Over 4,000 systems sold in less than two years since FlashSystem family was announced
The ability to dynamically build and accumulate context at scale from data sources is an opportunity for IBM to deliver new client value
Traditional IT systems have focused processing, recording and managing the core transactions of business operations. These so-called “systems of record” form the backbone of today’s enterprise systems and continue to be essential to the business. But while the structured data, linear processes and highly repeatable operations of todays data warehouses and enterprise application platforms are well suited to their requirements for the business, a new approach has emerged that is focused on the dynamic engagement with customers and partners.
Trends such as mobile and social computing have transformed so-called “systems of engagement” into the digital front office of the business, and require a highly dynamic, more exploratory approach focused on both structured and unstructured data from a wide variety of public and private sources. This high variety of data is a poor fit to traditional fixed database schemas and the relatively brittle systems based on them. Gathering and connecting all these data sources over time is crucial to establishing the data pipeline required for achieving the contextual enterprise.
Storage IT and Storage vendors look at this opportunity and think HW [time it with the last animation] but the reality is that you need more than HW, SW is key and conversations that start with a HW focus will surely fail.
IBM has been studying why and how clients (even those who currently aren’t IBM clients) have deployed Big Data projects, the study keeps going and it is now much more than 1000 examples. This study, together with the 3 year work of the IBM Institute for Business Value have given us unparalleled insights and, most importantly, which are the use cases of big data projects that are yielding most value for clients already.
[quickly list the 5 and tell them we’ll go over in a little bit detail later]
Before we get started and dig deeper, I want to ask the question: Why flash? What is driving this rapid adoption?
We’ve arrived here by way of three technology transformations and one epic problem:
Processor performance increases have followed Moore’s law
Unprecedented growth in concurrent users and transactions, as we mentioned previously in the “systems of engagement”
Rapid adoption of flash in consumer electronics drove up flash density and drove down flash price
But in the meantime, the underlying problem behind all of this is the disk drive. Disk drives continue to increase capacity, but at the expense of performance. In fact, because the performance has remained stagnant, the performance per capacity has actually gone down.
So how have our customers tried to solve this problem?
The first method that our customers use to avoid the performance gap is adding more DRAM to the server, which is costly, but may have some limited benefit.
The second method is to add more CPUs, which is also expensive, both in hardware and the resulting software licensing, and this is ineffective for solving the storage performance issues.
Another dynamic we have seen is disk drive performance enhancement – striping small capacity, high speed drives, with low utilization, large caches, all wasteful, inefficient, and expensive, with limited returns.
And finally, our customers try tuning and modifying the applications, which is VERY time consuming and expensive. It can also be quite risky to re-architect critical applications that your business and your customers depend on…
Not only do these measures not solve the root of the problem, they actually add to many of the IT woes…
Countless software licenses, ever expanding inefficient infrastructure, slow or unreliable application performance… but most importantly dissatisfied users.
As a result, loss of business opportunities
So how IT should be…
We want a balanced infrastructure…
Fewer software licenses
Fewer servers
Energy efficient data centers
Satisfied repeat customers
Stable applications
Feature rich software
Developers working on new capabilities
Fast applications
Predictable performance
Balanced architectures
So how are we going to solve this problem?
We’ve seen that disk drives are a really inefficient solution… Where does flash come into all of this?
