More Related Content More from IBM India Smarter Computing (20) The IBM zEnterprise EC12: A Technology/Strategy Review1. Research Report
The IBM zEnterprise EC12: A Technology/Strategy Review
Executive Summary
When IBM introduces a new generation of its System z mainframe architecture as it has just done
with its zEnterprise EC12, the first three questions that mainframe customers, prospects, reporters
and research analysts ask are:
1. What has IBM done to improve System z from a technology perspective?
2. What is IBM trying to accomplish strategically? And,
3. How much does it cost?
In short, the answers are:
From a technology perspective, IBM has introduced a new z processor that offers up to
25% more computing power than the previous generation z processor thanks to
improvements in cache, the processor nest, and thanks to new instructions.
o In cache, IBM has expanded Level 2 cache by 33%, and has doubled the cache in
Levels 3 and 4 from 24 MB to 48 MB, and from 192MB to 384MB respectively.
Cache serves data to the processor — so, with faster processors and more cache
IBM is now able to serve more data to the process, thus delivering more compute
power;
o Instruction set improvements and new facilities enable certain workloads to
execute more quickly (for instance, new Java workload performance can increase
by 45%; DB2 by 30%; and PL/1 workload performance will also increase
significantly as PL/1 exploits the new decimal format conversion facility).
Other technology advances include new manageability, availability and security offerings
— as well as advances in hybrid computing.
o The new management improvements include IBM zAware (embedded firmware
that provides advanced analytics-based monitoring for Sysplex environments)
o Enhancements that enable mainframes to manage System x more effectively by
improving mainframe Unified Resource Manager (the mainframe hybrid
computing management environment) interaction with System x Systems Director;
o To improve availability, IBM introduced Flash Express (a new tier of solid state
disk-based memory that absorbs paging challenges as customers transition from
one mode of operations such as batch processing to another mode such as
interactive processing);
o In security IBM has introduced a new version of its cryptography card (the Crypto
Express 4S) with new security extensions including PKCS#11 compliance; and,
o For hybrid environments, IBM has introduced a new zBX enclosure (the zBX
Model 003).
What is IBM trying to do strategically with the new System z? For the past several years
the System z strategy has been focused on: 1) making successive generations of System z
more efficient at processing new and existing workloads optimally in order to ensure the
2. The IBM zEnterprise EC12: A Technology/Strategy Review
August, 2012 © 2012 Clabby Analytics and IT Market Strategy Page 2
longevity of System z; 2) pushing-the-envelope in reliability, availability, and security
(RAS) in order to separate System z from other servers in terms of enterprise-class
resiliency; 3) driving efficiency at scale — making System z the only choice for very large
enterprise computing jobs; and, 4) positioning System z as a central governance node for
management and security within the enterprise.
o With the new zEC12, IBM has placed new emphasis on optimizing Java, SAP and
business analytics workloads;
o The company has pushed-the-RAS-envelope with new availability and security
products (described earlier).
o By adding more processing power and more capacity, IBM has continued to drive
its efficiency at scale initiative; and,
o By introducing a new zBX blade enclosure and by adding management extensions
to its Unified Resource Manager, IBM is proving its continued commitment to
hybrid computing.
From a cost perspective, IBM is reducing $/MIPS (dollars per millions of instruction per
second) consistent with reductions of the past — meaning that the cost for processing
existing workloads on standard, general purpose z workloads is going down. But even
more interesting, IBM is offering a cost reduction of @20% on the $/MIPS on its specialty
engines (IFLs, zIIP, and zAAP processors). Maintenance costs are also decreasing from 2-
20%.
o This shows that IBM is getting very aggressive when it comes to capturing new
workloads on its System z mainframe architecture.
As we look more closely at this announcement, we think the big news is that the zEnterprise EC12 can
execute certain workloads significantly faster than ever before (with major improvements in the processing
of Java workloads, in the throughput of DB2/operational analytics applications, in the speed of processing
compute-intensive C/C++ applications, and with a bigimprovement in the throughput of SAP workloads).
