Here are the main topics we’ll cover today. We’ll begin by introducing the new DS8800 model and talk about what’s new and what’s not new about it. Then we’ll give you an idea about the various benefits this new model offers, which are mostly centered around space efficiency and energy efficiency, and, of course, performance. Then, we have a couple other enhancements to talk about on both the DS8800 and DS8700. Then, we’ve reserved some time to share with you some new storage software enhancements in our storage management and encryption key management offerings.
Improving Storage Efficiency is, by far, the best way IBM can help clients manage information growth. In real world examples, both inside IBM and with actual client environments, IBM has demonstrated the ability to reduce physical storage requirements up to 50% . But it’s not just about cost reductions. As recessionary times end, more clients are looking for ways to enable future growth. Capital expenses that help lower long term operating expenses are more interesting now, as clients have more to invest. Investments in automated tiered storage , consolidated scale-out storage , and virtualized storage are valuable because they allow clients to manage growth without complexity. IBM offers more storage tier options for more server platforms than most other major vendors, more consolidation options, and proven storage virtualization solutions. You can see some of the highlights in the boxes on the bottom of the slide.
OK. Here’s the quick summary of what we’re introducing. This new DS8800 model is the fastest, most efficient high-end disk we’ve ever delivered. We’ve given it almost a complete overhaul with new hardware throughout the system. We’ve upgraded the controller from the POWER6 to the POWER6+ processors, and we’ve got faster adapters on the front end and the back end. And probably the most important change is the migration to new small-form-factor enterprise drives, which enable us to offer many more drives in the same footprint with a much lower energy profile. And, of course, we continue to focus on the reliability and high availability of the system, which, along with pure, high-end performance, is one of the most important attributes of any high-end system. The system was announced on October 7 and will be generally available on November 19.
Before we get into the details of what’s new, let me point out that the DS8800 represents the 4 th generation of our flagship disk system, which we introduced in 2004… after several generations of the ESS, or Shark, as many still refer to it. All of these generations were built on the IBM POWER processor architecture, which has enjoyed great success because of its continuous and incremental increase in performance, reliability, scalability, etc. This market-proven legacy has been a big reason the DS8000 has such a high reliability profile and reputation. In fact, the DS8800 has over 85% of the same microcode as the DS8700, even with all the HW updates that we’ve included, which is a real testament to the stability of our microcode. We introduced the DS8700 about a year ago, and the uptake was beyond our expectations. While we don’t publish sales results at the product level, this particular model, or the success of the 8700 has been driven by the performance and the core functionality and, from what our customers tell us, the bullet-proof reliability. You can see on the chart how the DS8000 has benefitted from basing its architecture on the market-proven IBM POWER architecture with the binary compatibility across generations, and we expect this to continue through more generations. LET’S GO TO SLIDE 5
The introduction of this new DS8800 model pretty much marks the evolution away from the earlier generations of DS8300, so we now have two distinct models in the portfolio based on the POWER6 processor family… the DS8700 and the DS8800. If you remember back when we introduced the DS8700, in addition to the POWER6 processor and PCIe internal fabric, we made it much easier for the system to scale up from the smallest configuration to the largest configuration non-disruptively. The DS8800 also has this capability, so both modes can scale up as the clients’ needs grow. As they add new workloads through consolidation and their existing workloads grow, clients can scale both models from the 2-way to the 4-way model non-disruptively, as well as add more capacity and frames and memory… all non-disruptively. Another difference between the two base models is that we’ve upgrade both the host adapters and RAID adapters to 8Gb/s, so we’ve got 8Gb on the front and the back ends. And probably the biggest difference is the introduction of the 2.5”, small-form-factor enterprise SAS-2 drives, which replace the 3.5” drives high-end systems have been using for years. Now, while the DS8800 has all of these hardware upgrades, we’re still expecting clients to continue buying the DS8700s, at least in the near term, since it’s going to be the more cost-effective and field-proven system since we’re pricing it a bit lower than the new model. LET’S MOVE TO SLIDE 6
OK. If there’s one slide that captures what this release is about, this is it. [BUILD SLIDE with the last 3 bullets] We’ve got a new 4 th -generation DS8000 system with all new HW inside… BUT this new hardware is built on the same proven architecture that has served us well for over 10 years. So, this is going to provide faster performance with the upgraded hardware throughout; much greater space efficiency and energy efficiency with the new smaller SAS-2 drives; and provide the same outstanding reliability that our clients continue to tell us is of primary importance.
