Tag: hpc

Posted on

Dell Technology World 2018 Announcement Summary

Dell Technology World 2018 Announcement Summary

Dell Technology World 2018 Announcement Summary
This is part one of a five-part series about Dell Technology World 2018 announcement summary. Last week (April 30-May 3) I traveled to Las Vegas Nevada (LAS) to attend Dell Technology World 2018 (e.g., DTW 2018) as a guest of Dell (that is a disclosure btw). There were several announcements along with plenty of other activity from sessions, meetings, hallway and event networking taking place at Dell Technology World DTW 2018.

Major data infrastructure technology announcements include:

  • PowerMax all-flash array (AFA) solid state device (SSD) NVMe storage system
  • PowerEdge four-socket 2U and 4U rack servers
  • XtremIO X2 AFA SSD storage system updates
  • PowerEdge MX preview of future composable servers
  • Desktop and thin client along with other VDI updates
  • Cloud and networking enhancements

Besides the above, additional data infrastructure related announcements were made in association with Dell Technology family members including VMware along with other partners, as well as customer awards. Other updates and announcements were tied to business updates from Dell Technology, Dell Technical Capital (venture capital), and, Dell Financial Services.

Dell Technology World Buzzword Bingo Lineup

Some of the buzzword bingo terms, topics, acronyms from Dell Technology World 2018 included AFA, AI, Autonomous, Azure, Bare Metal, Big Data, Blockchain, CI, Cloud, Composable, Compression, Containers, Core, Data Analytics, Dedupe, Dell, DFS (Dell Financial Services), DFR (Data Footprint Reduction), Distributed Ledger, DL, Durability, Fabric, FPGA, GDPR, Gen-Z, GPU, HCI, HDD, HPC, Hybrid, IOP, Kubernetes, Latency, MaaS (Metal as a Service), ML, NFV, NSX, NVMe, NVMeoF, PACE (Performance Availability Capacity Economics), PCIe, Pivotal, PMEM, RAID, RPO, RTO, SAS, SATA, SC, SCM, SDDC, SDS, Socket, SSD, Stamp, TBW (Terabytes Written per day), VDI, venture capital, VMware and VR among others.

Dell Technology World 2018 Venue
Dell Technology World DTW 2018 Event and Venue

Dell Technology World 2018 was located at the combined Palazzo and Venetian hotels along with adjacent Sands Expo center kicking off Monday, April 30th and wrapping up May 4th.

The theme for Dell Technology World DTW 2018 was make it real, which in some ways was interesting given the focus on virtual including virtual reality (VR), software-defined data center (SDDC) virtualization, data infrastructure topics, along with artificial intelligence (AI).

Virtual Sky Dell Technology World 2018
Make it real – Venetian Palazzo St. Mark’s Square on the way to Sands Expo Center

There was plenty of AI, VR, SDDC along with other technologies, tools as well as some fun stuff to do including VR games.

Dell Technology World 2018 Commons Area
Dell Technology World Village Area near Key Note and Expo Halls

Dell Technology World 2018 Commons Area Drones
Dell Technology World Drone Flying Area

During a break from some meetings, I used a few minutes to fly a drone using VR which was interesting. I Have been operating drones (See some videos here) visually without dependence on first-person view (FPV) or relying on extensive autonomous operations instead flying heads up by hand for several years. Needless to say, the VR was interesting, granted encountered a bit of vertigo that I had to get used to.

Dell Technology World 2018 Commons Area Virtual Village
More views of the Dell Technology World Village and Commons Area with VR activity

Dell Technology World 2018 Commons Area Virtual Village
Dell Technology World Village and VR area

Dell Technology World 2018 Commons Area Virtual Village
Dell Technology World Bean Bag Area

Dell Technology World 2018 Announcement Summary

Ok, nuff with the AI, ML, DL, VR fun, time to move on to the business and technology topics of Dell Technologies World 2018.

What was announced at Dell Technology World 2018 included among others:

Dell Technology World 2018 PowerMax
Dell PowerMax Front View

Subsequent posts in this series take a deeper look at the various announcements as well as what they mean.

Where to learn more

Learn more about Dell Technology World 2018 and related topics via the following links:

Additional learning experiences along with common questions (and answers), as well as tips can be found in Software Defined Data Infrastructure Essentials book.

Software Defined Data Infrastructure Essentials Book SDDC

What this all means

On the surface it may appear that there was not much announced at Dell Technology World 2018 particular compared to some of the recent Dell EMC Worlds and EMC Worlds. However turns out that there was a lot announced, granted without some of the entertainment and circus like atmosphere of previous events. Continue reading here Part II Dell Technology World 2018 Modern Data Center Announcement Details in this series, along with Part III here, Part IV here (including PowerEdge MX composable infrastructure leveraging Gen-Z) and Part V (servers and converged) here.

Ok, nuff said, for now.

Cheers Gs

Greg Schulz – Microsoft MVP Cloud and Data Center Management, VMware vExpert 2010-2017 (vSAN and vCloud). Author of Software Defined Data Infrastructure Essentials (CRC Press), as well as Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press), Resilient Storage Networks (Elsevier) and twitter @storageio. Courteous comments are welcome for consideration. First published on https://storageioblog.com any reproduction in whole, in part, with changes to content, without source attribution under title or without permission is forbidden.

All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.

Posted on

Part II Dell Technology World 2018 Modern Data Center Announcement Details

Part II Dell Technology World 2018 Modern Data Center Announcement Details

Dell Technology World 2018 Modern Data Center Announcement Summary
This is Part II Dell Technology World 2018 Modern Data Center Announcement Details that is part of a five-post series (view part I here, part III here, part IV here and part V here). Last week (April 30-May 3) I traveled to Las Vegas Nevada (LAS) to attend Dell Technology World 2018 (e.g., DTW 2018) as a guest of Dell (that is a disclosure btw).

Dell Technology World 2018 Venue
Dell Technology World DTW 2018 Event and Venue

What was announced at Dell Technology World 2018 included among others:

Dell Technology World 2018 PowerMax
Dell PowerMax Front View

Dell Technology World 2018 Modern Data Center Announcement Details

Dell Technologies data infrastructure related announcements included new solutions competencies and expanded services deployment competencies with partners to boost deal size and revenues. An Internet of Things (IoT) solution competency was added with others planned including High-Performance Computing (HPC) / Super Computing (SC), Data Analytics, Business Applications and Security related topics. Dell Financial Services flexible consumption models announced at Dell EMC World 2017 provide flexible financing options for both partners as well as their clients.

Flexible Dell Financial Services cloud-like consumption model (e.g., pay for what you use) enhancements include reduced entry points for the Flex on Demand solutions across the Dell EMC storage portfolio. For example, Flex on Demand velocity pricing models for Dell EMC Unity All-Flash Array (AFA) solid state device (SSD) storage solution, and XtremIO X2 AFA systems with price points of less than USD 1,000.00 per month. The benefit is that Dell partners have a financial vehicle to help their midrange customers run consumption-based financing for all-flash storage without custom configurations resulting in faster deployment opportunities.

In other partner updates, Dell Technologies is enhancing the incentive program Dell EMC MyRewards program to help drive new business. Dell EMC MyRewards Program is an opt-in, points-based reward program for solution provider sales reps and systems engineers. MyRewards program is slated to replace the existing Partner Advantage and Sell & Earn programs with bigger and better promotions (up to 3x bonus payout, simplified global claiming).

What this means for partners is the ability to earn more while offering their clients new solutions with flexible financing and consumption-based pricing among other options. Other partner enhancements include update demo program, Proof of Concept (POC) program, and IT transformation campaigns.

Powering up the Modern Data Center and Future of Work

Powering up the modern data center along with future of work, part of the make it real theme of Dell Technologies world 2018 includes data infrastructure server, storage, I/O networking hardware, software and service solutions. These data infrastructure solutions include NVMe based storage, Converged Infrastructure (CI), hyper-converged infrastructure (HCI), software-defined data center (SDDC), VMware based multi-clouds, along with modular infrastructure resources.

