Tag:

NVMe

HPE PRIMERA

by Roger Lund October 22, 2019

written by Roger Lund

HPE PRIMERA

The World’s Most Intelligent storage

Coming to you live from California at the Tech Field Day event, #TFDx, @HPE_Storage , #NthSymp2019

Presenting Alex Veprinsky

Constant Challenges for I. T.

Complex Infrastructure

Constant Fire-Fighting
Fragmented Data Silos

Introducing HPE PRIMERA

HPE PRIMERA Foundation elements

Powerful New Architecture

Build for Intelligence
Redefined Management

Architected for speed and intelligence

Flexible and Powerful Architure

All-Active

All-active and lock-less with no single point of failure

Memory-Driven

Optimized for SCM and NVMe

Intelligent

Predicts and prevents disruptions that prevent outages

Manage performance at Scale

Today vs Ideal

Latency at scale as IOPS increase, SCSI has challenges. NVMe comes into play.

Understanding Storage

Servers > Fabric > Nodes > Backend.

Built for NVME, Optimized for SCM.

CPU , CPU

ASIC, ASIC, ASIC,

no cpu on ASIC’s

Latency response is important.

Cache hits and IO deliver to the backend have a IOPS impact.

PRIMERA delivers most I/O in .03 MS , without NVMe

Predictable, Low Latency even at scale.

All-Active
Memory-driven
Lockless

1.5m IOPS at 44GB/S

NVME ROADMAP

Current support for Front End NVMe with Fiber Support. Example Linux. NVMe with roce on 200G as a example,

The Service-Centric OS Architecture

Monolithic

Moving away from this

Service-centric

To this. Software features will be service deliverable as a example.

Intgelligance Changes Everything.

On-Array

Tune the sytem in real-time

On-Prem

View statturation, trouble shooting at the datacenter level

Cloud

Embedded, Self-optimizing AI Engine

Anomaly detection
Hotspot detection
Workload prediction

“what if” scenarios
Estimate saturation
Forecast system load

PRIMERA AI Pipeline

Collect > Learn > Identify, score > Correlate > Act.

AI/ML Offers Multiple Insights into Data Center

Topology insights

Performance insights
Workload insights

Example

Performance insights

Key Performance Indicators
Saturation
Problem space

Performance

Allows the identification in real time, to shorten resolution.

Workload Insights

Allows you to compare workloads with other customers.

Whats Next?

Next-Step AI-RUN Storage

Traditional Storage System vs intelligent Storage System.

Using AI Model to Optimize Tunevv

Impact of tunevv
Forcast
AI in action

Allows identification of customer workload patterns.

Upgrading has never been so easy fo High-End Storage.

Efficient scaling

Simple refresh
Intelligent updates.

Scale up, Scale out, Scale up and out – non-disruptively

Either and or flexibility

HPE PRIMERA is truly timeless

All inclusive licensing
Free controller upgrade options
Availability & data reduction guarantee

Protect your applications

Resilient

Protected
Predictive

Powerful Native Replication

peer persistence
synchronous

periodic

Secure Quorum witness

peer persistence – Zero RPO RTT latency < 5ms Automatic failover.

Sync, IP over 10Gbe or FC.

Synchronous – Zero PRO RTT latency < 5 ms manual failover

Async periodic 5 + min interval RTT latency < 1s manual failover.

Heterogeneous Data protection and mobility

HPE Recovery manager central between platforms.

This is part of the Tech Field Day Exclusive with HPE Storage 2019

https://techfieldday.com/event/tfdxhpe19/

Roger Lund

October 22, 2019 1 comment

Datacenter Storage

NVMe and Storage Spaces Direct are a match made in SQL heaven

by Philip Sellers December 20, 2018

written by Philip Sellers

SQL Server is ready to take a ride directly on Microsoft’s Storage Spaces Direct (S2D), its own software defined storage (SDS) platform. It is not new news – Microsoft has been talking about SQL on S2D since 2015 – but changes in the NVMe product offerings are changing the possibilities of very high performance from SDS architectures.

NVMe has evolved and with the acceleration of new technologies like Intel’s Optane storage class memory drives and expanding capacities like the Intel P4800 series, building highly-performant, software-defined storage is getting really easy. That provides another opportunity for acceleration beyond what current NVMe drives can do today.

While Microsoft has been busy building Storage Space Direct as a hyper-converged, software-defined storage platform for its Hyper-V hypervisor – it also built a robust replicating file system suitable for SQL Server and even highly-available File Services platforms.

