Hot Sexy call girls in Patel Nagar🔝 9953056974 🔝 escort Service
z13: New Opportunities – if you dare!
1. FOR MORE INFORMATION PLEASE CONTACT
Xact Consulting A/S
Arnold Nielsens Boulevard 68A
DK-2650 Hvidovre
+45 7023 0100
info@xact.dk
www.xact.dk
Enterprise Modernization
z13: New Opportunities –
if you dare!
Michael Erichsen,
April 2015
2. Lix Level Warning
• This is not a buzzword
compliant sales pitch
• This is low level stuff,
aimed at hard core
programmers
3. Agenda
• New hardware
• New machine
instructions
• Business use
• Development
improvements
• New application types?
4. z12 broadens the Scope of Mainframes
• Processor chip optimized for software performance
– Exploited by Java, compilers, and DB2
– Transactional Execution Facility for parallelism and scale
– Runtime Instrumentation Facility to reduce Java overhead
• zEC12 has the industry’s fastest chip with each core at 5.5 GHz
– z13 processor actually a little bit slower
5. z12 Transactional Execution
• Sequence of instructions
executed as “atomic”
operation
• Eliminates locks
• Improved software
parallel scaling
6. z12 Runtime Instrumentation
• Real-time information
to software on
dynamic program
characteristics
• Increased optimization
in JVM/JIT
recompilations
7. z196/z12 Blade Extension
• Unified Resource
Manager
• Inside the same
security perimeter
• Windows added in z13
8. z13: Big Data, Analytics, Cloud
• Decimal-floating-point packed-conversion facility
• Message-security-assist extension
• Multithreading facility
• Vector facility for z/Architecture (Single Instruction / Multiple
Data (SIMD) technology)
9. Don't forget the Machine Instructions added in earlier Models
• UTF conversion instructions
• Cryptographic instructions
• Gzip acceleration in hardware
10. New Instructions – why?
• Competition from the fast moving distributed world and cheap
commodity hardware
– Java, dot.net, XML, Unicode, BI, Big Data, Analytics, Cloud…
• Basis for compiler improvements: COBOL, PL/I, C, C++, Java,
PHP…
– E.g. decimal floating point format conversions for COBOL
• Also the basis for hybrid applications
– Run traditional and new workloads on the same system
11. Why Java on the Host?
• New developers
• Application portability
• Requires improved interfaces to COBOL,PL/I, C/C++, HLASM
– Language Environment, OO COBOL, DWARF debugging, service
interfaces
• Requires performance improvements and economical tricks
like zIIP and zAAP processors
12. Why XML?
• Standard protocol
• Very extensible
• Broad use
• But terrible performance on all platforms
– Requires processor improvements and zIIP/zAAP processors
• Programming languages use the new hardware and
instructions through XML System Services
13. Why UNICODE?
• All languages, dead or alive
• But is a character 1, 2, 3 or 4
bytes?
– UTF-8, 16, 32 – and 1, 5, 6, 7,
9, and 18…
• COMPARE LOGICAL LONG
UNICODE
• CONVERT UNICODE TO UTF-8
• CONVERT UTF-16 TO UTF-32
• CONVERT UTF-16 TO UTF-8
• MOVE LONG UNICODE
• SEARCH STRING UNICODE
17. New COBOL Instructions
• ULENGTH
• USUBSTR
• UNSUPPLEMENTARY
• VALID
• UWIDTH
• Also supported
natively in DB2 and
MQ
18. Big Data
• Unstructured data
Processable data
• A new paradigm
–Batch
–OLTP
–Grid and Big Data
19. Data Types
• Structured
– Relational, hierarchical, XML
• Unstructured
– Documents, web sites, e-mail, social media
• Streaming
– Sensors, RFID's, Internet of Things, Smart Phone apps, log data
20. Application types
• Search
• Analytics
• Patterns, trends, anomalies
• Traffic control, water and electricity supply, personal health
monitoring, earth quake warnings…
• Sentiment analysis
21. Infrastructure types
– Low latency networking
• Fire-and-forget
– NoSQL (Not only SQL)
• Dynamic, semi-structured data with low latency
– Hadoop
• Processing, storing and analysing data
– UIMA (Unstructured Information Management Architecture)
22. The z Implementation
• Hybrid boxes
• Data accelerators
• Hybrid applications
• Instructions
– Vector processing
– SIMD (Single Instruction, Multiple Data)
23.