We are also intrigued by IBM’s new IBM zAware reporting facility because it offers the potential to expand
analytics beyond the management of System z environments to other environments that can make use of
pattern analysis and predictive analytics (such as cryptography and fraud detection).
In this Research Report, Clabby Analytics takes a closer look at the new zEnterprise EC12 from
technology, strategic, and pricing perspectives. And our key findings are that the zEC12 is
significantly faster than previous generations; it has been tuned and optimized to execute certain
workloads more efficiently; that System z RAS has been improved — and that IBM has made a
serious effort to offer all of these improvements while lowering the $/MIPS costs for its customers.
Technological Improvements
When we evaluate new generations of System z, we look for hardware improvements (including
processor improvements and the introduction of new hardware elements) as well as optimization
efforts in the areas of firmware, instruction sets and compilers.
A Closer Look at the Hardware Improvements
The first thing we look at when evaluating systems is the microprocessor (we look for speed
improvements as well as new instruction set inclusions). We then look for advances in memory
and in memory management (because the more data that can be placed in memory and cache, the
faster that data can be processed). Next, we look for improvements in the System z internal
throughput rate (ITR – how much work the system can do at 100% busy). And finally, we look at
the input/output (I/O) subsystem to see what IBM has done to improve system I/O bandwidth
3. The IBM zEnterprise EC12: A Technology/Strategy Review
August, 2012 © 2012 Clabby Analytics and IT Market Strategy Page 3
(measured in GB per second — to see how fast IBM can move data between the system and
external devices such as storage subsystems and tightly coupled zBX blade environments).
Our key technology findings are:
At the microprocessor level— the speed of the z microprocessor has been increased from
5.2 GHz to 5.5 GHz, and new instructions have been added to streamline execution for
certain workloads. What this means is that the zEC12 is the fastest superscalar processor in
the industry, enabling it to get more work done than any other processor — while new
instructions and facilities streamline the execution of certain workloads. Additionally, the
number of processors has increased from 80 up to 101 configurable cores.
IBM has added more on-chip cache (described earlier) to speed data serving to these
processors. IBM’s cache improvements have doubled the amount of on-chip data that can
be fed to the processor — leading to significantly faster transaction and data processing.
Memory has remained the same as the predecessor z196 at 3TB of RAIM (redundant array
of independent memory). From our perspective, 3TB is a lot of memory capacity — plenty
for most of today’s most challenging enterprise computing applications — so we don’t
think IBM needed to focus its development efforts on expanding memory in the new
generation zEC12. But, we note that with the new zEC12, IBM has continued to focus on
memory reliability and availability.
o IBM introduced RAIM memory in the last generation System z (the z196) to
improve memory reliability. But now, with Flash Express, IBM has added a new
tier of memory — a solid state disk-based product that can be used to improve the
availability of workloads that cannot tolerate paging spikes or inconsistent
performance during a transition period (for instance, when moving from a batch
processing to interactive processing state).
To further improve performance, we note that the new z microprocessor offers second
generation out-of-order design (a means to keep the microprocessor busy by executing jobs
when they become available, even if they are out of order). Further, IBM has added multi-
level branch prediction (a means to speed processing using a predictive method).
To optimize workload performance, IBM has introduced:
o A transactional execution facility that improves parallelism and scalability;
o A runtime instrumentation facility that helps reduce Java overhead; and,
o Page frames have been increased to 2GB (enabling faster paging for DB2 buffer
pools and Java heaps).
o The performance benefits of these actions are illustrated in Figure 1 (next page).
IBM has increased page frames to 2GB (page frame size is important because larger frames
improve performance in DB2 database buffer pools and speed the execution of Java heaps);
optimized IMS performance (IMS is IBM’s information management system database
management environment), leading to a 30% performance improvement in processing; and
IBM announced that it is planning to release a new PL/1 compiler that should deliver a
large performance boost from the PL/1 decimal format conversions utility.