OK… how about some details? On the right side, you’ll see the inside of the system, which looks very similar to the DS8700 model. The red highlights show where the new upgrades are, so we’ve got the new small-form-factor drives and drive enclosures at the top, the upgraded controllers near the middle, and the new 8Gb adapters at the bottom. Let’s start with the new drives and enclosures This new model will support only small-form-factor 2.5” drives. These will be the 6 Gb/s SAS-2 drives, which will provide not only faster performance than their 3.5” counterparts, but they’ll also provide higher reliability and scalability than the Fibre Channel interface provides a much lower cost. These drives offer all the enterprise attributes of the FC drives but at a lower cost and footprint. This is going to offer our clients outstanding space and energy efficiency benefits and faster performance. The drives offer better performance at the same rotational speeds, plus a 6 Gb/s transfer rate vs. 4 Gb/s, better energy usage per drive, at a lower cost per gigabyte. They also offer outstanding signal integrity and support longer cable lengths up to 10 meters. And since they’re smaller than the 3.5” drives, we’re able to pack many more drives into a frame than ever before. The result, of course, is better performance at a lower cost and with a more compact and efficient footprint. This has clients excited… very excited….mostly about the fact that they can have a dramatic effect on lowering their overall footprint in data center. One client is looking at consolidating at least a dozen older systems onto the DS8800. This has really become the most interesting benefit for most of the clients we’ve spoken with. In terms of options, we’ll be offering these 2.5” drives in SSD, 15k RPM and 10k RPM options Additional notes: Regarding SATA (7.2k RPM) options for 2.5” drives, I believe I saw that Seagate currently offers capacities up to 500GB, so we’re in no hurry support these until the capacities get much larger. And then we’ll have to compare these to the 10k options they have to see what makes the most sense. Drive Roadmap - Higher capacity options are expected for each family; Nearline SAS options will also be available to replace existing, 3.5” SATA capacities Next, let’s talk about the upgraded POWER6+ controllers As we move from POWER6 to POWER6+, we take advantage mostly of the increased chip speed. The clock speed is going from 4.7 Ghz to 5 Ghz, and we’re seeing faster IOPS and overall throughput. This P6 line also enables the system to support higher overall capacities, which will enable more consolidation of workloads and applications onto one system. At the bottom, we’ve got the upgraded 8Gb/s host and device adapters, which are connected to the PCI Express (PCI-E) internal fabric, which also contributes to higher performance and scalability. host adapters We’re making available both four-port and eight-port options, and these new adapters provide up to 100% improvement in single-port performance and up to 400% improvement in single adapter throughput. They’re also include dual core processors and have 4 times the memory of the DS8700’s adapters. This increased throughput is likely going to reduce the number of host ports needed, which is going to mean lower costs for clients These adapters will support attachment to FICON Express8 on our zEnterprise and z10 systems. device adapters We’ll offer four-port models, which is the same as what we offer on the DS8700, but these new DAs will offer up to 4 times the throughput of the previous adapters (single adapter). Like the host adapters, they’ll include dual core processors, which will enable faster and more effective automated tiering with Easy Tier since the system will be able to accelerate the movement of data across the various drives tiers. they also have 4 times the memory of those in the DS8700. And the new adapters also offer the ability to scale the system to much higher capacity and more controllers in the future. With respect to airflow changes , the system now has complete front-to-back airflow, which can help clients deploy a consistent hot aisle / cold aisle data center design. The airflow in the DS8700 is slightly different. The processor complexes and I/O enclosures in the bottom half of the frame took cool air from the front and sent hot air out the back, but the drive enclosures at the top took cool air from the front and the back and sent the hot air out of the top. This new front-to-back design can help clients design a more energy efficient data center with a hot-aisle/cold-aisle design. And since energy efficiency has been a hot topic for most clients, they’ve been telling us this is good news and that this should help with their power needs. Additional notes: Device adapter interconnect remains Fibre Channel In order to take advantage of DS8700 architecture, Gigapack connects to the DA via Fibre Channel protocol and converts to SAS protocol through the c ontrol card switch in the RAID adapter Improved backbone throughput, increased from 2Gbps to 8Gbps Moved to optical cabling Allows for increased cabling distance, especially at 8Gbps speeds Believed to be improved reliability over passive copper cabling Power distribution unit (PDU) is the upper left
Since the biggest change in this new release is the new high-density drives and enclosures, I thought it would be helpful to take a look at how they differ from the DS8700 drive enclosures. Integrated power and cooling allows us to move more efficiently with the new front-to-back cooling design. [Another smaller benefit is that it provides more redundancy if a fan our power supply fails. If this happens, we can quickly replace that power supply or fan without affecting the other enclosure.]