In addition to server and storage data infrastructure resources form desktop to data center, Dell also has a focus of enabling traditional as well as emerging Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL) as well as analytics applications. Besides providing data infrastructure resources to support AI, ML, DL, IoT and other applications along with their workloads, Dell is leveraging AI technology in some of their products for example PowerMax.

Other Dell Technologies announcements include Virtustream cloud risk management and compliance, along with Epic and SAP Digital Health healthcare software solutions. In addition to Virtustream, Dell Technologies cloud-related announcements also include VMware NSX network Virtual Cloud Network with Microsoft Azure support along with security enhancements. Refer here to recent April VMware vSphere, vCenter, vSAN, vRealize and other Virtual announcements as well as here for March VMware cloud updates.

Where to learn more

Learn more about Dell Technology World 2018 and related topics via the following links:

Additional learning experiences along with common questions (and answers), as well as tips can be found in Software Defined Data Infrastructure Essentials book.

Software Defined Data Infrastructure Essentials Book SDDC

What this all means

The above set of announcements span business to technology along with partner activity. Continue reading here (Part III Dell Technology World 2018 Storage Announcement Details) of this series, and part I (general summary) here, along with Part IV (PowerEdge MX Composable) here and part V here.

Ok, nuff said, for now.

Cheers Gs

Greg Schulz – Microsoft MVP Cloud and Data Center Management, VMware vExpert 2010-2017 (vSAN and vCloud). Author of Software Defined Data Infrastructure Essentials (CRC Press), as well as Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press), Resilient Storage Networks (Elsevier) and twitter @storageio. Courteous comments are welcome for consideration. First published on https://storageioblog.com any reproduction in whole, in part, with changes to content, without source attribution under title or without permission is forbidden.

All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.

Posted on

Part III Dell Technology World 2018 Storage Announcement Details

Part III Dell Technology World 2018 Storage Announcement Details

Part III Dell Technology World 2018 Storage Announcement Details

This is Part III Dell Technology World 2018 Storage Announcement Details that is part of a five-post series (view part I here, part II here, part IV (PowerEdge MX Composable) here and part V here). Last week (April 30-May 3) I traveled to Las Vegas Nevada (LAS) to attend Dell Technology World 2018 (e.g., DTW 2018) as a guest of Dell (that is a disclosure btw).

Dell Technology World 2018 Storage Announcements Include:

  • PowerMax – Enterprise class tier 0 and tier 1 all-flash array (AFA)
  • XtremIO X2 – Native replication and new entry-level pricing

Dell Technology World 2018 PowerMax back view
Back view of Dell PowerMax

Dell PowerMax Something Old, Something New, Something Fast Near You Soon

PowerMax is the new companion to VMAX. Positioned for traditional tier 0 and tier 1 enterprise-class applications and workloads, PowerMax is optimized for dense server virtualization and SDDC, SAP, Oracle, SQL Server along with other low-latency, high-performance database activity. Different target workloads include Mainframe as well as Open Systems, AI, ML, DL, Big Data, as well as consolidation.

The Dell PowerMax is an all-flash array (AFA) architecture with an end to end NVMe along with built-in AI and ML technology. Building on the architecture of Dell EMC VMAX (some models still available) with new faster processors, full end to end NVMe ready (e.g., front-end server attachment, back-end devices).

The AI and ML features of PowerMax PowerMaxOS include an engine (software) that learns and makes autonomous storage management decisions, as well as implementations including tiering. Other AI and ML enabled operations include performance optimizations based on I/O pattern recognition.

Other features of PowerMax besides increased speeds, feeds, performance includes data footprint reduction (DFR) inline deduplication along with enhanced compression. The DFR benefits include up to 5:1 data reduction for space efficiency, without performance impact to boost performance effectiveness. The DFR along with improved 2x rack density, along with up to 40% power savings (your results may vary) based on Dell claims to enable an impressive amount of performance, availability, capacity, economics (e.g., PACE) in a given number of cubic feet (or meters).

There are two PowerMax models including 2000 (scales from 1 to 2 redundant controllers) and 8000 (scales from 1 to 8 redundant controller nodes). Note that controller nodes are Intel Xeon multi-socket, multi-core processors enabling scale-up and scale-out performance, availability, and capacity. Competitors of the PowerMax include AFA solutions from HPE 3PAR, NetApp, and Pure Storage among others.

Dell Technology World 2018 PowerMax Front View
Front view of Dell PowerMax

Besides resiliency, data services along with data protection, Dell is claiming PowerMax is 2x faster than their nearest high-end storage system competitors with up to 150GB/sec (e.g., 1,200Gbps) of bandwidth, as well as up to 10 million IOPS with 50% lower latency compared to previous VMAX.

PowerMax is also a full end to end NVMe ready (both back-end and front-end). Back-end includes NVMe drives, devices, shelves, and enclosures) as well as front-end (future NVMe over Fabrics, e.g., NVMeoF). Being NVMeoF ready enables PowerMax to support future front-end server network connectivity options to traditional SAN Fibre Channel (FC), iSCSI among others.

PowerMax is also ready for new, emerging high speed, low-latency storage class memory (SCM).  SCM is the next generation of persistent memories (PMEM) having performance closer to traditional DRAM while persistence of flash SSD. Examples of SCM technologies entering the market include Intel Optane based on 3D XPoint, along with others such as those from Everspin among others.

IBM Z Zed Mainframe at Dell Technology World 2018
An IBM “Zed” Mainframe (in case you have never seen one)

Based on the performance claims, the Dell PowerMax has an interesting if not potentially industry leading power, performance, availability, capacity, economic footprint per cubic foot (or meter). It will be interesting to see some third-party validation or audits of Dell claims. Likewise, I look forward to seeing some real-world applied workloads of Dell PowerMax vs. other storage systems. Here are some additional perspectives Via SearchStorage: Dell EMC all-flash PowerMax replaces VMAX, injects NVMe


Dell PowerMax Visual Studio (Image via Dell.com)

To help with customer decision making, Dell has created an interactive VMAX and PowerMax configuration studio that you can use to try out as well as learn about different options here. View more Dell PowerMax speeds, feeds, slots, watts, features and functions here (PDF).

Dell Technology World 2018 XtremIO X2

XtremIO X2

Dell XtremIO X2 and XIOS 6.1 operating system (software-defined storage) enhanced with native replication across wide area networks (WAN). The new WAN replication is metadata-aware native to the XtremIO X2 that implements data footprint reduction (DFR) technology reducing the amount of data sent over network connections. The benefit is more data moved in a given amount of time along with better data protection requiring less time (and network) by only moving unique changed data.

Dell Technology World 2018 XtremIO X2 back view
Back View of XtremIO X2

Dell EMC claims to reduce WAN network bandwidth by up to 75% utilizing the new native XtremIO X2 native asynchronous replication. Also, Dell says XtremIO X2 requires up to 38% less storage space at disaster recovery and business resiliency locations while maintaining predictable recovery point objectives (RPO) of 30 seconds. Another XtremIO X2 announcement is a new entry model for customers at up to 55% lower cost than previous product generations. View more information about Dell XtremIO X2 here, along with speeds feeds here, here, as well as here.

What about Dell Midrange Storage Unity and SC?

Here are some perspectives Via SearchStorage: Dell EMC midrange storage keeps its overlapping arrays.

Dell Bulk and Elastic Cloud Storage (ECS)

One of the questions I had going into Dell Technology World 2018 was what is the status of ECS (and its predecessors Atmos as well as Centera) bulk object storage is given lack of messaging and news around it. Specifically, my concern was that if ECS is the platform for storing and managing data to be preserved for the future, what is the current status, state as well as future of ECS.

In conversations with the Dell ECS folks, ECS which has encompassed Centera functionality and it (ECS) is very much alive, stay tuned for more updates. Also, note that Centera has been EOL. However, its feature functionality has been absorbed by ECS meaning that data preserved can now be managed by ECS. While I can not divulge the details of some meeting discussions, I can say that I am comfortable (for now) with the future directions of ECS along with the data it manages, stay tuned for updates.

