experience.”
HPE says GreenLake Azure Stack HCI and SQL Server have pay-per-use pricing, point-and-click self-service, scalability, cloud-like flexibility and the security of dedicated, on-premises IT at 30 to 40 per cent TCO savings with no need for over-provisioning.
Customers can control their services through the Azure portal in the same way that they manage their Azure public cloud services.
On the Microsoft side Roanne Sones, Corporate VP for Microsoft Azure Edge + Platform, said “The shared work to integrate HPE GreenLake with Microsoft offerings gives our customers a managed services solution that offers simplicity, flexibility, and the economics that work for their environments, plus the benefit of ongoing security expertise from Azure and HPE.”
HPE GreenLake with the HPE Validated System for Microsoft Azure Stack HCI and SQL Server is available immediately, with an Integrated System becoming available later this year.
A cloudy theme this week, with Tesla camera footage stored in a cloud and cloud service provider OVH offering a Kubernetes service. We also have Komprise enabling its management dashboard facility to look at multiple sites.
It’s a peculiar week — there’s almost no backup news. None.
Storj and Tesla
Storj has a way for Tesla owners to store and share their car camera footage.
The company produces decentralised cloud object storage (DCS) software, previously known as Tardigrade, which uses MinIO and is S3-compatible.
Tesla vehicles come with built-in sensors and cameras that can record Dashcam or Sentry Mode video. Dashcam footage is captured when a Tesla driver is operating their vehicle. Sentry Mode video is captured when the vehicle is parked, and someone approaches or movement is detected. Both modes must be manually enabled to record video, which is primarily stored on a USB drive inside the vehicle. The data must then be manually transferred to another device if the driver wishes to view, share, or store the video long-term.
Tesla auto.
Storj has integrated its DCS software with Rclone open source software to enable Tesla car owners to store the car-generated footage in DCS. Storj DCS has a free tier so that Tesla owners can upload up to 150GB of footage to the service at no cost.
When video is uploaded to Storj DCS, it is encrypted, encoded for redundancy, and stored across a diverse set of Nodes worldwide. Once the footage is uploaded, Tesla drivers can share the video with friends, family, or other interested parties with just a click of a button through Storj’s object browser using Storj Share functionality. From there, video can be streamed or downloaded to other devices.
JT Olio, Storj CTO, said “This integration not only enables Tesla owners to privately and securely manage their footage, but it also showcases how easy it is to integrate Storj DCS with open source software and how well the product can upload and stream data to locations all around the globe.”
A Storj blog by Infrastructure Engineering Manager Krista Spriggs provides more details.
Komprise extends Intelligent Data Manager
File lifecycle manager Komprise has added multi-site control to its Intelligent Data Manager with a v4.0 release.
Customers and CSPs can get a consolidated view across sites with this data management as a service offering.
Kumar Goswami, co-founder and CEO of Komprise, said “Aside from the cloud, many organisations are geographically dispersed or have security protocols and organisational boundaries that require them to localise where data movement occurs. Komprise brings granular flexibility and control with new multi-site data management capabilities.”
The 4.0 release introduces:
Centralised management across multiple sites — Set up multiple sites, each with their own storage and separate data management policies and activities. All sites are managed from a single, on-premises or cloud-based Komprise Director. This enables centralised management for cost and performance optimisation through a single dashboard, providing a consolidated view across all sites in the entire deployment.
Localised policy management and execution — Each site has its own Komprise Observer virtual machines to connect to all the storage and clouds for that site and perform the associated data-management tasks. Policies and execution are controlled locally within each site, according to distinct departmental needs and security requirements.
Global search and deep analytics across sites and clouds — With a centralised dashboard, enterprises now have a single searchable virtual data lake of all unstructured data that can be shared across the enterprise. Customers can develop custom queries to identify specific data sets such as ex-employee data, or data belonging to a project that might be strewn across data centres and clouds.