The effect of these new changes are the dramatic increase in space efficiency and energy efficiency, which, should dramatically lower costs in the data center. You can see on the right that a fully-configured DS8700 can hold 1,024 drives in 5 frames. The DS8800 can hold slightly more than that in only 3 frames, which is a 40% reduction in footprint. In fact, each frame can hold considerably more drives than its DS8700 counterpart. For example, the base frame and expansion frames can now support 88% more drives than prior models (base: 240 drives compared to 128; first expansion frame: 336 drives compared to 256 for DS8700; second expansion frame: 480 drives compared to 256 for DS8700 ). The benefits of this space and energy efficiency, of course, is the consolidation that clients can put into place by replacing many of their older systems with a handful of these new systems, and the energy savings that go with it. They’ll have a much smaller footprint to power and cool and the costs of managing few systems is going to also help lower operating costs. And client feedback has been overwhelmingly positive on this. We’ve spoken with clients that are preparing for a major consolidation with this new system where they’re planning on replacing a lot of their older systems with these newer ones.
Here we have a comparison of the energy usage for two fully-configured systems. You can see the new system uses over a third less energy than the existing DS8700 that we tested with. While we didn’t calculate the total with a DS8300 model, the energy savings that come from replacing these older models is more compelling since they would require a larger configuration to match the performance we’re delivering with the new DS8800 model. [READ summary findings at bottom of chart on money saved] Contributors to energy efficiency include: The new small-form-factor drives which use 10.2 watts, compared to 19.4 watts each in the DS8700 (90% reduction) The new, high-density disk enclosures , or gigapacks, which use 245 watts in the DS8800 vs. 310 watts each in the DS8700 (27%), despite the fact that the new enclosures hold 50% more drives Compares 15k drives vs. 10k drives running at maximum I/O (capacity does not affect comparison) Includes controller card power, power for cooling fans, and power for disks, but most efficiencies result from new drives) Assumptions in savings calculations: Floor space costs in a data center or even in an office can add up since the space taken by your storage is not usable for any other general business purposes. Your year 1 2010IBM DS8700 Turbo floor space costs are based on 5 racks that each require 65 square feet of floor space for a total of 325 square feet. Consumption of electricity to power and cool your storage can add up, especially if you have a large number of disk drives or disk arrays. Your year 1 2010IBM DS8700 Turbo power and cooling costs are based on 4,400 Watts of input power for your base controller, array frames, and disk expansion units plus 13,824 Watts of input power for your hard disks. Annual power and cooling requirements are based on the following formulas: Power : 1 Watt / 1000 Watts per kiloWatt * 8760 hours/year Cooling : 1 Watt * 3.413 BTU per hour per Watt / 10 BTU cooled per Watt according to average SEER enery ratings / 1000 Watts per kiloWatt * 8760 hours per year Frame comparison details A rack - 240 disks in DS8800 vs. 128 disks B rack – 336 disks vs. 256 disks DS8800 shows better energy per performance Energy per GB, DS8800 is better on energy usage / GB (nearline drives are not an option for the DS8800) Both cases are fully configured. DS8700 extrapolated from 3 frame to 5 frame. Compares 15KRPM in DS8700 to 15KRPM in DS8800. R5 DS8700 15K RPM A-rack: 6.8kW B-rack: 7.1kW C-rack: 6.1kW C-rack: 6.1kW C-rack: 3 kW -----------> Total 29.1KW R6 DS8800 10KRPM A-rack: 6.8kW B-rack: 5.4kW C-rack: 6.5kW ------------> Total 18.7 KW Approx 36% power reduction Recent measurement DS8700 extrapolated from 3 frame to 5 frame. Compare with 15k RPM in both DS87000 and DS8800 R5 DS8700 15KRPM A-rack: 6.8kW B-rack: 7.2kW C-rack: 6.4kW C-rack: 6.4kW C-rack: 3.2kW -------------> total 30KW R6 DS8800 15KRPM A-rack: 7.5kW B-rack: 6.2kW C-rack: 7.3kW -----------> Total 21 KW Approx 30% power reduction