Dell Data Protection

What about Data Protection? Security was mentioned in several different contexts during Dell Technology World 2018, as was a strong physical security presence seen at the Palazzo and Sands venues. Likewise, there was a data protection presence at Dell Technologies World 2018 in the expo hall, as well as with various sessions.

What was heard was mainly around data protection management tools, hybrid, as well as data protection appliances and data domain-based solutions. Perhaps we will hear more from Dell Technologies World in the future about data protection related topics.

Where to learn more

Learn more about Dell Technology World 2018 and related topics via the following links:

Additional learning experiences along with common questions (and answers), as well as tips can be found in Software Defined Data Infrastructure Essentials book.

Software Defined Data Infrastructure Essentials Book SDDC

What this all means

If there was any doubt about would Dell keep EMC storage progressing forward, the above announcements help to show some examples of what they are doing. On the other hand, lets stay tuned to see what news and updates appear in the future pertaining to mid-range storage (e.g. Unity and SC) as well as Isilon, ScaleIO, Data Protection platforms as well as software among other technologies.

Continue reading part IV (PowerEdge MX Composable and Gen-Z) here in this series, as well as part I here, part II here, part IV (PowerEdge MX Composable) here, and, part V here.

Ok, nuff said, for now.

Cheers Gs

Greg Schulz – Microsoft MVP Cloud and Data Center Management, VMware vExpert 2010-2017 (vSAN and vCloud). Author of Software Defined Data Infrastructure Essentials (CRC Press), as well as Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press), Resilient Storage Networks (Elsevier) and twitter @storageio. Courteous comments are welcome for consideration. First published on https://storageioblog.com any reproduction in whole, in part, with changes to content, without source attribution under title or without permission is forbidden.

All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.

Posted on

Part IV Dell Technology World 2018 PowerEdge MX Gen-Z Composable Infrastructure

Part IV Dell Technology World 2018 PowerEdge MX Gen-Z Composable Infrastructure

Part IV Dell Technology World 2018 PowerEdge MX Gen-Z Composable Infrastructure
This is Part IV Dell Technology World 2018 PowerEdge MX Gen-Z Composable Infrastructure that is part of a five-post series (view part I here, part II here, part III here and part V here). Last week (April 30-May 3) I traveled to Las Vegas Nevada (LAS) to attend Dell Technology World 2018 (e.g., DTW 2018) as a guest of Dell (that is a disclosure btw).

Introducing PowerEdge MX Composable Infrastructure (the other CI)

Dell announced at Dell Technology World 2018 a preview of the new PowerEdge MX (kinetic) family of data infrastructure resource servers. PowerEdge MX is being developed to meet the needs of resource-centric data infrastructures that require scalability, as well as performance availability, capacity, economic (PACE) flexibility for diverse workloads. Read more about Dell PowerEdge MX, Gen-Z and composable infrastructures (the other CI) here.

Some of the workloads being targeted by PowerEdge MX include large-scale dense SDDC virtualization (and containers), private (or public clouds by service providers). Other workloads include AI, ML, DL, data analytics, HPC, SC, big data, in-memory database, software-defined storage (SDS), software-defined networking (SDN), network function virtualization (NFV) among others.

The new PowerEdge MX previewed will be announced later in 2018 featuring a flexible, decomposable, as well as composable architecture that enables resources to be disaggregated and reassigned or aggregated to meet particular needs (e.g., defined or composed). Instead of traditional software defined virtualization carving up servers in smaller virtual machines or containers to meet workload needs, PowerEdge MX is part of a next-generation approach to enable server resources to be leveraged at a finer granularity.

For example, today an entire server including all of its sockets, cores, memory, PCIe devices among other resources get allocated and defined for use. A server gets defined for use by an operating system when bare metal (or Metal as a Service) or a hypervisor. PowerEdge MX (and other platforms expected to enter the market) have a finer granularity where with a proper upper layer (or higher altitude) software resources can be allocated and defined to meet different needs.

What this means is the potential to allocate resources to a given server with more granularity and flexibility, as well as combine multiple server’s resources to create what appears to be a more massive server. There are vendors in the market who have been working on and enabling this type of approach for several years ranging from ScaleMP to startup Liqid and Tidal among others. However, at the heart of the Dell PowerEdge MX is the new emerging Gen-Z technology.

If you are not familiar with Gen-Z, add it to your buzzword bingo lineup and learn about it as it is coming your way. A brief overview of Gen-Z consortium and Gen-Z material and primer information here. A common question is if Gen-Z is a replacement for PCIe which for now is that they will coexist and complement each other. Another common question is if Gen-Z will replace Ethernet and InfiniBand and the answer is for now they complement each other. Another question is if Gen-Z will replace Intel Quick Path and another CPU device and memory interconnects and the answer is potentially, and in my opinion, watch to see how long Intel drags its feet.

Note that composability is another way of saying defined without saying defined, something to pay attention too as well as have some vendor fun with. Also, note that Dell is referent to PowerEdge MX and Kinetic architecture which is not the same as the Seagate Kinetic Ethernet-based object key value accessed drive initiative from a few years ago (learn more about Seagate Kinetic here). Learn more about Gen-Z and what Dell is doing here.

Where to learn more

Learn more about Dell Technology World 2018 and related topics via the following links:

Additional learning experiences along with common questions (and answers), as well as tips can be found in Software Defined Data Infrastructure Essentials book.

Software Defined Data Infrastructure Essentials Book SDDC

What this all means

Dell has provided a glimpse of what they are working on pertaining composable infrastructure, the other CI, as well as Gen-Z and related next generation of servers with PowerEdge MX as well as Kinetic. Stay tuned for more about Gen-Z and composable infrastructures. Continue reading Part V (servers converged) in this series here, as well as part I here, part II here and part III here.

Ok, nuff said, for now.

Cheers Gs

Greg Schulz – Microsoft MVP Cloud and Data Center Management, VMware vExpert 2010-2017 (vSAN and vCloud). Author of Software Defined Data Infrastructure Essentials (CRC Press), as well as Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press), Resilient Storage Networks (Elsevier) and twitter @storageio. Courteous comments are welcome for consideration. First published on https://storageioblog.com any reproduction in whole, in part, with changes to content, without source attribution under title or without permission is forbidden.

All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.

Posted on

HPE Announces AMD Powered Gen 10 ProLiant DL385 For Software Defined Workloads

HPE Announces AMD Powered Gen 10 ProLiant DL385 For Software Defined Workloads

server storage I/O data infrastructure trends

By Greg Schulz – www.storageioblog.com November 20, 2017

HPE Announced today a new AMD EPYC 7000 Powered Gen 10 ProLiant DL385 for Software Defined Workloads including server virtualization, software-defined data center (SDDC), software-defined data infrastructure (SDDI), software-defined storage among others. These new servers are part of a broader Gen10 HPE portfolio of ProLiant DL systems.

HPE AMD EPYC Gen10 DL385
24 Small Form Factor Drive front view DL385 Gen 10 Via HPE

The value proposition being promoted by HPE of these new AMD powered Gen 10 DL385 servers besides supporting software-defined, SDDI, SDDC, and related workloads are security, density and lower price than others. HPE is claiming with the new AMD EPYC system on a chip (SoC) processor powered Gen 10 DL385 that it is offering up to 50 percent lower cost per virtual machine (VM) than traditional server solutions.