Storage vendors are starting to introduce Optane storage class memory as a caching layer, with HPE announcing this for its 3PAR and Nimble lines last week. Pure Storage and NetApp are also busy working with storage class memory for their products and integrating NVMe into already shipping products.

Microsoft, for its part, has been busy building out support for its Storage Spaces Direct feature, which pools direct attached storage into a software SAN within Windows 2016 servers. And about 2 years ago, during Microsoft Ignite 2016, Microsoft announced support for SQL Server 2016 on Storage Spaces Direct. While it didn’t click initially, the use of NVMe and even Storage Class Memory with Storage Spaces direct may drive adoption for uses other than Hyper-V in many environments.

Storage Spaces Direct offers benefits and flexibility that traditional storage arrays do not. While storage vendors are busy trying to establish ways to handle NVMe on fabric, Storage Spaces Direct and its use of RDMA have already solved some of these issues.

NVMe bypasses the traditional SCSI protocol for storage and uses its own protocol. In a lot of use cases, it is not ready for prime time because it lacks the robust features like multipathing and bus sharing. But under an SDS solution, NVMe can shine quite well.

The rise of high speed NVMe drives has given way to all sort of inventive ways to use them. Hyperconverged architectures have been using NVMe for acceleration in their solutions for many years now, but with capacities of NVMe expanding, 2019 is going to be the year for primary storage shifting onto NVMe drives. Because of the added cost, however, you won’t see a mass market shift, but you’ll see the highest value workloads exploring ways to make NVMe its primary storage. I see a lot of bare bones arrays embracing NVMe rather than the big, traditional storage arrays whose data services (like compression and deduplication) would slow things down.

December 20, 2018 0 comments

Datacenter News Storage

E8 STORAGE ACCELERATES PERFORMANCE FOR GENOMIC WORKLOADS AT QUEEN MARY UNIVERSITY OF LONDON

by Phil Marsden November 13, 2018

written by Phil Marsden

E8 STORAGE ACCELERATES PERFORMANCE FOR GENOMIC WORKLOADS AT QUEEN MARY UNIVERSITY OF LONDON

Having worked both as a researcher in computational chemistry and as IT support for the Department of Chemistry at a major University I understand the demards and inportance of storage in such an environment. I congratulate Queen Mary University of London for choosing E8, who are cementing there place as a top storage provider running some of the most important institutions.

Read more in E8`s official press release bellow :

Delivering record breaking performance speeds, E8 Storage underpins leading university scientific research department

SANTA CLARA, Calif., October 30, 2018 – E8 Storage today announced that Queen Mary University of London (QMUL) has deployed its E8-D24 array, to underpin its genomic workflow and significantly increase performance. With the newly updated high-performance cluster, supporting both researchers and students, the university will be able to leverage the benefits of the industry’s first shared NVMe storage solution to optimize its data intensive workloads.

QMUL is part of the Russell Group and is ranked fifth in the UK in terms of research quality. Integral to this ranking is its study of genomics, which requires high computing and processing power. IT services has a fundamental role to play in supporting this research, and with the need to cater to students and influential researchers, the institution decided to research the market and with the support of its channel partner, OCF, looked for the latest cutting-edge technology with performance and optimization.

With a shared NVMe platform in place, QMUL will be able to accelerate the performance of the fast-tier scratch space behind a distributed file system to support genomic sequencing. The high performance and capacity of E8 Storage NVMe-over-Fabrics system will be used for both data and metadata, enabling QMUL to push more jobs through the cluster at a faster rate. The previous high-performance system that the university used to support its genomics research was no longer fast enough to support its needs. Upon seeing the new performance benchmarks published by E8 Storage, the institution decided to upgrade, opting for the E8-D24 array to support genomics and moving the existing storage to a bulk storage tier. With E8 Storage, QMUL now has equipment that can scale for the future and deliver the fastest performance speed in the industry.

“Performance and data optimization are essential to the work behind genomic processing, and E8 Storage overcomes the traditional storage constraints with its first-to-market shared NVMe storage platform,” commented Tom King, Assistant Director for Research, IT Services at Queen Mary University of London. “The team have been extremely responsive right from the start, and we were able to discuss our main use case and pain points to see how the technology would enable us to overcome this. In addition, we were extremely pleased with the latest benchmark performance tests which showed E8 Storage as a leader in NVMe.”