No application runs better with storage than it does without storage
Scalable Performance: Tailor solution architectures to your workload; scale-up and scale-out with consistent and predictable IBM MicroLatency™
Enduring Economics : Realize the immediate and long-term economic advantages of FlashCore Technology and virtualization management (Infrastructure Continuity - Future proof your data center by managing, protecting and preserving data)
Agile Integration: Accelerate time to value with an easy to implement, fully integrated system
Feature Rich (Tier 1) Solution Pillars
Dynamic Performance: Tailor solution architectures to your workload; scale-up and scale-out with consistent and predictable IBM MicroLatency™
- Scalable to 2.5 million IOPS
- Scalable to 19.2 GB/s bandwidth
- Sustained IBM MicroLatency™
- Mixed workload consolidation
- Quality of Service
Data Economics: Realize the immediate and long-term economic advantages of IBM FlashCore™ Technology and virtualization management
- Scalable to 2.2PB effective capacity
- Flash for less than the cost of disk with IBM Real-time Compression
- Low power and cooling requirements
- Virtualize storage
Agile Integration: Accelerate time to value with an easy to implement, fully integrated system
- Fully integrated system management
- Application-aware data services
- AES, data at rest encryption
- Tier or mirror to existing storage
Infrastructure Continuity: Future proof your data center by managing, protecting, and preserving data
- Flash Wear Warranty
- Non-disruptive data migrations
- Enable disaster recovery
- Concurrent code load
- Space efficient snapshots, cloning and replication
Performance Focused (Tier 0) Solution Pillars
Extreme Performance: Enable business to unleash the power of performance, scale, and insight to drive services and products to market faster
IBM MicroLatency™: Deliver a microsecond response time to accelerate critical applications to achieve competitive advantages
Macro Efficiency: Driven by consolidation of hardware and software, deployment speed, efficient use of IT staff as well as power and cooling savingsEnterprise Reliability: Durable and reliable designs that use enterprise class flash and patented data protection technology (Variable Stripe RAID)
IBM FlashCore™ Technology
Hardware Accelerated I/O:
- RAID Controllers
- Hot Swappable
- Redundant Backplane Design
- Concurrent Code Loading & Maintenance
- Advance Switch Fabric
MicroLatency Module™:
- Offload Encryption Engine
- Higher Capacity Density
- Massively parallel
- Hot Swappability
- High Speed Interface
Advanced Flash Management:
- Variable Stripe RAID
- IBM Engineered ECC
- Overprovisioning
- Wear Levelling
- Write Buffer & Hardware Offload
- Garbage Collection
With all of this exciting innovation engineered into every FlashSystem, I am happy to introduce IBM’s premier all-flash offering, the IBM FlashSystem 900.
This system will be GA March 20.
This is our next generation, lowest latency offering. These systems still deliver over 1 Million IOPS but with over 40% more capacity and at 10% less the cost of the prior generation. We have also improved bandwidth performance by 25%.
You will here a bit more later, but some of the innovation we have been able to achieve on a the Flash chip level is due to our strategic collaborate with our flash chip vendor, Micron Technology.
We also have greater integration with Vmware which provides system monintoring from Vspehere.
In addition, we are also delivering the ability to manage multiple FlashSystem units from a single GUI interface.
With all of this exciting innovation engineered into every FlashSystem, I am happy to introduce IBM’s premier all-flash offering, the IBM FlashSystem 900.
This system will be GA March 20.
This is our next generation, lowest latency offering. These systems still deliver over 1 Million IOPS but with over 40% more capacity and at 10% less the cost of the prior generation. We have also improved bandwidth performance by 25%.
You will here a bit more later, but some of the innovation we have been able to achieve on a the Flash chip level is due to our strategic collaborate with our flash chip vendor, Micron Technology.
We also have greater integration with Vmware which provides system monintoring from Vspehere.
In addition, we are also delivering the ability to manage multiple FlashSystem units from a single GUI interface.
Support: Up to 7 years support available with optional price protection and flash media retention as well as an optional TA services available during the MA period for at least 5 years.
To dive a little deeper into these decisions it’s important to understand the difference in the design approaches to other flash solutions
Here we are comparing our MicroLatency module to a regular SSD. Might I mention, that many of the all-flash arrays that you will encounter in the market today are populated with these.
The microlatency module that is present in every Flash system are:
CUSTOM, Meaning it requires hardware engineering as opposed to an SSD that is Commodity, which requires no hardware skills
Microlatency modules are FAST because they are Designed for low latency
SSDs on the other hard are notably slower because they are designed for cost.
MicroLatency modules are also RELIABLE because they integrate significant IBM R&D with their advanced flash management
SSDs by comparison are meant for a variety of non-specific cross vendor solutions, meaning that the SSD suppliers are not motivated to make efforts do mitigate failures within the media itself. They are motivated to develop cheap, one-size fits most, modular flash components.
These design decions that make up a key tenant of FlashCore results in
Better Performance (higher IOPS, higher bandwidth, lower latency)
Hardware data path (FPGAs)
Parallelism (more flash chips per module)
More powerful processors
No SAS/SATA protocol
Better connectors (higher bandwidth)
Optimized for performance over low cost
More overprovisioning
Better Reliability
Hardware data path implements Variable Stripe RAID™
Optimized for reliability and not low cost
More overprovisioning improves wear life
Better Density
More flash chips per module
And Lower Price/IOPS
Compared to SSD
So we add Module level RAID protection – By spreading data and parity across flash chips, we can protect data from a chip failure. At this point we have a RAID protected flash module (very few actually go this far with data protection), but we still have problems:
Flash chips fail in a lot of different ways
Flash chips cannot be hot swapped like disk drives
A failed chip puts the flash module in a degraded state and it must be replaced immediately
Variable Stripe RAID - Because the flash chip fails in different ways than a hard disk drive, we can innovate new solutions that improve on our old ways. Variable Stripe RAID – a patented IBM approach to data protection that sees a flash chip as more than a single entity but as a collection of other components: cells, blocks and layers. By implementing Variable Stripe RAID across more granular levels our flash module is more reliable. At this point, when a cell, layer or a flash chip fails, this module does not have to be replaced. The module is not capacity or performance degraded. This means less maintenance touches will still preserving the life, protection and performance of the system.
