Your summer storage picnic

We have spread out a tasty array of storage summer dishes on our white linen cloth. Take a deck chair and feast your eyes on Attala’s smart NIC nuts, Hitachi Vantara chips, Micron savoury munchies, Pivot3 crackers. Let’s finish with some fruity desserts; Acronis apples, Cobalt Iron cookies, Hammerspace dates, and IBM big blueberries.

Smart Attala NICS prevent CPU core robbery

Attala has made hardware-accelerated Smart NICS for NVMe-over Fabrics (NVMe-oF) available for use by cloud service providers. The company said the technology prevents mediocre performance and CPU core robbery.

NVMe-oF supplies RDMA-based access to NVMe SSDs. This enables low latency access to remote SSDs, adding single-digit microseconds to the latency involved in a direct-attached NVMe drive access.

A local NVME drive access can take, say 100 microsecs. A remote NVMe drive access, using Fibre Channel or iSCSI can take 1,000 microsecs or more.

By using NVME-oF – with its RDMA – the remote NVME drive access takes about 110 microsecs. This is a massive speed up and more or less the same as accessing a local, direct-attached NVMe drive.

The Storage Smart NIC uses FPGA hardware, SR-IOV, peer-to-peer PCIe, and RDMA to hook up remote NVMe drives to a server.

Taufik Ma, CEO of Attala Systems said: “As cloud service providers started to transition from hard drives and SAS/SATA SSDs to NVMe SSDs, they found that their traditional software-based architectures could not support NVMe levels of performance. And to achieve even mediocre performance required the reservation of several CPU cores worth of compute resources.”

He said the Storage Smart NIC “eliminates costly software cycles that limit the full value of NVMe SSD performance and that lower cloud service provider revenue due to reserved compute resources that rob from tenant use.”

Hitachi Vantara improves Pentaho data pipeline and visibility

Hitachi Vantara’s 8.3 update of its Pentaho data integration and analytics software has better data pipeline and data governance capabilities plus  expanded multi-cloud. New features include:

Improved drag and drop data pipeline capabilities to access and blend data that’s difficult to access  

  • A connector to SAP for drag and drop blending, enriching and offloading data from SAP ERP and Business Warehouse,
  • Integration with Amazon Kinesis for real-time data capability in an AWS environment, for ingesting and processing streaming data in a visual environment as opposed to writing code.

Data visibility for improved governance   

  • Improved integration with Hitachi Content Platform (HCP).
  • IBM Information Governance Catalog (IGC) Integration, initially offered in beta with restricted distribution. Customers can view lineage data from Pentaho within IBM’s IGC to reduce the amount of manual effort required to govern data.  
  • Streaming data from popular protocols such as AMQP, JMS, Kafka and MQTT. 

Expanded multicloud support

  • AWS Redshift bulk load capability,
  • Snowflake connectivity allowing blending, enrichment and analysis of Snowflake data along with other data sources,
  • Customers can access data from existing supported cloud platforms, including AWS and Google Cloud, in addition to Snowflake. 

Thanks for the memory, Micron

Micron discussed its DRAM roadmap during an investor call earlier this month.

The company fabricates 16Gb DDR4 chips using a second generation 10nm-class process known as 1Y nm. The gen 1 10nm-class process is called 1X nm. A third generation process is coming called 1Z nm.

A 10 nm-class process has a size range between 19nm and 10nm.

We understand that 1X nm refers to 18nm, while 1Y is in the 15-16nm area. 1Z would be smaller still; say 14 nm. The smaller the process size, the more DRAM chips can be carved out from a wafer, lowering cost if the yield is high enough and any extra process steps involved are affordable.

Micron plans to follow this with three more 10nm-class processes; 1α, 1β, and 1y; alpha, beta and gamma. These would also increase wafer yield by using progressively small cell sizes. For example, 13 nm, 12 nm and 10-11nm.

There is also a die-shrink version 1X; 1Xs nm. That gives Micron 7 separate 10nm-class processes. The post 1X nm processes use deep ultraviolet lithography (DUVL) and extreme ultraviolet lithography (EUV L) may be required after that.

EUV could facilitate a move to even smaller processes; 7nm and possibly 5nm.

Read an Anandtech article for a thorough discussion of DRAM process size.

Pivot3 hooks up with VMware Site Recovery Manager

HCI software vendor Pivots3 has completed integration with VMware’s new version of Site Recovery Manager (SRM) 8.2.

Mike Koponen, senior director of product marketing at Pivot3, said: “We are now extending Pivot3’s policy-based management approach to organisations implementing DR strategies with containerised VMware Site Recovery Manager.”

Pivot3 storage replication is now fully integrated with VMware Photon OS, a new operating system purpose-built for VMware infrastructures. Site Recovery Manager 8.2 is deployed as a Photon OS-based virtual appliance and can be installed directly from vCenter Server. The new Photon OS-based deployment model simplifies setup of VMware Site Recovery Manager components and disaster recovery (DR) plans.

With this, Pivot3 can automate and orchestrate non-disruptive testing and execution of recovery plans to minimise downtime in case of disasters.

Blocks & Files asked Pivot3 a couple of questions.

B&F; Did Pivot3 provide DR through VMware Site Recovery Manager before v8.2 of that product?

Pivot3; Yes, Pivot3 has had a Storage Replication Adapter (SRA) in place for previous versions of SRM as well. This integration required doing a containerised version of our SRM compatible with this new version of SRM.

B&F; Does it provide DR any other way?

Pivot3; Yes, Pivot3 also has Technology Alliance relationships with Zerto, Veeam, CloudEndure (now part of AWS) and can provide customer options based on their needs/preferences.

Storage snacks

Acronis Backup has added backup and recovery for SAP HANA, via scripting and with verification. It said this is the first step toward greater connectivity between Acronis’s cyber-protection services and SAP HANA’s data management capabilities.

Backup supplier Cobalt Iron has received a patent for its techniques related to electronic discovery when searching for data in backup storage. It involves extracting metadata from the backup data set and finding the right subset within it. Users of Cobalt Iron’s Adaptive Data Protection platform can fulfil e-discovery requests more quickly and cost-effectively, with verifiable data discipline.Ethos Technology, a UK-based value-add distributor has announced it will be offering the latest systems and capabilities from Pure Storage.

The Hammerspace data-as-a-service offering is available on the AWS Marketplace. It is now possible for DevOps, application owners, and data operations to consume data-as-a-service via on-premises, cloud, and container environments.

IBM has a published a Spectrum Discover Redbook entitled “Metadata Management for Deep Insight of Unstructured Storage”. This is timely as IBM recently updated Spectrum Discover with support for heterogeneous environments.

Western Digital’s SANdisk has announced a 2TB Extreme Pro M.2 format NVME SSD. It uses 64-layer TLC (3 bits/cell) flash and complements the existing 500GB and 1TB products.

The IT management division of Zoho Corporation, today announced that ManageEngine ADManager Plus, its Active Directory (AD) management and reporting solution, can now back up and recover all AD objects, including users, OUs, GPOs, groups and their members. Apart from creating, modifying and de-provisioning objects in AD, administrators can now back up objects belonging to multiple domains and recover deleted objects, from a single console.