Mahesh Pancholi, business development manager at OCF noted: “OCF are pleased to be working with E8 Storage to provide the industry leading NVMe-over-Fabric solution to QMUL. The combination of great performance figures and integration with IBM Spectrum Scale make the E8 Storage appliance a great way for QMUL to provide a high-performance scratch tier to their cutting-edge researchers. OCF are proud to have been working with QMUL to expand their cluster and provide cutting edge technologies and services to their world-class researchers.”

“The combination of Mellanox InfiniBand and E8 Storage provides the high computing and storage performance needed for QMUL’s research activities,” said Gilad Shainer, Vice President of Marketing at Mellanox. “Connecting the shared NVMe storage with high speed InfiniBand network is an ideal design for genomic applications.”

About E8 Storage

E8 Storage is a pioneer in shared NVMe storage for data-intensive, high-performance applications that drive business revenue. E8 Storage’s affordable, reliable and scalable solution is ideally suited for the most demanding low-latency workloads, including real-time analytics, financial and trading applications, genomics and large-scale file systems. Driven by the company’s patented architecture, E8 Storage’s high-performance NVMe over Fabrics certified storage delivers record breaking performance at half the cost of existing storage products. When performance matters, enterprise data centers turn to E8 Storage for unprecedented storage performance, density and scale, without compromising on reliability and availability. Privately held, E8 Storage is based in Santa Clara with R&D in Tel Aviv, and channel partners throughout the US and Europe. For more information, please visit www.e8storage.com, and follow us on Twitter @E8Storage and LinkedIn.

November 13, 2018 0 comments

Datacenter News Storage

NETAPP INSIGHT 2018 – TFDX- MAX Data

by Roger Lund October 24, 2018

written by Roger Lund

It’s Wednesday. We are here at the Netapp Insight event. And Netapp again is presenting on MAX Data.

A Few of the delegates.

@Arjantim @ChanEk81 @DeepStorageNet @settlersoman @darkkavenger @RayLucchesi @sfoskett @MBLeib @BenTGage https://tfd.bz/2pXpD82

also we have @davidchapa @ChrisMaki @mcbride_ruairi at the table and @PeytStefanova and @timwaldron in the room!

Yes it’s the hashtag #NetAppInsight and #TFDx !

Here we go!

Andrew Grimes is presenting

See his blog post on this technology prior to the release.

An Update on the Plexistor Acquisition: Introducing NetApp Memory Accelerated Data

Netapp Max Data : In-Memory Application Data Services

Product Page for Max Data a Acquisition of plexi store.

https://www.netapp.com/us/products/data-management-software/max-data.aspx

MAX Data Datasheet

Podcast: Tech ONTAP Podcast Episode 154: NetApp MAX Data

Agenda

Netapp Flash Memory Fabric
Max Data Overview

Use cases
Resources

Netapp Data Fabric, this product fits well.

Real-Time Applications : the next phase of digital transformation.

Deep Learning
Real-Time Analytics

Started with Disk
Focus on the future of Solid-State Media.
Then NVMe and now persistent memory.

Layer 4-Layer 0

Lower latency, more IOPS as you move to layer 0 but costs rise as you move to Layer 0.
- ( note i’ll insert a picture here )
Persistent Memory Technology.

NVDIMM , based on 3D Xpoint.
Costly – intel optane dimm’s should lower costs.

Comprehensive NVMe : End-to-End & non-disruptive.

Add FAS, AFF, and AFF with NVMe non disruptively.
Storage-Class Memory SCM as cache.

Persistent Memory PMEM in Server.
NVMe Over fabrics.

What is Max Data Design Goals

Faster application response time – focus on low latency

leverage server-side memory data locality ( close to application )
software-only approach + PMEM

Acceleration for read and write workloads
- not a cache ( page caching not needed )
current software stack written to disk. no re architecture of the software stack for memory

Application plug and play
enterprise data protection services.
Put the LUN on the bus Enterprise Data Protection Tools

Linux IO/ Stack vs Linux Stack leveraging Netapp MAX Data File System

MAX Data on the memory bus PMEM Data Acceleration

Primary Tier

Memory. DRAM, NVDIMM, D3XP

Secondary Tier

Flash ( must be block)
Local flash SSD, or LUN.

Memory Optimized File System

Single Mount Point
Acceleration any workload

Focus on acceleration Writes.
Read after read acceleration.