About HPE AMD Powered Gen 10 DL385

HPE AMD EPYC 7000 Gen 10 DL385 features:

  • 2U (height) form factor
  • HPE OneView and iLO management
  • Flexible HPE finance options
  • Data Infrastructure Security
  • AMD EPYC 7000 System on Chip (SoC) processors
  • NVMe storage (Embedded M.2 and U.2/8639 Small Form Factor (SFF) e.g. drive form factor)
  • Address server I/O and memory bottlenecks

These new HPE servers are positioned for:

  • Software Defined, Server Virtualization
  • Virtual Desktop Infrastructure (VDI) workspaces
  • HPC, Cloud and other general high-density workloads
  • General Data Infrastructure workloads that benefit from memory-centric or GPUs

Different AMD Powered DL385 ProLiant Gen 10 Packaging Options

Common across AMD EPYC 7000 powered Gen 10 DL385 servers are 2U high form factor, iLO management software and interfaces, flexible LAN on Motherboard (LOM) options, MicroSD (optional dual MicroSD), NVMe (embedded M.2 and SFF U.2) server storage I/O interface and drives, health and status LEDs, GPU support, single or dual socket processors.

HPE AMD EPYC Gen10 DL385 Look Inside
HPE DL385 Gen10 Inside View Via HPE

HPE AMD EPYC Gen10 DL385 Rear View
HPE DL385 Gen10 Rear View Via HPE

Other up to three storage drive bays, support for Large Form Factor (LFF) and Small Form Factor (SFF) devices (HDD and SSD) including SFF NVMe (e.g., U.2) SSD. Up to 4 x Gbe NICs, PCIe riser for GPU (optional second riser requires the second processor). Other features and options include HPE SmartArray (RAID), up to 6 cooling fans, internal and external USB 3. Optional universal media bay that can also add a front display, optional Optical Disc Drive (ODD), optional 2 x U.2 NVMe SFF SSD. Note media bay occupies one of three storage drive bays.

HPE AMD EPYC Gen10 DL385 Form Factor
HPE DL385 Form Factor Via HPE

Up to 3 x Drive Bays
Up to 12 LFF drives (2 per bay)
Up to 24 SFF drives ( 3 x 8 drive bays, 6 SFF + 2 NVMe U.2 or 8 x NVMe)

AMD EPYC 7000 Series

The AMD EPYC 7000 series is available in the single and dual socket. View additional AMD EPYC speeds and feeds in this data sheet (PDF), along with AMD server benchmarks here.

HPE AMD EPYC Specifications
HPE DL385 Gen 10 AMD EPYC Specifications Via HPE

AMD EPYC 7000 General Features

  • Single and dual socket
  • Up to 32 cores, 64 threads per socket
  • Up to 16 DDR4 DIMMS over eight channels per socket (e.g., up to 2TB RAM)
  • Up to 128 PCIe Gen 3 lanes (e.g. combination of x4, x8, x16 etc)
  • Future 128GB DIMM support

AMD EPYC 7000 Security Features

  • Secure processor and secure boot for malware rootkit protection
  • System memory encryption (SME)
  • Secure Encrypted Virtualization (SEV) hypervisors and guest virtual machine memory protection
  • Secure move (e.g., encrypted) between enabled servers

Where To Learn More

Learn more about Data Infrastructure and related server technology, trends, tools, techniques, tradecraft and tips with the following links.

  • AMD EPYC 7000 System on Chip (SoC) processors
  • Gen10 HPE portfolio and ProLiant DL systems.
  • Various Data Infrastructure related news commentary, events, tips and articles
  • Data Center and Data Infrastructure industry links
  • Data Infrastructure server storage I/O network Recommended Reading List Book Shelf
  • Software Defined Data Infrastructure Essentials (CRC 2017) Book

    • What This All Means

    With the flexible options including HDD, SSD as well as NVMe accessible SSDs, large memory capacity along with computing cores, these new solutions provide good data infrastructure server density (e.g., CPU, memory, I/O, storage) per cubic foot or meter per cost.

    I look forward to trying one of these systems out for software-defined scenarios including virtualization, software-defined storage (SDS) among others workload scenarios. Overall the HPE announcement of the new AMD EPYC 7000 Powered Gen 10 ProLiant DL385 looks to be a good option for many environments.

    Ok, nuff said, for now.

    Gs

    Greg Schulz – Microsoft MVP Cloud and Data Center Management, VMware vExpert 2010-2017 (vSAN and vCloud). Author of Software Defined Data Infrastructure Essentials (CRC Press), as well as Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press), Resilient Storage Networks (Elsevier) and twitter @storageio. Courteous comments are welcome for consideration. First published on https://storageioblog.com any reproduction in whole, in part, with changes to content, without source attribution under title or without permission is forbidden.

    All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO. All Rights Reserved. StorageIO is a registered Trade Mark (TM) of Server StorageIO.

    Posted on

    July 2017 Server StorageIO Data Infrastructures Update Newsletter

    Volume 17, Issue VII

    Hello and welcome to the July 2017 issue of the Server StorageIO update newsletter.

    It has been busy time with a lot going on, so let’s get right to this months topics which include software defined, data infrastructures, server, I/O networking, storage and related topics.

    Software-Defined Data Infrastructure Essentials SDDI SDDC

    A quick update following up from the June newsletter is that my new book is now available via Amazon.com, CRC Press and other venues in hardcopy hardcover as well as electronic versions. Visit storageio.com/book4 to learn more including view table of contents, preface, how organized among other items.

    In This Issue

    Enjoy this edition of the Server StorageIO update newsletter.

    Cheers GS

    Data Infrastructure and IT Industry Activity Trends

    Some recent Industry Activities, Trends, News and Announcements include:

    Arcserve who gained their independence from CA a few years ago has now acquired Zetta to expand its data protection solution portfolio for legacy, software defined virtual and cloud data infrastructure environments.

    Software Defined Data Infrastructure (SDDI) solution provider Compuverde has been awarded a US Patent for optimal server selection in software defined data centers (SDDC).

    The InfiniBand Trade Association (IBTA) announced its April 2017 integrators list including RDMA over Converged Ethernet (RoCE) along with other recent plugfest updates. Learn more about IBTA and related topics at www.infinibandta.org

    Next Platform has an interesting piece about a slump in the High Performance Computing (HPC) also known as High Profit Compute (for financial firms among others) along with Super Compute (SC) industry sector. Some of the vendors in the HPC and SC space include AMD, Cisco, Cray, DDN, Dell EMC, Fujifilm, IBM, Intel, HPE/SGI, Mellanox, NetApp, Oracle, Seagate and Western Digital (WD) among others. Check out the piece over at Next Platform here.

    To beef up its storage solution offerings, Cray has done a deal with Seagate who is one of their suppliers of HDDs as well as software defined storage software (ClusterStore) and related systems. In this new deal, Cray will get (e.g. now own) Lustre based ClusterStore storage solution which besides the intellectual property (IP) including software, around 100 or so workers will shift from being Seagate to becoming Cray employees. Note that Seagate acquired ClusterStore solution as part of its acquisition a few years ago of Xyratex (enclosures and Lustre storage system software).

    Meanwhile Seagate CEO Stephen Luczo is steeping down from his top leadership position after posting poor earnings and more job cuts. We have heard that HDDs are dead for decades, yet they continue to be consumed, granted with ups and downs, lately more downs for Seagate than its competitor Western Digital (WD). Seagate has dabbled in flash solid state devices (SSDs) ranging from drives to cards including acquisition of the flash business from LSI/Avago a few years ago, along with some partnerships. As a component supplier, does Seagate need to take a bold move and get closer or do a deal with one of the nand flash or other persistent memory chip makers? Or, does Seagate need to continue to move up the stack with solutions building on their previous deals buying Xyratex (enclosure and ClusterStor), Evault (data protection software), DotHill (storage systems), LSI/Avago flash (flash cards)? Lets see what happens with Seagate both from a big picture as well as more focused product, technology, portfolio perspective.

    Watch for a flurry of NVMe, along with flash SSD and other non-volatile memory (NVM), persistent memories along with storage class memories over the next few weeks. Keep in mind that If Answer is NVMe, what are the questions? Also watch for a flurry of software defined, cloud, virtual and related news towards the end of the month and into September.