Krishna Subramanian, Komprise President and COO, told us: “[We] are now providing a global multi-site Director and dashboard across multiple Komprise Directors along with a Global Data Index so you can search and find across all files and objects across all sites. This is significant engineering because the scale can be hundreds of billions of files and we have a fast, efficient, elastic index to do this.”
“Just last weekend, we had a customer who ran into a compliance issue and had to quickly purge ex-employee data globally and was able to use this capability to do so.”
Komprise customers will appreciate the opportunity to manage all their Komprise instances through a centralised single dashboard.
OVH Cloud gets into managing Kubernetes
OVH Cloud, one of whose data centres was struck by fire earlier this year, has made a managed Kubernetes Service available.
It manages both the underlying hardware and the Kubernetes software stack, keeping it up to date with critical updates associated with bugs and security patches, and providing load-balancer and pod auto-scaling capabilities. Developers can start small with entry-level instances and then upgrade to more powerful instances when moving to larger-scale production.
OVH Strasbourg data centre showing fire damage.
OVH expects the overwhelming majority of production applications to be-cloud native within the next 18–24 months.
Jeffrey Gregor, General Manager, OVHcloud US, said “The ability to deploy, manage, and scale consistent and secure-by-default Kubernetes instances on demand defines what it means to be cloud-native. The power of our managed Kubernetes in our Public Cloud environment allows teams to iterate faster, automate more aggressively, and exploit modern application-lifecycle paradigms.”
The OVH service includes:
CNCF-certified managed Kubernetes offer;
Persistent volumes — Add persistent volumes to worker nodes, ensuring durability for application data;
Simple, easy-to-use interface — Configure, add and delete nodes via the OVHcloud Control Panel;
Role-based access control — Deploy services with specific levels of access;
Multiple versions and upgrades — Choose from any of the latest three stable versions of Kubernetes when creating a cluster;
Integration with OVHcloud’s vRack — Flexibility to expose cluster(s) services publicly, or through the OVHcloud private network.
Data migrator Datadobi says its product suite can be sold on US government GSA IT 70 contracts, as it has been approved to be on Climb Channel Solutions’, a subsidiary of Wayside Technology Group, Inc. That gives Datadobi access to US federal, state, and local government business.
HPE has added Qumulo’s File Data Platform to its list of GreenLake Cloud Services (metered subscription) offerings. Qumulo is expanding its operations to include the Asia-Pacific region in partnership with HPE. Qumulo’s scalable file system software runs on HPE Apollo and ProLiant servers.
HPE is certified for the SplunkOperator for Kubernetes, which allows users to deploy and manage Splunk Enterprise in a Kubernetes infrastructure. The HPE Ezmeral Container Platform works with Splunk to create a single datastore that leverages open-source Kubernetes and S3, and is available as a fully managed as-a-service solution from HPE GreenLake.
AI pipeline supplier Iguazio is partnering Boston Limited, which supplies high-performance, mission-critical server and storage kit, to sell into the data science market. Boston Limited will offer data centre hardware and technical services, while Iguazio provides its data science platform, said to save time and cost on getting AI to production.
Micro Focus’s Vertica Eon Accelerator is available under an early access program, and Micro Focus has bought Full360 — its strategic partner which supplies Vertica as a service. Vertica is a massively scalable analytics platform using database machine learning. The Eon Accelerator provides analytics-as-a-service.
Smart infrastructure SaaS supplier Nebulon has added support for ProLiant servers using gen-3 AMD EPYC processors to its smartCore and smartEdge products.
Next Pathway, an automated cloud migration company, announced it has extended Snowpark, Snowflake’s developer tool, to automatically migrate and run ETLs (Extract, Transform and Load) and ETL pipelines (data workflows) natively in Snowflake. Snowpark, using Next Pathway’s language, enables ETL data pipelines to be executed in Snowflake’s single, governed core data platform, shortening migration time, minimising the technical footprint and reducing cost due to fewer systems in a customer’s architecture.
Redstor has added AI and analytics to its smart data management service and offers support for OneNote, Kubernetes and Salesforce. It has enhanced its RedApp, a purpose-built application for partners, enabling them to deliver scalable protection and management of customer data, more easily, from a single application.