Minimal Change to the application

Accelerate PM deployments.

Server and SAN can be supported.

1:25 ratio T1 should be backup-ed to T2

Sub 10 Micro Sec Latency.

T1 replication possible. adds 4 MS of latency.

Sync mirroring.

Max Data Data Services for PMEM

Protecting writes at memory speeds.
- Max Snapsync
- T1 to T2
- Integrated with Ontap snapshot
- Integrated with ontap Snapmiorror
Max Recovery
- T1 to T1 mirroring of writes
- Private RDMA Ethernet 100GB
- RPO 0 Data Protection for in memory.

Brings Enterprise Data Services to PMEM

MAX Data Cloud Integrations

Integrated Cloud Services

ontap fabric pool archive/offiste

ontap NDAS Backup
NPS
Cloud Volume Services ontap

Native Cloud Data Services

Future Releases

Resident Cloud instance
AWS, EC2, With DRAM and EBS
Azure, and GCP have been tested.

MAX Data Software Defined flash

Server

SAN
Cloud

Netapp MAX Data

use cases

MAX Data – Target Workloads

Real-time Analytics trading platforms
in-memory databases

AI
Data Wearhousing

Mongo DB

supports very large work sets
3x better OPS/SEC

4-5x lower latency

Planned on a PER Server license.

October 24, 2018 0 comments

Datacenter News Storage

E8 STORAGE SETS ANOTHER NEW RECORD THIS YEAR IN INDEPENDENT BENCHMARK TEST

by Phil Marsden September 2, 2018

written by Phil Marsden

Ok, We all like fast storage, and E8 are blazing new trails with their shared NVMe storage solutions, facilitating distributed storage at mind bending speeds, Utilizing the massively parallel capabilities of the NVMe protocol. With massive overhead in the NVMe protocol E8`s potential for expansion and optimization make them a storage provider to watch.

read more in their recent press release :

E8 STORAGE SETS ANOTHER NEW RECORD THIS YEAR IN INDEPENDENT BENCHMARK TEST

E8 Storage to demonstrate record-breaking performance at Flash Memory Summit (FMS) 2018

SANTA CLARA, Calif., August 7, 2018 – E8 Storage today announced that the company’s high-performance, shared accelerated storage solution has set a new record for the SPEC SFS^® 2014performance benchmark. As the only shared NVMe storage solution to break the 1ms overall response time (ORT) threshold, E8 Storage has once again beat all other previously published SPEC SFS2014_swbuild test results. The solution leverages a single E8-X24 array in combination with an IBM Spectrum Scale 5.0.1 high performance cluster, without requiring any special performance tuning on the storage subsystem.

More specifically, E8 Storage delivers the industry’s lowest sub-millisecond overall response time (ORT) of 0.57ms, while meeting previous records on the number of builds. The record-breaking results validate E8 Storage’s vision to bring high performance, low latency and simplified storage management without compromising on reliability, availability or scalability. The configuration powers financial trading systems, real-time data analytics, and genomics sequencing research. Demonstrations of the record-breaking configuration will take place during FMS 2018 at booth 745.

“E8 Storage’s architecture is designed for NVMe and we can use any industry standard drive, allowing us to deliver the power behind innovation, with the lowest possible latency at a much lower TCO than many other cache memory solutions,” says Zivan Ori, co-founder and CEO of E8 Storage. “With the SPEC SFS® 2014 validation, we are delighted to have achieved these ground-breaking results, which showcase our ability to power, accelerate and scale high performance applications, so customers can quickly and efficiently extract value from their data.”

Available since April 2018, the E8-X24 also set multiple STAC-M3™ benchmark records, with the fastest tests performing 17x faster than the previous results. These independently verified, record-breaking speeds reaffirm E8 Storage’s vision to extract the full performance of flash for massive scalability of data hungry applications such as data centers with data- and transaction-intensive applications.

For a copy of the full results visit: https://www.spec.org/sfs2014/results/res2018q3/sfs2014-20180706-00041.html

For more information on purchasing E8 Storage appliances, please contact [email protected].

About E8 Storage

September 2, 2018 0 comments

Datacenter News Storage

E8 Storage Announces InfiniBand Support Extending High Performance Compute (HPC) Options for New and Existing Customers

by Phil Marsden April 24, 2018

written by Phil Marsden

Big news from growing storage player, E8. Bringing InfiniBand to its blazingly fast storage solutions. In a very welcome move E8 have announced added support for InfiniBand to their storage solutions for customers looking to leverage the blazing speed of the centralized NVMe flash only storage, in storage intensive workloads via the speed of InfiniBand.