In addition to the energy efficiency at the system level, the DS8000’s front-to-back airflow will enable clients to design their data centers with a consistent hot-aisle / cold-aisle configuration. As IT vendors continue to offer front-to-back cooling designs, this will enable clients to become even more energy efficient.
DBOpen 70/30/50 (IOs/sec)
Here are some additional enhancements we’ve made on both the DS8800, as well as the DS8700. We announced Storage Pool Striping on the DS8300 several years ago, and this this feature was designed to automatically eliminate hot spots by spreading volumes across multiple RAID groups, rather than within a single RAID group. What’s new on the DS8800 is that we’re now setting this as the default for new volumes. So clients get this hotspot reduction automatically without having to configure the system. Unlike our competitors, this supports both open- and mainframe-formatted data. Looking at the wider perspective, this enhancement extends our leadership in simplifying the way clients manage their volumes, which is becoming more and more important as we’re seeing more and more clients are asking us to help them manage more data with less resources. Hopefully, you’re already familiar with the way we allocate capacity in the DS8000 into 1GB chunks, or extents, as we call them. These 1GB building blocks are much more efficient than the way our competitors allocate volumes from available capacity. If you think about how their volumes are allocated, they need to create volumes from contiguous blocks on a physical disk, (i.e., Symmetrix hypervolumes), and this method can lead to a lot of wasted capacity since it leaves a lot of random and isolated islands of unused capacity between the actual volumes the client has created. These random islands of unused capacity between existing volumes might never be used because the client might not have a volume that fits nicely into that space. And this becomes more wasteful as more and more volumes are created and deleted over time. Another volume management advantage DS8000 users have is the way we allow for non-disruptive volume expansion. Since we create volumes from consistent 1GB extents in the storage pool, our Dynamic Volume Expansion feature can expand volumes simply by adding available extents from the storage pool to an existing volume. Just specify the existing volume name and the new size volume you need… and viola! That’s it! The competitors don't have an extent-based design. So the high-end EMC and Hitachi systems aren’t as efficient as ours. When a user wants to expand an existing volume, they would need to create a brand new volume that matches the size of the increase they want… then they have &quot;concatenate“, or link, the two volumes together, which the system now pretends is a single volume. When a client wants to expand the volume again, they’d have to create another brand new volume of the size they desire and concatenate that new volume to the others they’ve already created. The end result might seem the same as how IBM does it, but it’s not. It’s like expanding your house by buying the house across the street, when all you really want to do is build another floor on top of your existing house. Clearly, our volume management is much more efficient. The IBM System Storage DS8800 business class cabling feature offers a streamlined, lower cost configuration than the standard configuration. It essentially reduces the number of device adapters and I/O enclosures while increasing the number of drive enclosures attached to the remaining device adapters. This allows for a cost-competitive configuration that aligns with the special needs of our clients who want smaller, cost-effective configurations. ******************* Additional competitive notes