As for the internal throughput rate, the maximum System z MIPS rating has changed from
52286 to 78426 — meaning that the zEC12 can now do 50% more work than its
predecessor; and,
4. The IBM zEnterprise EC12: A Technology/Strategy Review
August, 2012 © 2012 Clabby Analytics and IT Market Strategy Page 4
From a system I/O bandwidth perspective, I/O bandwidth remains the same at 288 GBps —
still the fastest internal throughput rate in the industry (this is particularly important for
read/writes to attached storage and for internal program-to-program communications).
Figure 1 — Huge Performance Improvements Due to Microprocessor Optimization Efforts
Source: IBM Corporation — August, 2012
Strategic Imperatives
From a strategic perspective, we look for system and software improvements that enable the
System z to run new workloads or to run existing workloads faster; that extend IBM’s already
strong position in RAS; that improve System z efficiency at scale; and that extend IBM’s
zEnterprise hybrid computing environment.
Our key strategic findings are that IBM:
Remains strongly focused on tuning System z for new workloads. At one point in the
history of System z, there was great concern that the future of the mainframe was at risk
because it had construed as a “legacy” server primarily dedicated to running batch and
transactional environments. Over the past dozen years, IBM has been working to change
that image by introducing an open operating environment (Linux) — and by focusing
efforts on tuning the mainframe for new applications (in particular Java-based applications,
enterprise-class business applications, and more recently business analytics applications).
o IBM’s runtime instrumentation facility reduces Java overhead and helps speed Java
program execution.
o In business analytics, IBM has focused on making it possible to run analytics and
on-line transaction processing side-by-side on System z. What this means is that a
single system (System z) can share the same data between operational applications
and analytics applications — without having to move the data set. This means that
data warehouses, analytics, and OLTP can all run as the same workload in a seam-
less, real time environment instead of on siloed, separate databases.
When it comes to RAS, resiliency and manageability, IBM wants System z to be the
industry standard bearer. In the previous section, we described how IBM has improved
availability and performance using a new product called Flash Express (a product that is
designed to support pageable large pages and ensure that workloads perform paging
operations more efficiently). But IBM has also improved its best-in-the-industry, EAL
Level 5+ position with PR/SM and also with the availability of a new cryptographic card
known as Crypto Express4S. This card improves security with new algorithms and with
5. The IBM zEnterprise EC12: A Technology/Strategy Review
August, 2012 © 2012 Clabby Analytics and IT Market Strategy Page 5
the addition of new PKCS #11 coprocessor firmware (it helps meet security requirement of
the European Union and public sector clients). From a manageability perspective, IBM has
improved its System z Sysplex configuration with a “smarter message monitoring” facility
known as IBM zAware (described earlier — it automates identifies potential problems to
help customers speed up problem determination/resolution).
With faster processing and extensive capacity, IBM is positioning the System z as a private
cloud server that offers a high degree of efficiency as well as the ability to scale into very
large cloud configurations. The new zEC12 is consistent with this goal.
Finally, IBM announced its zEnterprise hybrid systems environment when it announced its
z196 environment two years ago. zEnterprise enables a System z to manage and govern
attached “zBX” x86- and POWER-based blade environments, leading to all sorts of
advantages in workload optimization and balancing (workloads can easily be assigned to
the processors/systems best suited to execute them). Along with its announcement of the
zEC12, IBM has also announced a new, updated zBX blade chassis environment, signaling
that the company plans to continue to aggressively pursue hybrid workload opportunities
by improving the management of hybrid zEnterprise/zBX mainframe/blade environment.
Cost/Pricing
When looking at System z costs we look at the cost for MIPS (millions of instructions per second
— a measure of processing power), software pricing, and maintenance costs.
The big news from a cost perspective is that IBM is lowering its MIPS costs for traditional established
applications — and more aggressively for new application environments. What this shows us is that IBM
wants to reward its existing customers with lower costs, while incenting customers and prospects to place
new workloads on System z.