Testing house SANBlaze has doubled the tests in its Zoned Namespaces (ZNS) test suite for NVMe devices from 105 to over 200, and added more than 200 negative test cases. It claims it is the only ZNS testing vendor to provide all-inclusive comprehensive ZNS testing. ZNS technology allows for dividing physical storage media into logical zones governed by the NVMe ZNS specification and is part of the recently ratified NVMe V2.0 spec.
SMART Modular has added T5EN PCIe/NVMe M.2 2280 and U.2 flash drives for aerospace, defence and industrial applications that require durable, rugged, and secure memory storage. The drives offers capacities up to 8TB (M.2 up to 4TB) of TLC NAND, and PCIe Gen 3×4 performance using 3D TLC flash with pseudo single-level cell (pSLC) support.
People movers
HPC parallel file system supplier Panasas has hired Fred Vasofsky as Managing Director of its Vertical Solutions business. The upstanding Fred comes from being the GM for WekaIO’s federal business, and was at WD, SanDisk, EMC and StorageTek before that.
Storage stack rewriting software supplier StorONE has appointed Bill Cordero to be VP Worldwide Channels. He’ll launch the new PartnerONE channel program and expand StorONE’s global reach. Cordero was VP Worldwide Channel Sales at Yellowbrick Data for eight months and ran channel sales at Diamanti, Blue Medora and Upguard before that.
Courtesy of IDC, Nutanix has been named number one in the software-defined storage controller market, beating VMware into second place.
IDC recently redesigned its software tracking reports, producing a single and semi-annual Software-Defined Infrastructure Tracker report, with three segments: compute, storage controller, and networking.
The storage segment looks at storage software stacks delivering storage services in conjunction with COTS hardware to create a complete storage system. IDC states: “SDS-CS is the core software that virtualises and pools the storage resources across different servers that comprise the SDS. SDS-CS provides for data persistence, a set of data services (snapshots, replication, etc.), and a method of data organisation (block, file, and/or object) along with one or more defined access methods (block, file, and/or object). Another way to describe an SDS system is that it has modular building blocks, uses industry-standard hardware platforms, and typically employs distributed, scale-out architecture.”
Nutanix has revealed details of suppliers’ rankings in that segment in a blog by Tonya Chin, its SVP for Corporate Marketing and Investor Relations. Yes, the topic is that important.
Chin starts out by writing, “The hyperconverged infrastructure (HCI) market has been on a fast journey over the past few years.” It’s moved from being hardware-centric to software-centric.
Then she writes, “I’m ecstatic to share [that] Nutanix is clearly the market leader with 25 per cent of the market, well ahead of the nearest competitor at 15 per cent.”
Here is a pie chart of the IDC numbers that Nutanix has kindly revealed:
Blocks & Files chart using Nutanix-sourced IDC numbers.
Chin burbles, “We’re deeply grateful to IDC for realising that there was a gaping hole in the market when it came to HCI software, and bringing this project to completion while giving Nutanix the credit that we long knew it was due.”
She writes, “IDC asserts that SDI (Software-Defined Infrastructure) is a better proxy for where HCI is headed as it encompasses all three infrastructure software elements: compute, storage, and networking.”
Comment
Why is VMware so far behind Nutanix in the storage controller segment? We would have thought VSAN sales far outstripped Nutanix SW sales. Another oddity to our eyes is that NetApp has seven per cent of the market while HPE, with SimpliVity and Nimble dHCI products is, we suppose, either absent altogether or lumped into the Others category.
IBM-Red Hat is given ten per cent of the market, but its HCI market presence is generally perceived to be as near zero as makes no difference.
Until we get to see more details of the IDC tracker document, we are left scratching our heads in puzzlement at this apparently odd take on the market.
GigaOm has has added separate SMB and Enterprise hybrid cloud data protection supplier evaluation docs in its Radar series as the fast-developing market splits into two groups of products.