This will allow local flash storage speeds from networked centralized storage ensuring sufficient IOPS for even the most intensive workloads.

Their press release follows

E8 Storage Announces InfiniBand Support Extending High Performance Compute (HPC) Options for New and Existing Customers

Support for the solution highlights the close alignment with Mellanox Technologies and delivers unrivaled performance and ultra-low latency

SANTA CLARA, CALIF., April 17, 2018 – E8 Storage today announced availability of InfiniBand support to its high performance, NVMe storage solutions. The move comes as a direct response to HPC customers that wish to take advantage of the high speed, low latency throughput of InfiniBand for their data hungry applications. E8 Storage support for InfiniBand will be seamless for customers who now have the flexibility to connect via Ethernet or InfiniBand when paired with Mellanox ConnectX InfiniBand/VPI adapters.

As our economy becomes more digitally-defined, HPC environments play a critical role in unlocking and harnessing the transformational power of data,” commented Scott Sinclair, senior analyst at ESG. “InfiniBand connectivity combined with the full potential of flash as delivered by E8 Storage is an almost unstoppable partnership in terms of speed, low latency and scalability. Together, this makes E8 Storage and Mellanox a powerful combination for data center implementations and for databases that drive financial trading systems, real-time data analytics, and healthcare applications.”

The announcement also strengthens E8 Storage’s long-standing technology partnership with Mellanox that originally enabled the integration of E8 Storage’s rack scale flash architecture with selected Mellanox adapters. Today, this relationship allows E8 Storage systems to attach directly to existing RDMA high performance networks over Converged Ethernet (RoCE) or InfiniBand connections and makes E8 Storage’s full support of InfiniBand possible especially in HPC environments where the majority of InfiniBand adoption is occurring. Like Ethernet, the InfiniBand products perform at 100Gb/s allowing enhanced scalability as E8 Storage solutions are deployed.

According to the most recent Top 500 supercomputer rankings (Nov 2017), InfiniBand is the second most-used internal system interconnect technology. Further, in a recent evaluation by Gilad Shainer, vice president of marketing for HPC products at Mellanox Technologies, Mellanox accounts for 62 percent share of the compute clustering of the true HPC systems.

“We are very pleased that E8 Storage is expanding this partnership to offer full support for our InfiniBand products. Working closely together, our combined products can deliver the lightning-fast and highly scalable network necessary to power data intensive applications, such as artificial intelligence and machine learning that now drive today’s economy,” stated Scot Schultz, Senior Director, HPC/AI and Technical Computing, Mellanox Technologies.

“Today we demonstrate once again that E8 Storage’s architecture can expand, evolve and always extract the full potential of flash performance,” comments Zivan Ori, co-founder and CEO of E8 Storage. “Partnering with market leaders like Mellanox that deliver the very best network connectivity technology ensures we continue to meet and, frequently, exceed the needs of our HPC customers even in their most demanding environments.”

E8 Storage software and appliances with InfiniBand support are available now, contact E8 Storage or your reseller for information on pricing.

E8 Storage is also a sponsor of the IBM Spectrum Scale (GPFS) User Group which will be held in London from April 18-19, 2018. The event features Spectrum Scale filesystem experts sharing the latest updates. To schedule a meeting with executives, visit https://e8storage.com/events/.

About E8 Storage

E8 Storage is a pioneer in shared accelerated storage for data-intensive, high-performance applications that drive business revenue. E8 Storage’s affordable, reliable and scalable solution is ideally suited for the most demanding low-latency workloads, including real-time analytics, financial and trading applications, transactional processing and large-scale file systems. Driven by the company’s patented architecture, E8 Storage’s high-performance shared NVMe storage solution delivers 10 times the performance at half the cost of existing storage products. With E8 Storage, enterprise datacenters can enjoy unprecedented storage performance density and scale, delivering NVMe performance without compromising on reliability and availability. Privately held, E8 Storage is based in Santa Clara with R&D in Tel Aviv, and channel partners throughout the US and Europe. For more information, please visit www.e8storage.com, and follow us on Twitter @E8Storage and LinkedIn.

Who wouldn’t be excited by faster access to their data? i know i am!

Thanks for reading

Tchau for now!

April 24, 2018 0 comments