    Just for fun, Mike Acton (@mike_acton) has an interesting read on Introductory bullshit detection for non-technical managers that you can check out here. Over at Coding Horror there is another interesting read on the notion of paying down your technical debt, check it out here.

    Check out other industry news, comments, trends perspectives here.

    Server StorageIO Commentary in the news

    Recent Server StorageIO industry trends perspectives commentary in the news.

    Via EnterpriseStorageForum: Comments on Who Will Rule the Storage World?
    Via SearchDataCenter: Comments on Dell EMC Cisco VxBlock news

    View more Server, Storage and I/O trends and perspectives comments here

    Server StorageIO Featured White Paper Report

    This months featured Server StorageIO Data Infrastructure content is a new white paper industry trends perspective report looking at Weka.IO Matrix Multi-Dimensional Software Based Storage (e.g. software defined storage). In this report, we look at WekaIO Matrix, an elastic, flexible, highly scalable easy to use (and manage) software-defined (e.g. software-based) storage solution. WekaIO Matrix enables radically simple software defined storage that is flexible elastic scaling with stability and without compromise.

    Weka.io Matrix software defined storage

    • Matrix is a new storage solution that:
    • Installs on bare metal, virtual or cloud servers
    • Has POSIX, NFS, SMB, and HDFS storage access
    • Adaptable performance for little and big data
    • Tiering of flash SSD and cloud object storage
    • Distributed resilience without compromise
    • Removes complexity of traditional storage

    Read more in this StorageIO Industry Trends and Perspective (ITP) Report compliments of WekaIO by clicking here. Visit www.weka.io to learn more about WekaIO and their Matrix solution.

    View more Server StorageIO Data Infrastructure White Papers, Lab Reports, Solution Briefs, Research, Polls, Surveys and additional content here on our portfolio page.

     

    Server StorageIOblog Posts

    Recent and popular Server StorageIOblog posts include:

    • Intel Announces New Xeon Processors for Software Defined Data Infrastructures
      Intel announced a new family of Xeon Scalable Processors (aka Purely) that for some workloads Intel claims to be on average of 1.65x faster than their predecessors. Note your real improvement will vary based on workload, configuration, benchmark testing, type of processor, memory, and many other server storage I/O performance considerations.

    • Who Will Be At Top Of Storage World Next Decade?
      It is safe to say that each new year will bring new trends, techniques, technologies, tools, features, functionality as well as solutions involving data storage as well as data infrastructures. This means a usual safe bet is to say that the current year is the most exciting and has the most new things than in the past when it comes to data infrastructures along with resources such as data storage.

    • Zombie Technology Life after Death Tape Is Still Alive
      A Zombie Technology is one declared dead yet has Life after Death such as Tape which is still alive, despite having declared dead for decades.

    In Case You Missed It #ICYMI

    View other recent as well as past StorageIOblog posts here

    Server StorageIO Data Infrastructure Tips and Articles

    Recent Server StorageIO industry trends perspectives commentary in the news.

    Via IDG/NetworkWorld: Ensure your data infrastructure remains available and resilient
    Via IDG/NetworkWorld: Whats a data infrastructure?
    Via InfoStor: Object Storage Is In Your Future
    Via InfoStor: Cloud Storage Concerns, Considerations and Trends
    Via InfoStor: SSD Trends, Tips and Topics

    View more Server, Storage and I/O trends and perspectives comments here

    Events and Activities

    Recent and upcoming event activities.

    Sep. 21, 2017 – MSP CMG – Minneapolis MN

    Sep. 14, 2017 – Fujifilm IT Executive Summit – Seattle WA

    Sep. 12, 2017 – SNIA Software Developers Conference (SDC) – Santa Clara CA

    Sep. 7, 2017 – TBA – Enabling, Planning and Executing Your Software Defined Journey

    August 28-30, 2017 – VMworld – Las Vegas

    June 22, 2017 – Webinar – GDPR and Microsoft Environments

    May 11, 2017 – Webinar – Email Archiving, Compliance and Ransomware

    See more webinars and activities on the Server StorageIO Events page here.

    Server StorageIO Industry Resources and Links

    Useful links and pages:
    Microsoft TechNet – Various Microsoft related from Azure to Docker to Windows
    storageio.com/links – Various industry links (over 1,000 with more to be added soon)
    objectstoragecenter.com – Cloud and object storage topics, tips and news items
    OpenStack.org – Various OpenStack related items
    storageio.com/protect – Various data protection items and topics
    thenvmeplace.com – Focus on NVMe trends and technologies
    thessdplace.com – NVM and Solid State Disk topics, tips and techniques
    storageio.com/converge – Various CI, HCI and related SDS topics
    storageio.com/performance – Various server, storage and I/O benchmark and tools
    VMware Technical Network – Various VMware related items

    Ok, nuff said, for now.

    Cheers
    Gs

    Greg Schulz – Multi-year Microsoft MVP Cloud and Data Center Management, VMware vExpert (and vSAN). Author of Software Defined Data Infrastructure Essentials (CRC Press), as well as Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press), Resilient Storage Networks (Elsevier) and twitter @storageio.

    Courteous comments are welcome for consideration. First published on https://storageioblog.com any reproduction in whole, in part, with changes to content, without source attribution under title or without permission is forbidden.

    All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2023 Server StorageIO(R) and UnlimitedIO. All Rights Reserved.

    Posted on

    Announcing SAS SANs for Dummies book, LSI edition

    There is a new (free) book that I’m a co-author of along Bruce Grieshaber and Larry Jacob (both of LSI) along with foreword by Harry Mason of LSI and President of the SCSI Trade Association titled SAS SANs for Dummies compliments of LSI.

    SAS SANs for Dummies, LSI Edition

    This new book (ebook and print hard copy) looks at Serial Attached SCSI (SAS) and how it can be used beyond traditional direct attached storage (DAS) configurations for support various types of storage mediums including SSD, HDD and tape. These configuration options include as entry-level SAN with SAS switches for small clusters or server virtualization, or as shared DAS as well as being a scale out back-end solution for NAS, object, cloud and big data storage solutions.

    Here is the table of contents (TOC) of SAS SANs for Dummies

    Chapter 1: Data storage challenges

  • Storage Growth Demand Drivers
  • Recognizing Challenges
  • Solutions and Opportunities

    • Chapter 2: Storage Area Networks

  • Introducing Storage Area Networks
  • Moving from Dedicated Internal to Shared Storage

    • Chapter 3: SAS Basics

  • Introducing the Basics of SAS
  • How SAS Functions
  • Components of SAS
  • SAS Target Devices
  • SAS for SANs

    • Chapter 4: SAS Usage Scenarios

  • Understanding SAS SANs Usage
  • Shared SAS SANs Scenarios including:
    • SAS in HPC environments
    • Big data and big bandwidth
    • Database, e-mail, back-office
    • NAS and object storage servers
    • Cloud, wen and high-density
    • Server virtualization

    Chapter 5: Advanced SAS Topics

  • The SAS Physical Layer
  • Choosing SAS Cabling
  • Using SAS Switch Zoning
  • SAS HBA Target Mode

    • Chapter 6: Nine Common Questions

  • Can You Interconnect Switches?
  • What Is SAS Cable Distance?
  • How Many Servers Can Be In a SAS SAN?
  • How Do You Manage SAS Zones?
  • How Do You Configure SAS for HA?
  • How Does SAS Zoning Compare to LUN Mapping?
  • Who Has SAS Solutions?
  • How Do SAS SANs Compare?
  • Where Can You Learn More?

    • Chapter 7: Next Steps

  • SAS Going Forward
  • Next Steps
  • Great Take Away’s

    • Regardless of if you are looking to use SAS as a primary SAN interface, or leverage it for DAS or implementing back-end storage for big-data, NAS, object, cloud or other types of scalable storage solutions, check out and get your free copy of SAS SANs for Dummies here compliments of LSI.

    SAS SANs for Dummies, LSI Edition

    Click here to ask your free copy of SAS SANs for Dummies compliments of LSI, tell them Greg from StorageIO sent you and enjoy the book.