This separation follows on from the February 2020 unified hybrid cloud data protection Radar document, which evaluated 14 vendors: Actifio, Acronis, Clumio, Cohesity, Commvault, Delphix, Druva, HYCU, IBM, Rubrik, StorageCraft, Veeam, Veritas, and Zerto.
The GigaOM analysts have split the original document in two because the two markets have different characteristics. The enterprise market “is trending toward additional services built on top of data protection, and these services are becoming instrumental in collecting and consolidating data across the entire organisation.”
The SMB (Small and Medium Business) market, on the other hand, “is trending toward solutions with a broad feature set that can address the data protection requirements of hybrid and cloud workloads as well as SaaS applications.”
These Radar documents evaluate suppliers on a range of characteristics and then diagrammatically position them on a quasi-radar screen with two axes: feature vs platform play, and maturity vs innovation. The screen is separated into three concentric circles around a bullseye. These circles, from the outside in, represent positions for new entrants, challengers and then leaders.
Vendors are located with square dots and an arrow showing their direction and speed of progress. The nearer they are to the middle of the diagram the better as far as the analysts are concerned.
SMB Radar
Here is the 2020 unified radar diagram and, for comparison, the updated 2021 SMB diagram;
2020 unified hybrid cloud data protection radar diagram on left and 2021 SMB hybrid cloud data protection radar on right.
There are 14 vendors in the 2021 SMB radar: Acronis, Arcserve, Barracuda, Carbonite, Clumio, Cohesity, Commvault, Dell, Druva, IBM, NAKIVO, StorageCraft, Veeam and Zerto. It’s the same number as before but Actifio, Delphix, HYCU, Rubrik and Veritas are no longer included while Arcserve, Barracuda, Carbonite, Dell, and NAKIVO have been added. The included vendors are quite spread around the rings with just Veeam and Commvault identified as leaders. Cohesity was a leader but is now a challenger and moving into the inner leaders’ ring.
Carbonite, Clumio and StorageCraft are classed as new entrants. The remaining companies are all challengers, with Acronis set to enter the leaders’ ring.
The market is said to be in active development with all suppliers “looking at ways to disrupt traditional backup models and differentiate, while ensuring they can cover most data protection use cases.”
Enterprise
In the 2021 enterprise radar diagram we see a reduced list of 13 vendors: Acronis, Arcserve, Bacula, Clumio, Cohesity, Commvault, Dell, Druva, IBM, Rubrik, Veeam, Veritas and Zerto. Vendors that have been withdrawn from the original list are Actifio, Delphix, HYCU and StorageCraft. Arcserve, Bacula and Dell have been added.
2020 unified hybrid cloud data protection radar on left and 2021 enterprise hybrid cloud data protection radar on right.
There is a noticeable concentration of vendors on the right-hand side of the diagram, favouring platform plays rather than features. There are three outside this semi-circle: Clumio, Druva and Zerto — with Druva and Zerto set to move across into the platform play section, leaving Clumio as the sole feature player.
The four fastest-moving suppliers; Cohesity, Commvault, Veeam and Zerto, are all in the innovation semi-circle.
The analysts summarise the enterprise market as being active “with both startups and incumbents looking at ways to disrupt traditional backup models, as well as increase the value of the protected data by adding data management and data reuse capabilities to their solutions.”
Overall, the enterprise market suppliers tend to be classed more as innovative platform players while the SMB suppliers are more spread between feature and platform plays on the one hand, and mature and innovative vendors on the other. All the fastest-moving SMB-focussed vendors are innovative though.
GigaOM subscribers can read a separate “Key Criteria for Evaluating Hybrid Cloud Data Protection v 2.0” document to gain more insight into the analysts’ thinking.
Since raising $20m in January, NAS supplier OpenDrives has increased its headcount by 70 per cent and now expects to grow its 2021 revenues 200 per cent year on year as it expands into new markets.