Now that our building block, the flash module is created, we can start building a system using something clearly more advanced than a SSD.
The IBM flash module, on its own, is fast and reliable, but we can apply system level technology and improve on this. By having many highly available, high performing modules we can then offer Extreme Performance.
The IBM flash module, on its own, is fast and reliable, but we can apply system level technology and improve on this. By having many highly available, high performing modules we can then offer Extreme Performance.
At this point, we add System-Level RAID protection
VSR protects us from almost every flash cell, block, chip or module issue. System-level RAID protects against abrupt Flash module failure and controller failure. If a flash module has a failure, our system level RAID protects your data… Providing Two Dimensional RAID protection.
The next step is to equip our system with industry leading reliability, availability and serviceability features.
Redundant components – every active component is redundant:
flash modules
interface controllers
management controllers
RAID controllers and XBARs
power, fans, and backup power
A failure in any active component will not bring the system down. This is Enterprise Reliability.
Redundant components are necessary for highly available systems but you expect even more…
The next technology to layer into our system, from the IBM technology tool kit, are designs that promote serviceability and availability.
Starting with Serviceable components – every active component can be removed and replaced while the system is online without removing the chassis lid. Failed modules are easily replaced resulting in higher system availability and faster mean time to repair.
And providing Concurrent Code Load - Promoting the highest possible system availability means avoiding reboots when new firmware/software is loaded into the system, fortunately, IBM has engineered a solution. With concurrent code load, the system remains online during firmware updates.
With this array of features in place, we are ready for the enterprise and ready to deliver the data center of the future!
So what are the most common objections from clients when we start a conversation: They will say that Big Data is not for them. This slide again shows how IBM understands and can help them regardless of the industry or use case they pursue.
The common infrastructure looks like this at a high level and may end up being ‘heavier’ in certain areas. For example, a client deploying a security extension big data project would likely deploy more SW and HW that aid with real time analytics than one who is pursuing a project of data warehouse augmentation.
You’ll notice that we use the icons and names of IBM Infosphere product line. These can be replaced with other components from other vendors or even opensource community but our experience so far is that IBM Infospehere portfolio is highly differentiated and can help with the high level conversation that then brings in the IBM Storage components into it. For example, we have seen several instances where EMC is giving away Greenplum for free to customers and quickly outlining the benefits of Big Insights Enterprise Edition can be a great way to show the customer that IBM can provide value at every step. Big Insights Enterprise Edition include Platform Symphony, which delivers huge performance improvements over Hadoop open source or Cloudera/MapR/Greenplum by using advanced task mangement, workload balancing and other unmatched capabilities.
Link to Youtube Video: http://www.youtube.com/watch?v=D8IQkbUMX9Y
Challenges for RCCL:
-recognized business growth outside of the US…aggressive expansion
-# international guests rose significantly
Customer engagement is important – learning about guests to captivate their business
500M email sent out per year
50M customers are targeted each year
Increase revenue, reduce costs and continue to provide value to their guests
The Solution is business enablement
Big data
Marketing DWH is 15TB is size…oracle DB contains over 10k columns
SVC + FlashSystem ( 10 820 units)
Key Client Benefits
-Business response time to market improved - ability to run multiple marketing campaigns simultaneously
More reliability – NO DOWN TIME
Ability to create snapshots on demand
Reduced data center footprint by 82U
Reduced yearly operational expenses by $77,000
“Flash…the performance capabilities of this technology is what’s reality.”
“The decision was tactical. Performance, maintainability and cost were the highest factors in our consideration of the technology.”
“We were able to move the warehouse to it’s new flash-based home without any impact to the business.”
“The business is not complaining anymore about performance…flash to the rescue!”
“Our journey to flash storage delivered performance improvements and financial savings.”
Other customer quotes:
“Our mission statement is to make, sell and deliver Coca-Cola Company products better than anyone else…. By using IBM FlashSystem to accelerate our insights into customer demand, we’re better placed than ever before to offer unbeatable levels of service to our customers across the United States.”
“Installing the FlashSystem technology itself took just an hour or so—it really is a plug-and-play solution.”
Case study: http://ibm.co/1fFtHkx (external) https://ibm.biz/BdRmdR (internal)
Video: https://ibm.biz/BdRmMv