    Ok, nuff said.

    Cheers Gs

    Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press, 2011), The Green and Virtual Data Center (CRC Press, 2009), and Resilient Storage Networks (Elsevier, 2004)

    twitter @storageio

    All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2012 StorageIO and UnlimitedIO All Rights Reserved

    Posted on

    How can direct attached storage (DAS) make a comeback if it never left?

    Server and StorageIO industry trend and perspective DAS

    Have you seen or heard the theme that Direct Attached Storage (DAS), either dedicated or shared, internal or external is making a comeback?

    Wait, if something did not go away, how can it make a comeback?

    IMHO it is as simple as for the past decade or so, DAS has been overshadowed by shared networked storage including switched SAS, iSCSI, Fibre Channel (FC) and FC over Ethernet (FCoE) based block storage area networks (SAN) and file based (NFS and Windows SMB/CIFS) network attached storage (NAS) using IP and Ethernet networks. This has been particularly true by most of the independent storage vendors who have become focused on networked storage (SAN or NAS) solutions.

    However some of the server vendors have also jumped into the deep end of the storage pool with their enthusiasm for networked storage, even though they still sell a lot of DAS including internal dedicated, along with external dedicated and shared storage.

    Server and StorageIO industry trend and perspective DAS

    The trend for DAS storage has evolved with the interfaces and storage mediums including from parallel SCSI and IDE to SATA and more recently 3Gbs and 6Gbs SAS (with 12Gbs in first lab trials). Similarly the storage mediums include a mix of fast 10K and 15K hard disk drives (HDD) along with high-capacity HDDs and ultra-high performance solid state devices (SSD) moving from 3.5 to 2.5 inch form factors.

    While there has been a lot of industry and vendor marketing efforts around networked storage (e.g. SAN and NAS), DAS based storage was over shadowed so it should not be a surprise that those focused on SAN and NAS are surprised to hear DAS is alive and well. Not only is DAS alive and well, it’s also becoming an important scaling and convergence topic for adding extra storage to appliances as well as servers including those for scale out, big data, cloud and high density not to mention high performance and high productivity computing.

    Server and StorageIO industry trend and perspective DAS

    Consequently its becoming ok to talk about DAS again. Granted you might get some peer pressure from your trend setting or trend following friends to get back on the networked storage bandwagon. Keep this in mind, take a look at some of the cool trend setting big data and little data (database) appliances, backup, dedupe and archive appliances, cloud and scale out NAS and object storage systems among others and will likely find DAS on the back-end. On a smaller scale, or in high-density rack deployments in large cloud or similar environments you may also find DAS including switched shared SAS.

    Does that mean SANs are dead?
    No, not IMHO despite what some vendors marketers and their followers will claim which is ironic given how some of them were leading the DAS is dead campaign in favor of iSCSI or FC or NAS a few years ago. However simply comparing DAS to SAN or NAS in a competing way is like comparing apples to oranges, instead, look at how and where they can complement and enable each other. In other words, different tools for various tasks, various storage and interfaces for different needs.

    Thus IMHO DAS never left or went anywhere per say, it just was not fashionable or cool to talk about until now as it is cool and trend to discuss it again.

    Ok, nuff said for now.

    Cheers Gs

    Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press, 2011), The Green and Virtual Data Center (CRC Press, 2009), and Resilient Storage Networks (Elsevier, 2004)

    twitter @storageio

    All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2012 StorageIO and UnlimitedIO All Rights Reserved

    Posted on

    What industry pundits love and loathe about data storage

    Drew Robb has a good article about what IT industry pundits including vendors, analysts, and advisors loath including comments from myself.

    In the article Drew asks: What do you really love about storage and what are your pet peeves?

    One of my comments and perspectives is that I like Hybrid Hard Disk Drives (HHDDs) in addition to traditional Hard Disk Drives (HDD) along with Solid State Devices (SSDs). As much as I like HHDDs, I also believe that with any technology, they are not the best solution for everything, however they can also be used in many ways than being seen. Here is the fifth installment of a series on HHDDs that I have done since June 2010 when I received my first HHDD a Seagate Momentus XT. You can read the other installments of my momentus moments here, here, here and here.

    Seagate Momentus XT
    HHDD with integrated nand flash SSD photo courtesy Seagate.com

    Molly Rector VP of marketing at tape summit resources vendor Spectra Logic mentioned that what she does not like is companies that base their business plan on patent law trolling. I would have expected something different along the lines of countering or correcting people that say tape sucks, tape is dead, or that tape is the cause problem of anything wrong with storage thus clearing the air or putting up a fight that tape summit resources. Go figure…

    Another of my comments involved clouds of which there are plenty of conversations taking place. I do like clouds (I even recently wrote a book involving them) however Im a fan of using them where applicable to coexist and enhance other IT resources. Dont be scared of clouds, however be ready, do your homework, listen, learn, do proof of concepts to decide best practices, when, where, what and how to use them.

    Speaking of clouds, click here to read about who is responsible for cloud data loss and cast your vote, along with viewing what do you think about IT clouds in general here.

    Mike Karp (aka twitter @storagewonk ) an analyst with Ptak Noel mentions that midrange environments dont get respect from big (or even startup) vendors.

    I would take that a step further by saying compared to six or so years ago, SMB are getting night and day better respect along with attention by most vendors, however what is lacking is respect of the SOHO sector (e.g. lower end of SMB down to or just above consumer).

    Granted some that have traditional sold into those sectors such as server vendors including Dell and HP get it or at least see the potential along with traditional enterprise vendor EMC via its Iomega . Yet I still see many vendors including startups in general discounting, shrugging off or sneering at the SOHO space similar to those who dissed or did not respect the SMB space several years ago. Similar to the SMB space, SOHO requires different products, packaging, pricing and routes to market via channel or etail mechanisms which means change for some vendors. Those vendors who embraced the SMB and realized what needed to change to adapt to those markets will also stand to do better with the SOHO.

    Here is the reason that I think SOHO needs respect.

    Simple, SOHOs grow up to become SMBs, SMBs grow up to become SMEs, SMEs grow up to become enterprises and not to mention that the amount of data being generated, moved, processed and stored continues to grow. The net result is that SMBs along with SOHO storage demands will continue to grow and for those vendors who can adjust to support those markets will also stand to gain new customers that in turn can become plans for other solution offerings.

    Cloud conversations

    Not surprising Eran Farajun of Asigra which has been doing cloud backups decades before they were known as clouds loves backup (and restores). However I am surprised that Eran did not jump on the its time to modernize and re architect data protection theme. Oh well, will have to have a chat with Eran on that sometime.

    What was surprising were comments from Panzura who has a good distributed (e.g. read also cloud) file system that can be used for various things including online reference data. Panzura has a solution that normally I would not even think about in the context of being pulled into a Datadomain or dedupe appliance type discussion (e.g tape sucks or other similar themes). So it is odd that they are playing to the tape sucks camp and theme vs. playing to where the technology can really shine which IMHO is in the global, distributed, scale out and cloud file system space. Oh well, I guess you go with what you know or has worked in the past to get some attention.

    Molly Rector of Spectra also mentioned that she likes High Performance Computing, surprised that she did not throw in high productivity computing as well in conjunction with big data, big bandwidth, green, dedupe, power, disk, tape and related buzzword bingo terms.

    Also there are some comments from myself about cost cutting.

    While I see the need for organizations to cut costs during tough economic times, Im not a fan of simply cutting cost for the sake of cost cutting as opposed to finding and removing complexity that in turn remove costs of doing work. In other words, Im a fan of finding and removing waste, becoming more effective and productive along with removing the cost of doing a particular piece of work. This in the end meets the aim of bean counters to cut costs, however can be done in a way that does not degrade service levels or customer service experience. For example instead of looking to cut backup costs, do you know where the real costs of doing data protection exist (hint swapping out media is treating the symptoms) and if so, what can be done to streamline those from the source of the problem downstream to the target (e.g. media or medium). In other words, redesign, review, modernize how data protection is done, leverage data footprint reduction (DFR) techniques including archive, compression, consolidation, data management, dedupe and other technologies in effective and creative ways, after all, return on innovation is the new ROI.