Its business involves selling scale-out and high-performance NAS systems to media and entertainment companies. OpenDrives has hundreds of customers — including HBO, CBS Sports, Spotify and Sony Interactive Entertainment. It’s also into healthcare diagnostics, for customers including Steadman Clinic, and advertising with Deutsch.
David Buss.
CEO David Buss, a retired Vice-Admiral and not your usual tech startup boss, blew OpenDrives’ trumpet in a statement: ”OpenDrives’s continued momentum would not be possible without adding talent to our already strong team. Our new and existing team members are the heart and soul of our business. As the company has grown, so has the number of industries we’re able to serve. I’m thrilled to bring on so many talented people to our team this year.”
There’s a fair number of sales people being hired, with two new channel account managers — Mike Novoselac (formerly Dell) and Eric Zappia (formerly Quantum) — plus four regional territory managers — Daniel Hatch and Jason Matousek (both formerly Quantum), Craig Alger (formerly Dell EMC) and Chesed Patt (formerly Qumulo). These people will be working with more than 65 channel partners.
Buss has also appointed Ethan Cole (formerly AT&T) as Director of Product Management and Tim Chunn (formerly NetApp) as Director of Business Development.
He wants the company to expand into supplying video management software and enter the high-performance computing, AI and general enterprise markets. New hire Tim Chunn will oversee the company’s expansion strategy and help expand market reach.
Existing NAS suppliers should prepare to say hello to this hot new kid on block.
The Veeamster, now fully US-owned and -managed, is back in the US federal business after trying and failing by buying and then selling N2WS. It has set up a subsidiary called Veeam Government Solutions, Inc.
Veeam bought N2WS in January 2018, but one customer became quite unhappy about that — namely the US government, which didn’t want its backups provided by a Russian-led and -controlled company. Consequently, Veeam sold N2WS back to its original founders in the third quarter of 2019.
Since then, US private equity business Insight Partners bought Veeam in January 2020 for about $5B, with Veeam becoming a US-owned and -managed company, and thus eligible to bid for US government contracts. Since then it’s been building up its federal business.
Veeam CEO Bill Largent said in statement: “To help accelerate our federal business, we hired Earl Matthews last summer to develop and lead our VGS team, as well as join the Veeam executive staff.”
Earl Matthews.
Matthews is “An American government official and attorney who has held senior positions within the Department of the Army and at the White House. Earl’s expertise and impressive, extensive background in the public sector will accelerate our ability to reliably serve the current and future Modern Data Protection needs of federal customers during critical times.”
A Matthews statement read: “For the past decade, Veeam has delivered innovative data protection and data management solutions to government. We founded VGS to be even more hyper-focused on the highest level of security and nimbler in support of federal agency needs. We’re committed to continually finding better ways to serve the government, solve its biggest data challenges, and protect the information that is a strategic asset for everything government organisations do.”
VGS
Veeam Government Solutions (VGS) has its own board of directors and an independent advisory board.
VGS has US Navy DADMS, NIST – 1800-11, SOCOM Approved Products List (APL), and Section 508 certifications relevant to the US federal market. It says it has been sponsored by DISA and submitted for testing for the DoDIN APL, and is working to secure additional certifications important to federal organisations.
More bricks in its US federal business wall include announcing a DOD ESI Blanket Purchase Agreement with EC America, a subsidiary of immixGroup, an Arrow Electronics company, and a partnership with CACI, Inc. Federal. This is aimed to ensure a secure supply chain and provide auditable and traceable processes in line with the President’s May 12, 2021 Executive Order on Improving the Nation’s Cybersecurity provisions.
VGS has more than 1200 US government customers and says it’s focused on the backup and protection of mission-critical data for the US Department of Defense, Civilian Agencies, Native American Tribes, the Intelligence Community and Federal System Integrators.
Comment
We’re seeing the emergence of a superset of massive data protection companies — Acronis, Cohesity, Commvault, Dell Technologies, Rubrik, Veeam, and Veritas, with Dell probably the largest of them all in revenue terms. Our picture is fuzzy and we don’t have revenue numbers for all these suppliers.