Our key cost/pricing findings are that:
Existing customers running traditional workloads should expect to see traditional hardware
price points decline in a range consistent with the past several years;
The introduction of IBM’s Technology Upgrade Pricing will deliver a 2-7% lower MLC
(monthly license charge) software pricing for equivalent capacity;
Offers a cost reduction of @20% on the $/MIPS on its specialty engines (IFLs, zIIP, and
zAAP processors). (IBM is reducing costs by preserving the 120 PVU [processor value
unit] rating to help reduce IFL software costs). This reduction shows that IBM is getting
even more aggressive when it comes to capturing new workloads on zEnterprise;
Offers a minimum 2% price/performance decrease in maintenance for traditional workload
environments running standard MIPS with a greater discount offered with growth upgrades
— and a 20% maintenance reduction for customers running IFL (Integrated Facility for
Linux — a Linux specialty processor) MIPS;
Is continuing to try to attract new customers with cost-reduced, packaged mainframe/-
application software solutions (known as “Solution Editions”) that offer aggressive pricing
and terms for first-in-enterprise customers; and,
Is offering a no-fee carry-forward of zManager blade entitlements and zBX upgrades to its
zEnterprise customers.
6. The IBM zEnterprise EC12: A Technology/Strategy Review
August, 2012 © 2012 Clabby Analytics and IT Market Strategy Page 6
What Impresses Us Most: IBM’s Workload Optimization, Management and Business Analytics
Improvements
To us, it is perfectly clear that IBM is positioning the System z to go after new workloads —
particularly Java-based as well as analytics workloads. As proof, notice in Figure 1 the huge
improvement in processing Java workloads (a whopping 45% — this is a huge improvement!).
Also note the 30% performance improvement for DB2 and z/OS operational analytics.
From our perspective IBM is making a statement: The System z is more than a batch/transaction processor
— it is a highly scalable, highly resilient enterprise server capable of executing the most modern Java and
database/analytics applications available in today’s marketplace.
Workload Optimization: A System-wide Effort
Note, however, that these major improvements in workload optimization are not soley the result of
improving processor speed, expanding cache, and building in new instructions and facilities. They
are the result of a total systems effort that includes advances in semiconductor manufacture,
microprocessor design, systems design, virtualization and operating systems tuning, compiler and
Java virtual machine tuning, and middleware optimization (see Figure 2).
Figure 2 — From Semiconductor Design to Middleware Optimization: A Total Systems Effort
Source: IBM Corporation — August, 2012
Manageability Advances
One area that we constantly track is System z manageability. We are primarily interested in
mainframe management from two perspectives:
We want to ensure that the mainframe management environment is constantly made
simpler to manage such that a new generation of mainframe managers can easily step-in
and manage/tune the mainframes of the future. For more on this concept, please see our
report entitled “OMEGAMON V 5.1 Review: Client-driven Redesign of Mainframe
7. The IBM zEnterprise EC12: A Technology/Strategy Review
August, 2012 © 2012 Clabby Analytics and IT Market Strategy Page 7
Performance and Availability Monitoring” that describes how IBM clients are guiding the
design of mainframe management to simplify mainframe management; and,
We believe that one of the biggest advantages that mainframe architecture has over other
systems architectures is in the area of manageability. We have noted in other reports that it
takes far fewer systems administrators and managers to manage highly-virtualized
mainframe environments than it does to manage distributed systems environments. So we
constantly track advances in mainframe management in order to show our readers how
superior mainframe management is when compared with distributed systems environments.
IBM’s latest manageability announcements included improvements in IBM’s Unified Resource
Manager (a management product used to support workload aware resource optimization in hybrid
zEnterprise/zBX environments. Further IBM announced enhancements to its System Director x86
management environment that make it easier for System Director to support IBM’s zBX enclosures.
But, in addition to these management improvements, IBM also announced a new monitoring
product known as IBM zAware. As described earlier, IBM zAware is embedded firmware that
provides advanced analytics-based monitoring for mainframe environments. IBM zAware has a set
of restful APIs that provide information, in XML format, that can feed other processes or tools. It
takes samples every 2 minutes at 10 minute intervals — and it uses a 90 day rolling baseline that it
establishes by monitoring system behavior. In this manner, IBM zAware is able to isolate potential
problems — and enterprises are able to spot problems before they mushroom.