    Checkout Drews article here to read more on the above topics and themes.

    Ok, nuff said for now

    Cheers gs

    Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press, 2011), The Green and Virtual Data Center (CRC Press, 2009), and Resilient Storage Networks (Elsevier, 2004)

    twitter @storageio

    All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2011 StorageIO and UnlimitedIO All Rights Reserved

    Posted on

    Does IBM Power7 processor announcement signal storage upgrades?

    IBM recently announced the Power7 as the latest generation of processors that the company uses in some of its mid range and high end compute servers including the iSeries and pSeries.


    IBM Power7 processor wafers (chips)

     

    What is the Power7 processor?
    The Power7 is the latest generation of IBM processors (chips) that are used as the CPUs in IBM mid range and high end open systems (pSeries) for Unix (AIX) and Linux as well as for the iSeries (aka AS400 successor). Building on previous Power series processors, the Power7 increases the performance per core (CPU) along with the number of cores per socket (chip) footprint. For example, each Power7 chip that plugs into a socket on a processor card in a server can have up to 8 cores or CPUs. Note that sometimes cores are also known as micro CPUs as well as virtual CPUs not to be confused with their presented via Hypervisor abstraction.

    Sometimes you may also here the term or phrase 2 way, 4 way (not to be confused with a Cincinnati style 4 way chili) or 8 way among others that refers to the number of cores on a chip. Hence, a dual 2 way would be a pair of processor chips each with 2 cores while a quad 8 way would be 4 processors chips each with 8 cores and so on.


    IBM Power7 with up to eight cores per processor (chip)

    In addition to faster and more cores in a denser footprint, there are also energy efficiency enhancements including Energy Star for enterprise servers qualification along with intelligent power management (IPM also see here) implementation. IPM is implanted in what IBM refers to as Intelligent Energy technology for turning on or off various parts of the system along with varying processor clock speeds. The benefit is when there is work to be done, get it down quickly or if there is less work, turn some cores off or slow clock speed down. This is similar to what other industry leaders including Intel have deployed with their Nehalem series of processors that also support IPM.

    Additional features of the Power7 include (varies by system solutions):

    • Energy Star for server qualified providing enhanced performance and efficiency.

    • IBM Systems Director Express, Standard and Enterprise Editions for simplified management including virtualization capabilities across pools of Power servers as a single entity.

    • PowerVM (Hypervisor) virtualization for AIX, iSeries and Linux operating systems.

    • ActiveMemory enables effective memory capacity to be larger than physical memory, similar to how virtual memory works within many operating systems. The benefit is to enable a partition to have access to more memory which is important for virtual machines along with the ability to support more partitions in a given physical memory footprint.

    • TurboCore and Intelligent Threads enable workload optimization by selecting the applicable mode for the work to be done. For example, single thread per core along with simultaneous threads (2 or 4) modes per core. The trade off is to have more threads per core for concurrent processing, or, fewer threads to boost single stream performance.

    IBM has announced several Power7 enabled or based server system models with various numbers of processors and cores along with standalone and clustered configurations including:

    IBM Power7 family of server systems

    • Power 750 Express, 4U server with one to four socket server supporting up to 32 cores (3.0 to 3.5 GHz) and 128 threads (4 threads per core), PowerVM (Hypervisor) along with main memory capacity of 512GB or 1TByte of virtual memory using Active Memory Expansion.
    • Power 755, 32 3.3Ghz Power7 cores (8 cores per processor) with memory up to 256GB along with AltiVec and VSX SIMD instruction set support. Up to 64 755 nodes each with 32 cores can be clustered together for high performance applications.
    • Power 770, Up to 64 Power7 cores providing more performance while consuming less energy per core compared to previous Power6 generations. Support for up to 2TB of main memory or RAM using 32GB DIMM when available later in 2010.
    • Power 780, 64 Power7 cores with TurboCore workload optimization providing performance boost per core. With TurboCore, 64 cores can operate at 3.8 GHz, or, enable up to 32 cores at 4.1 GHz and twice the amount of cache when more speed per thread is needed. Support for up to 2TB of main memory or RAM using 32GB DIMM when available later in 2010.

    Additional Power7 specifications and details can be found here.

     

    What is the DS8000?
    The DS8000 is the latest generation of a family of high end enterprise class storage systems supporting IBM mainframe (zSeries), Open systems along with mixed workloads. Being high end open systems or mainframe, the DS8000 competes with similar systems from EMC (Symmetrix/DMX/VMAX), Fujitsu (Eternus DX8000), HDS (Hitachi) and HP (XP series OEM from Hitachi). Previous generations of the DS8000 (aka predecessors) include the ESS (Enterprise Storage System) Model 2105 (aka Shark) and VSS (Versatile Storage Server). Current generation family members include the Power5 based DS8100 and DS8300 along with the Power6 based DS8700.

    IBM DS8000 Storage System

    Learn more about the DS8000 here, here, here and here.

     

    What is the association between the Power7 and DS8000?
    Disclosure: Before I go any further, lets be clear on something, what I am about to post on is based entirely on researching, analyzing, correlating (connecting the dots) of what is publicly and freely available from IBM on the Web (e.g. there is no NDA material being disclosed here that I am aware of) along with prior trends and tendency of IBM and their solutions. In other words, you can call it speculation, a prediction, industry analysis perspective, looking into the proverbial crystal ball or educated guess and thus should not be taken as an indicator of what IBM may actually do or be working on. As to what may actually be done or not done, for that you will need to contact one of the IBM truth squad members.

    As to what is the linkage between Power7 and the DS8000?

    The linkage between the Power7 and the DS8000 is just that, the Power processors!

    At the heart of the DS8000 are Power series processors coupled or clustered together in pairs for performance and availability that run IBM developed storage systems software. While the spin doctors may not agree, essentially the DS8000 and its predecessors are based on and around Power series processors clustered together with a high speed interconnect that combine to host an operating system and IBM developed storage system application software.

    Thus IBM has been able to for over a decade leverage technology improvement curve advantages with faster processors, increased memory and I/O connectivity in denser footprints while enhancing their storage system application software.

    Given that the current DS8000 family members utilize 2 way (2 core) or 4 way (4 core) Power5 and Power6 processors, similar to how their predecessors utilized previous generation Power4, Power3 and so forth processors, it only makes sense that IBM might possibly use a Power7 processor in a future DS8000 (or derivative perhaps even with a different name or model number). Again, this is just based all on historical trends and patterns of IBM storage systems group leveraging the latest generation of Power processors; after all, they are a large customer of the Power systems group.

    Consequently it would make sense for IBM storage folks to leverage the new Power7 processors and features similar to how EMC is leveraging Intel processor enhances along with what other vendors are doing.

    There is certainly room in the DS8000 architecture for growth in terms of supporting additional nodes or complexes or controllers (or whatever your term preference of choice is for describing a server) each equipped with multiple processors (chips or sockets) that have multiple cores. While IBM has only commercially released two complex or dual server versions of the DS8000 with various numbers of cores per server, they have come nowhere close to their architecture limit of nodes. In fact with this release of Power7, as an example, the model 755 can be clustered via InfiniBand with up to 64 nodes, with each node having 4 sockets (e.g. 4 way) with up to 8 cores each. That means on paper, 64 x 4 x 8 = 2048 cores and each core could have up to 4 threads for concurrency, or half as many cores for more cache performance. Now will IBM ever come out with a 64 node DS8000 on steroids?

    Tough to say, maybe possibly some day to play specmanship vs EMC VMAX 256 node architectural limit, however Im not holding my breath just yet. Thus with more and faster cores per processor, ability to increase number of processors per server or node, along with architectural capabilities to boost the number of nodes in an instance or cluster, on paper alone, there is lots of head room for the DS8000 or a future derivative.