Dell Technologies is public and IDC implied it had a $1.5B data replication and protection revenue run rate in 2020, with Veritas enjoying a $1B run rate.
Commvault is the other public company in the list and reported $723.5M in revenues for its fiscal 2021. Any IPOs, such as potential ones from Cohesity and Rubrik, will providing more visibility into what’s going on in what’s becoming a billion-dollar data protection business suppliers’ club.
Cohesity has announced another record quarter of results. It’s selling to more customers, who are spending more, via partners who are bringing in more business. Subscriptions are soaring as well.
The company sells data protection and management software for on-premises and hybrid cloud workloads, and is moving a software-as-a-service product offering with customers paying through subscriptions. Founded in 2013, it was valued at $3.7B in March and has taken in $660M in funding, with the most recent round bringing in $250M last year.
Mohit Aron.
CEO and founder Mohit Aron issued a statement: “Our record-breaking results show that more businesses are trusting Cohesity to manage their data in a world in which hybrid cloud is the norm, ransomware attacks are soaring, and the need to derive value from data has never been greater.
The business momentum is continuing following its last quarterly results. Subscriptions are growing, customer count is increasing, more partners are selling its software, and customer lifetime spend is increasing. We’ve tabulated the various bullet points from last quarter’s announcement and the latest one to enable as direct a comparison as possible:
The net expansion rate increase number is the rate of expansion net of churn from existing customers over the last year.
Comment
What we see here is, in our view, a company heading towards an IPO. It is distancing itself from another massively funded data protection and management startup, Rubrik, according to a GigaOm analysis of suppliers competing in the hybrid cloud enterprise data protection market.
GigaOm identifies Cohesity, Commvault and Veeam as the three leading vendors, with better products and faster progress.
The GigaOm analysis identifies a missing Kubernetes-orchestrated container data protection offering in Cohesity’s BaaS product set; it’s present in the enterprise product. Our understanding is that this gap will be filled by the end of the year.
Rubrik faced some exec churn earlier in the year. Since then it has signed up NetApp as a reseller and has also announced protection for containers in the Kubernetes environment. Both moves position it well for the future and its own potential IPO.
French supplier Kalray has launched a K200-LP acceleration card for accessing NVMe storage.
The card uses fabless semi-conductor company Kalray’s in-house-developed VLIW (Very Long Instruction Word) Massively Parallel Processing Array (MPPA) processor, with 80 cores. A PCIe Gen 4 link is used to connect the card to a host server CPU. The card has dual 100Gbit/s Ethernet ports for the NVMe links to SSDs.
Kalray K200-LP.
K200-LP features.
Eric Baissus, President and CEO of Kalray, provided a prepared quote: “This [MPPA] family of Ethernet/PCIe acceleration cards are natively capable of managing multiple workloads with no bottleneck to enable smarter, more efficient, and energy‐wise applications on Cloud and Edge data centres. To address the new generation of storage array solutions for Cloud and Edge, K200‐LP is a game changer solution in terms of performance per Watt and per dollar.”
Kalray says the K200-LP card runs all the critical functions of a disaggregated storage appliance, including NVMe and NVMe‐over‐Fabric protocols using NVMe SSDs. The card can deliver two million IOPS, with 12GB/sec bandwidth for both RoCE and NVMe/TCP. The actual latency added by the card is between 30μs for RoCE and 50μs for NVMe/TCP.
The firm claims it has an outstanding performance-per-Watt rating, without providing actual numbers. Software functions such as encryption, erasure coding, deduplication or other low-level storage tasks can be developed for the card using Kalray’s AccessCore Storage software suite.
The K200-LP is an example of SmartNIC technology and Kalray will be competing with AMD-Xilinx, Nvidia with BlueField and VMware’s Project Monterey with ESX running on the BlueField NIC, also Marvell, Intel, and others.
Kalray’s K200‐L smart storage card and associated tools are available now, the card being manufactured by Taiwanese EMS Wistron. Read a technical brief on MPPA here (registration and documentation choice required).