on Storage Pool Striping: Storage Pool Striping : Everyone does this for thin provisioned volumes. However, for any and all volume types, we’re the only one that offer this in a simple way by striping across the whole pool. With Symm, you got to different types of volumes on different disks and link them together manually, and this only works for open volumes, so no wide striping for mainframes. Now, this is our default. Volume expansion in DS8000 is easier than in EMC or HDS because we don't have to create new volumes then link them to the existing volumes. And DS8 supports this for open and MF, while Symm/Hit don't have DVE for mainframe. The only advantage Symm/Hit really have for managing volumes is the maximum size of a volume (60TB for both Symm/Hit, vs. 2TB for DS8)
06/01/09 *Business Continuity solution for AIX environments Description Open HyperSwap for AIX/DS8700 allows I/O to the primary volumes in a Metro Mirror relationship to be swapped, without human intervention, to the secondary volumes in the Metro Mirror relationship, with minimal disruption to the applications. The intent is for applications to see nothing more than a brief pause in their I/O. Once the swap has occurred, the application is automatically writing to the secondary site without noticing the switch. Supported Configuration Multiple AIX hosts supported in one TPC-R session DS8700 Release 5.1 microcode Supported on AIX 5.3 and 6.1 Requires new version of SDDPCM to be installed
Elevator pitch for the new DS8700.
IBM System Storage DS8800 (Release 6.0) October, 2010 Client Presentation October 7, 2010 Allen Marin, IBM Enterprise Disk Marketing
Storage Efficiency Getting the most from your storage resources Virtualized Storage Tiered Storage Consolidated Storage Efficiently use solid state disk to increase performance up to 300% on critical apps Reduce costs by migrating less critical data to less expensive media Build out an automated tiered storage architecture to maximize performance and reduce operating expenses Consolidate storage to reduce administrative costs and improve cycle time Increase utilization of existing storage by up to 30% Reduce TCO up to 66% by deploying automated, virtualized storage Quickly provision storage, support virtualized servers and increase flexibility Scale out storage to manage billions of files from a single, consolidated NAS system Scale up storage with systems that can scale to store petabytes of structured data
DS8800 4 th -generation DS8000 enterprise disk platform
4 th -generation DS8000 enterprise disk system Binary Compatibility 2004 2009 POWER5 POWER6 The IBM POWER processor has been behind the success of IBM enterprise storage beginning with the Enterprise Storage Server in 1999 2010 POWER6+ DS8800 DS8000 DS8700 2006 POWER5+ DS8000 Turbo DS8800 builds on a market-proven, reliable code base!
Client feedback is very positive on space-saving design
High-density drive enclosures
Almost double the drives in same frame footprint
More effective consolidation can lower operating costs
Support more workloads with smaller footprint
Reduce number of systems to manage
Reduce power and cooling costs
Saving money with high-density drives, enclosures, frames Extremely positive c lient feedback about substantial footprint reduction 1,024 drives 1,056 drives More drives in 60% floor space DS8700 DS8800
~ 90% more drives in a single frame
More drives in 60% floor space fully-configured
Energy consumption comparison with DS8700 and DS8300*
DS8800 with 1056 drives
Base frame: 6.8kW
Exp frame: 5.4kW
Exp frame : 6.5kW
DS8700 with 1024 drives
Base frame: 6.8kW
Exp frame: 7.1kW
Exp frame: 6.1kW
Exp frame: 6.1kW
Exp frame: 3.0kW
36% less energy usage; 40% less floor space
$41,336 less for power/cooling
$71,624 less for floor space
Saves client $112,960 3-year cost savings * Figures are based on IBM lab estimates; For planning purposes with client, please use peak power figures listed in the DS8000 Introduction and Planning Guide, which add safety margin to protect client installations.