24. Vector Facility
• The vector facility provided in the z/Architecture architectural
mode provides fixed-sized vectors ranging from one to sixteen
elements
– For most instructions, all of the data contained in a vector is
operated on by the instructions defined in this facility
– Some instructions only operate on a subset of the elements within a
vector
– If a vector is made up of multiple elements, each element is
processed in parallel with the others
– Instruction completion does not occur until processing of all
elements is complete
25. COBOL Compiler Implementations
• V5.1:
– UTF-8 support improved
– ARCH(10): z12 adds execution-hint facility, load-and-trap facility,
miscellaneous-instructions-extension facility, transactional-execution
facility speeds up each instruction
• V5.2:
– ARCH(11): Exploitation of the z13 instructions helps e.g. INSPECT
TALLYING, INSPECT REPLACING and COMP-3
29. How to keep Costs down doing this?
• System z New Application Licence Charges (zNALC)
– New workload moved from other platforms in separate LPAR
– Value Unit Edition (VUE)
• One time charge for eligible workloads in such LPAR
• Mobile Workload Pricing for z/OS (MWP)
– Extra load must be documented
• IFL, zIIP, zAAP, and Blade Extension
Editor's Notes
DataPower
IBM Data Accelerator Appliance
AIX
Linux
Windows
Braille
Klingon
Tengwar
The ULENGTH function returns an integer equal to the length, in UTF-8 characters, of a character string argument that is encoded in UTF-8.
The USUBSTR function returns a substring of a character string argument that is encoded in UTF-8.
The UNSUPPLEMENTARY function returns an integer equal to the index of the first Unicode supplementary character in a character string argument that is encoded in UTF-8 or UTF-16.
The UVALID function returns an integer which has the value zero if a character string contains valid Unicode UTF-8 or UTF-16 data, and which has the index of the first invalid element if the character string does not contain valid Unicode data.
The UWIDTH function returns an integer equal to the width in bytes of the nth UTF-8 character in a character string argument that is encoded in UTF-8.
Unstructured data Processable data
Volume
Variety: Different forms
Velocity: Streaming data
Veracity: Uncertainty of data
A new paradigm
Batch (Punched cards)
OLTP (Sabre, Saturn 5, customer Information control systems)
Grid (SETI) and Big Data (Internet search engines, Hadoop)
The grid can be thought of as a distributed system with non-interactive workloads that involve a large number of files.
Same program – sliced up data
Hadoop: Distributed processing of large data sets across clusters of commodity servers.
Same data – different programs – voting
The term MapReduce actually refers to two separate and distinct tasks that Hadoop programs perform.
The first is the map job, which takes a set of data and converts it into another set of data, where individual elements are broken down into tuples (key/value pairs).
The reduce job takes the output from a map as input and combines those data tuples into a smaller set of tuples.
As the sequence of the name MapReduce implies, the reduce job is always performed after the map job.
UIMA is a component software architecture for the development, discovery, composition, and deployment of multi-modal analytics for the analysis of unstructured information and its integration with search technologies
Single instruction, multiple data (SIMD), describes computers with multiple processing elements that perform the same operation on multiple data points simultaneously.
Thus, such machines exploit data level parallelism, but not concurrency: there are simultaneous (parallel) computations, but only a single process (instruction) at a given moment.
SIMD is particularly applicable to common tasks like adjusting the contrast in a digital image or adjusting the volume of digital audio
An application that may take advantage of SIMD is one where the same value is being added to (or subtracted from) a large number of data points, a common operation in many multimedia applications. One example would be changing the brightness of an image.
Traditionally you would run data intensive workloads on the mainframe because of the optimized I/O system
And numeric intensive on distributed servers because of the processors optimized for gaming and video streaming
A hybrid workload would be split between platforms
Now the mainframe processors have been optimized for numeric intensive workloads too.
Imagine a warehouse system that keeps track of your inventory
Combined with a sentiment analysis system that predicts your future sales
Running as a single hybrid application on a mainframe
This is an example of the technical setup of a hybrid applications