At the zEC12 announcement, IBM provided a statement of direction on IBM zAware: IBM plans
to provide new capability within the Tivoli Integrated Service Management family of products
designed to leverage analytics information from IBM zAware, and to provide alert and event
notification. (Note that because IBM zAware delivers information in XML format, so many other
management tools can also take advantage of IBM zAware information).
What we like about IBM zAware is that it simplifies management and streamlines troubleshooting —
making it easier for managers and administrators to tune and fix their System z environment without
requiring any rules to be written. We think this product, as it matures and becomes more broadly available
across the System z line, will go a long way to simplifying the mainframe for next generation managers —
as well as lower management costs.
Business Analytics Advances
In the past we have written about how IBM’s System z makes an ideal business analytics
processing environment (see our report entitled: “Choosing IBM’s z Enterprise for Next Gen
Business Analytics Applications” for more details on why System z offers “superior economics”
when it comes to business analytics processing).
What we are now seeing with the new zEnterprise EC12 is a system that offers more capacity, more
processing power, and a systems design that is superior to distributed systems when it comes to
analytics processing. A closer look at the System z network/bus/IO subsystem design shows why
mainframes communicate more efficiently than distributed systems designs. Distributed systems
rely heavily on networks to pass information between servers and to fetch data from databases. And
networks can easily become congested — especially during peak workload time. System z, on the
other hand, has a huge, high-bandwidth internal bus and a supporting subsystem of specialized
input/output processors that enable applications to work efficiently and cooperatively without the
network congestion that plagues distributed computing designs. Further, the System z can be
tightly-coupled to external storage that supports backend data stores, delivering data at high speeds
to the processors performing the work.
8. The IBM zEnterprise EC12: A Technology/Strategy Review
Clabby Analytics
http://www.clabbyanalytics.com
Telephone: 001 (207) 846-6662
© 2012 Clabby Analytics
All rights reserved
August, 2012
Clabby Analytics is an independent technology research and
analysis organization. Unlike many other research firms, we
advocate certain positions — and encourage our readers to find
counter opinions —then balance both points-of-view in order to
decide on a course of action. Other research and analysis
conducted by Clabby Analytics can be found at:
www.ClabbyAnalytics.com.
This final point regarding back-end data stores is very important. With its increased capacity and
processing power, the zEC12 is able to run data warehouse, analytics and OLTP applications using the
same data store (instead of relying on multiple, snap-shotted, siloed individual databases). The ability to
use the same data store simplifies data management (no need to worry about data duplication, for
instance) — and it paves the way for enterprises to develop new insights. (IBM customers are now telling
IBM that they are able to develop new insights that they could not have developed when using separate
data silos). The ability to use a common data set across data warehouses, analytics, OLTP, and other
operational environments is a distinct advantage for the highly-scalable System z.
Summary Observations
The new zEnterprise EC12 is a fast, powerful, highly-integrated, well-balance computing
environment designed to serve a wide variety of workloads (transactional, batch , interactive,
scientific, industry, and analytics applications).
What impressed us most from a technology perspective about this new System z is its raw
processing power (5.5Ghz) aided by significantly more on-chip cache (to rapidly feed data to the
superscalar processor).
We are also impressed with all of the work that IBM has done to accelerate the processing of
various workloads on System z. The zEC12 has new instructions and facilities that greatly improve
Java, PL/1 and C++ applications handling. Capturing new applications on System z is strategically
vital for IBM — and given the software performance streamlining that IBM has done on the
zEC12, we expect more and more independent software vendors (ISVs) to port their software into
the mainframe environment (and we expect that IBM customers with custom applications will build
more applications on z).
We also liked all of the work that IBM has done in manageability. Simplifying and improving
manageability helps pave the way for next generation System z managers and administrators to
more easily troubleshoot and tune System z environments. Further, by simplifying management by
using advanced analytics, we believe that IBM has poured the foundation for a slew of new
management and security products that will serve to greatly lower human-labor-related
management costs.
To us, this new zEnterprise (the zEC12) showcases all of the major System z differentiators: powerful
processing, advanced RAS, simplified manageability, efficiency at scale — and most importantly: the
ability to optimally execute a wide range of workloads across System z and hybrid environments.
ZSL03191-USEN-01