New airflow design is also more energy efficient
More data centers are moving to hot aisle / cold aisle designs to optimize energy efficiency
DS8800 is now designed with complete front-to-back airflow
Front-to-back airflow for hot-aisle-cold-aisle data centers
Benefit: Greater energy efficiency and contributes to lower energy costs
Performance comparison across DS8000 models Open Results Summary DS8800 Full Box Results – 96x RAID5 Arrays 768x 15K RPM HDDs, 16x SSDs, 8x DA Pair, 16x HA w/ 32x 8Gb ports DS8300 DS8700 DS8800 % increase (vs. DS8300) Seq Read GBps 3.9 9.7 11.5 19% (195%) Seq Write GBps 2.2 4.7 6.7 43% (205%) Database Open 4KB K IOps 165 191 196 3% (19%) 4K Read Miss 4KB K IOps 111 137 160 17% (44%) 4K Read Hits 4KB K IOps 425 523 530 1% (25%) 4K Write Hits 4KB K IOps 164 203 222 9% (35%) DS8800 Component Results – 4port HA; RAID5 arrays DS8700 DS8800 % increase HA 4KB Read 4KB K IOps 78 190 144% HA 4KB Write 4KB K IOps 42 110 162% HA 64KB Read GBps .53 2.53 377% HA 64KB Write GBps .40 1.42 255% DA Pair 4KB Read 4KB K IOps 53 77 45% DA Pair 4KB Write 4KB K IOps 19 32 68% DA Pair 64KB Read GBps 1.2 3.2 167% DA Pair 64KB Write GBps 0.77 2.2 186%
KMIP – Key Management Interoperability Protocol V1.0 standard from OASIS
Partners include Emulex, Brocade,…
RBAC – Role Based Access Control
Grouping of permissions
Grouping of Devices
Grouping of Administrators
Who can do what action
What device can see what keys
Continues with Ease of Use, simple install and administration
Key store, key serving, key rotation, auditing
IBM is rated in the top 5 of “Most Trusted Companies in [Data] Privacy”* Advanced security for data at rest
Announcement Summary The DS8800 announcement introduces the most advanced model in our high-end disk portfolio with another increase in performance, greatly improved energy efficiency, and almost double the capacity in the same footprint. This new model not only offers the ability to process information faster and more efficiently, it also builds on the DS8000’s unrivaled reputation for reliability and investment protection by maintaining its market-proven, IBM POWER-based architecture over generations of new models.
Trademarks and Disclaimers IBM Corporation 1994-2010. All rights reserved. References in this document to IBM products or services do not imply that IBM intends to make them available in every country. Trademarks of International Business Machines Corporation in the United States, other countries, or both can be found on the World Wide Web at http://www.ibm.com/legal/copytrade.shtml . Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both. Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the United States, other countries, or both. UNIX is a registered trademark of The Open Group in the United States and other countries. Java and all Java-based trademarks are trademarks of Sun Microsystems, Inc. in the United States, other countries, or both. Other company, product, or service names may be trademarks or service marks of others. Information is provided "AS IS" without warranty of any kind. The customer examples described are presented as illustrations of how those customers have used IBM products and the results they may have achieved. Actual environmental costs and performance characteristics may vary by customer. Information concerning non-IBM products was obtained from a supplier of these products, published announcement material, or other publicly available sources and does not constitute an endorsement of such products by IBM. Sources for non-IBM list prices and performance numbers are taken from publicly available information, including vendor announcements and vendor worldwide homepages. IBM has not tested these products and cannot confirm the accuracy of performance, capability, or any other claims related to non-IBM products. Questions on the capability of non-IBM products should be addressed to the supplier of those products. All statements regarding IBM future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only. Some information addresses anticipated future capabilities. Such information is not intended as a definitive statement of a commitment to specific levels of performance, function or delivery schedules with respect to any future products. Such commitments are only made in IBM product announcements. The information is presented here to communicate IBM's current investment and development activities as a good faith effort to help with our customers' future planning. Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput or performance that any user will experience will vary depending upon considerations such as the amount of multiprogramming in the user's job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be given that an individual user will achieve throughput or performance improvements equivalent to the ratios stated here. Photographs shown may be engineering prototypes. Changes may be incorporated in production models.