    What about software and functionality, sure IBM could in theory simply turn the crank and use a new hardware platform that is faster, more capacity, denser, better energy efficiency, however what about new features?

    Can IBM enhance its storage systems application software that it evolved from the ESS with new features to leverage underlying hardware capabilities including TurboCore, PowerVM, device and I/O sharing, Intelligent Energy efficiency along with threads enhancements?

    Can IBM leverage those and other features to support not only scaling of performance, availability, capacity and energy efficiency in an economical manner, however also add features for advanced automated tiering or data movement plus other popular industry buzzword functionality?

     

    Additional thoughts and perspectives
    One of the things I find interesting is that some IBM folks along with their channel partners will go to great lengths to explain why and how the DS8000 is not just a pair of Power enabled based servers tightly coupled together. Yet, on the other hand, some of those folks will go to great lengths touting the advantages of leveraging off the shelf or commercial enabled servers based on Intel or AMD based systems such as IBMs own XIV storage solution.

    I can understand in the past when the likes of EMC, Hitachi and Fujitsu were all competing with IBM building bigger and more function rich monolithic systems, however that trend is shifting. The trend now as is being seen with EMC and VMAX is to decouple and leverage more off the shelf commercially available technology combined with custom ASICs where and when needed.

    Thus at a time where more attention and discussion is around clustered, grid, scalable storage systems, will we see or hear the IBM folks change their tune about the architectural scale up and out capabilities of the Power enabled DS8000 family?

    There had been some industry speculation that the DS8000 would be the end of the line if the Power7 had not been released which will now (assuming that IBM leverages the Power7 for storage) shift to if there will be a Power8 or Power9 and so forth.

    From a storage perspective, is the DS8K still relevant?

    I say yes given its installed base and need for IBM to have an enterprise solution (sorry, IMHO XIV does not fit that bill just yet) of their own, lest they cut an OEM deal with the likes of Hitachi or Fujitsu which while possible, I do not see it as likely near term. Another soft point on its relevance is to gauge reaction from their competitors including EMC and HDS.

    From a server perspective, what is the benefit of the new Power7 enabled servers from IBM?

    Simple, increase scale of performance for single thread as well as concurrent or parallel application workloads.

    In other words, supporting more web sites, partitions for virtual machines and guest operating system instances, databases, compute and other applications that demand performance and economy of scale.

    This also means that IBM has a platform to aggressively go after Sun Solaris server customers with a lifeline during the Oracle transition, not to mention being a platform for running Oracle in addition to its own UDB/DB2 database. In addition to being a platform for Unix AIX as well as Linux, the Power7 series also are at the heart of current generation iSeries (the server formerly known as the AS400).

    Additional links and resources:

    Closing comments (for now):
    Given IBMs history of following a Power chip enhancement with a new upgraded version of the DS8000 (or ESS/2105 aka Shark/VSS) and its predecessors by a reasonable amount of time, I would be surprised if we do not see a new DS8000 (perhaps even renamed or renumbered) within the year.

    This is similar to how other vendors leverage new processor chip technology evolution to pace their systems upgrades for example how many vendors who leverage Intel processes have done announcements over the past year since the Nehalem series rolled out including EMC among others.

    Lets see what the IBM truth squads have to say, or, not have to say :)

    Ok, nuff said.

    Cheers gs

    Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press) and Resilient Storage Networks (Elsevier)
    twitter @storageio

    All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved

    Posted on

    Green IT Confusion Continues, Opportunities Missed!

    I continue to see those looking for fast silver bullets in the quest to be green, efficient, optimized or sustainable while addressing issues ranging from power/energy, cooling, floor-space/footprint, EH&S (environmental health & safety) not to mention recycling. Yet, I’m also continued to be  amazed by the focus and emphasis around reduce as in reduce your capacity and your performance or processing capabilities in the form of consolidation or aggregation along with energy avoidance which for some is applicable.

    However, there is also the other side of the tale which is shifting from avoidance to becoming more efficient, that is doing more with what you have or with less while boosting productivity. For example, having a server or processor that can do more work in the same or smaller physical footprint drawing the same or less energy and requiring less cooling is a form of reducing overall impact yet boosting productivity. The same can be done with data and I/O networks, storage and even software.

    Similar to automobiles after the 1970s oil and energy crisis, the focus was on reduction, conservation and avoidance as the form of being efficient. Over time, this approach gave way to levering more efficient engines and vehicles that boosted the MPG city and highway, change in driving or usage habits, awareness of issues including applicable metrics and energy costs, as well as the continuing quest for alternative fuels.

    This is no different than what is happening with the IT organizations or compute focused entities in that there has been an initial focus of avoidance to meet short term tactical requirements, not to mention all of the green hype of a few years ago. Today there is a shift taking place towards efficiency and awareness that optimization and efficiency is more than consolidation, that it also includes boosting productivity as part of achieving reduced energy and cooling demands.

    How this can be done is to leverage multiple different techniques including new servers with processors that have intelligent power management (IPM) also known as adaptive voltage scaling (AVS) or other marketing terms enabling variable performance and energy consumption. For example, vary clock cycles and turn on cores when needed, then to turn off cores, slow clock speed down when there is less work to be done. Likewise there are improvements with cooling closer to the heat source ranging from leveraging inert liquid cooling inside the cabinet of computers to surface attached cooling to emerging micro cooling located inside silicon. There is a fascination with using virtualization to consolidate and reduce servers that are underutilized, which again is applicable for some environments and applications.

    However not all servers including many that are underutilized lend themselves to being consolidated for various reasons including quality of service (QoS) or performance, security, vendor support or software compatibility, politics or finance among others. This however does not mean that they cannot be virtualized, it more than likely mean that they cannot be consolidated. There is a common myth that virtualization equals consolidation and vice versa, however virtualization can also be used for abstraction, transparency, emulation and enabling agility including support for load-balancing, scale-up and scale-out performance oriented clustering among other uses. Thus there is another side of virtualization and that is to achieve   efficiency, life beyond consolidation.

    Needless to say there are many more technologies and techniques to address various issues now along with those that are emerging. The good news in all of this is the growing awareness that there are many different faces or facets of being green. That green wash and green hype may be on the endangered species list, that green means more than reducing carbon footprints or recycling or energy avoidance. That green is really about shifting and becoming more efficient, more optimized to support more processing, more work in a cost effective manner to sustain growth on a go forward basis. For high performance compute (HPC) or other large scale IT organizations, there is a notion that small improvements on a large broad scale have significant impact.

    Some organizations are in pursuit of technologies of solutions that promise significant saving ratios over small sets or instances, solutions that provide  smaller reduction or savings over a larger basis can prove to be more effective. For example, if power is a concern, powering down servers or storage that promises 85-100% savings might only be applicable to less than 5% of the devices. However, if 85-100% of the devices can be upgraded to newer models that boost productivity by 5-15% (or more) in the same or smaller footprint, using 5-15% (or more) less power, the results add up quickly. Think of it this way, a 1% saving for an environment using 1,000 kilo watt hour (kWh) or 1mWh of energy is a savings of 10kWh. The point being that for large environments, don’t forget to look at small savings that apply to a large installed base that then add up to big benefits.

    The net result is that one can pursue being green or being perceived as being green which can have a high cost, or, can pursue various efficiency that help the overall organization by boosting productivity, helping the top and bottom line, doing more in a smaller footprint and guess what, the result is not only economic, it’s also environmental positive. Thus, the byproduct of shifting towards efficiency (and not just avoidance) is to become green!

    Ok, nuff said.

    Cheers gs

    Greg Schulz – Author Cloud and Virtual Data Storage Networking (CRC Press), The Green and Virtual Data Center (CRC Press) and Resilient Storage Networks (Elsevier)
    twitter @storageio

    All Comments, (C) and (TM) belong to their owners/posters, Other content (C) Copyright 2006-2024 Server StorageIO and UnlimitedIO LLC All Rights Reserved