26 September 2017

Oracle Sparc M8 and Oracle Advanced Analytics

Oracle SPARC M8 released with 32 cores 256 threads 5.0GHz

Oracle announced its eighth generation SPARC platform, delivering new levels of security capabilities, performance, and availability for critical customer workloads. Powered by the new SPARC M8 microprocessor, new Oracle systems and IaaS deliver a modern enterprise platform, including proven Software in Silicon with new v2 advancements, enabling customers to cost-effectively deploy their most critical business applications and scale-out application environments with extreme performance both on-premises and in Oracle Cloud.

Oracle’s Advanced Analytics & Machine Learning 12.2c New Features & Road Map; Bigger, Better, Faster, More!

SPARC M8 processor-based systems, including the Oracle SuperCluster M8 engineered systems and SPARC T8 and M8 servers, are designed to seamlessly integrate with existing infrastructures and include fully integrated virtualization and management for private cloud. All existing commercial and custom applications will run on SPARC M8 systems unchanged with new levels of performance, security capabilities, and availability. The SPARC M8 processor with Software in Silicon v2 extends the industry's first Silicon Secured Memory, which provides always-on hardware-based memory protection for advanced intrusion protection and end-to-end encryption and Data Analytics Accelerators (DAX) with open API's for breakthrough performance and efficiency running Database analytics and Java streams processing. Oracle Cloud SPARC Dedicated Compute service will also be updated with the SPARC M8 processor.

Spark SQL: Another 16x Faster After Tungsten: Spark Summit East talk by Brad Carlile

"Oracle has long been a pioneer in engineering software and hardware together to secure high-performance infrastructure for any workload of any size," said Edward Screven, chief corporate architect, Oracle. "SPARC was already the fastest, most secure processor in the world for running Oracle Database and Java. SPARC M8 extends that lead even further."

The SPARC M8 processor offers security enhancements delivering 2x faster encryption and 2x faster hashing than x86 and 2x faster than SPARC M7 microprocessors. The SPARC M8 processor's unique design also provides always-on security by default and built-in protection of in-memory data structures from hacks and programming errors.

SPARC M8's silicon innovation provides new levels of performance and efficiency across all workloads, including: 
  • Database: Engineered to run Oracle Database faster than any other microprocessor, SPARC M8 delivers 2x faster OLTP performance per core than x86 and 1.4x faster than M7 microprocessors, as well as up to 7x faster database analytics than x86.
  • Java: SPARC M8 delivers 2x better Java performance than x86 and 1.3x better than M7 microprocessors.  DAX v2 produces 8x more efficient Java streams processing, improving overall application performance.
  • In Memory Analytics: Innovative new processor delivers 7x Queries per Minute (QPM)/core than x86 for database analytics.
Oracle is committed to delivering the latest in SPARC and Solaris technologies and servers to its global customers. Oracle's long history of binary compatibility across processor generations continues with M8, providing an upgrade path for customers when they are ready. Oracle has also publicly committed to supporting Solaris until at least 2034.

Oracle Sparc M8 is available for:

  • Oracle SPARC M8
  • Oracle SPARC T8-1 server
  • Oracle SPARC T8-2 server
  • Oracle SPARC T8-4 server
  • Oracle SPARC M8-8 server
  • Oracle SuperCluster M8 engineered system

More information in: Oracle SPARC M8 Launch Webcast:  http://www.oracle.com/us/corporate/events/next-gen-secure-infrastructure-platform/index.html

About Oracle 

The Oracle Cloud offers complete SaaS application suites for ERP, HCM and CX, plus best-in-class database Platform as a Service (PaaS) and Infrastructure as a Service (IaaS) from data centers throughout the Americas, Europe and Asia. For more information about Oracle (NYSE: ORCL), please visit us at oracle.com.

Big data analytics using oracle advanced analytics and big data sql

The Oracle SPARC M8 is now out and is a monster of a chip. Each SPARC M8 processor supports up to 32 cores and 64MB L3 cache. Each core can handle 8 threads for up to 256 threads. Compare this to the AMD EPYC 7601, the world’s only 32 core x86 processor as of this writing, which handles 64 threads and also has 64MB L3 cache. The cores can also clock up to 5.0GHz faster than current x86 high-core count server chip designs from Intel and AMD. That is quite astounding given the SPARC M8 is still using 20nm process technology.

Beyond simple the simple core specs, there is much more going on. Oracle has specific accelerators for cryptography, JAVA performance, database performance and ETC. For example, there are 32 on-chip Data Analytics Accelerator (DAX) engines. DAX engines offload query processing and perform real-time data decompression. Oracle’s software business for the Oracle Database line is still strong and these capabilities are what is often referred to as “SQL in Silicon.” Oracle claims that Oracle Database 12c is up to 7 times faster by using M8 with DAX than competing CPUs. That is a big deal for software licensing costs. Another interesting feature is the inline decompression feature allows decompression of data stored in memory with no claimed performance penalty.

Oracle SPARC M8 Processor Key Specifications

Here are the key specs for the new Oracle SPARC CPUs:

  • 32 SPARC V9 cores, maximum frequency: 5.0 GHz
  • Up to 256 hardware threads per processor; each core supports up to 8 threads
  • Total of 64 MB L3 cache per processor, 16-way set-associative and inclusive of all inner caches
  • 128 KB L2 data cache per core; 256 KB L2 instruction cache shared among four cores
  • 32 KB L1 instruction cache and 16 KB L1 data cache per core
  • Quad-issue, out-of-order integer execution pipelines, one floating-point unit, and integrated cryptographic stream processing per core
  • Sophisticated branch predictor and hardware data prefetcher
  • 32 second-generation DAX engines; 8 DAX units per processor with four pipelines per DAX unit
  • Encryption instruction accelerators in each core with direct support for 16 industry-standard cryptographic algorithms plus random-number generation: AES, Camellia, CRC32c, DES, 3DES, DH, DSA, ECC, MD5, RSA, SHA-1, SHA-3, SHA-224, SHA-256, SHA-384, and SHA-512
  • 20 nm process technology
  • Open Oracle Solaris APIs available for software developers to leverage the Silicon Secured Memory and DAX technologies in the SPARC M8 processor
  • On Solaris Support Until 2034

In the official Oracle SPARC M8 release, Oracle has a note that is a clear nod to its Organizationals changes (we mentioned in a recent Oracle server release.)

Oracle is committed to delivering the latest in SPARC and Solaris technologies and servers to its global customers. Oracle’s long history of binary compatibility across processor generations continues with M8, providing an upgrade path for customers when they are ready. Oracle has also publicly committed to supporting Solaris until at least 2034.

Oracle is clearly hearing from its customers about the mass layoffs of Solaris engineering teams.

New Oracle SPARC M8 Systems

There are five new SPARC V9 systems are available from Oracle today:

  • Oracle SPARC T8-1 server
  • Oracle SPARC T8-2 server
  • Oracle SPARC T8-4 server
  • Oracle SPARC M8-8 server
  • Oracle SuperCluster M8 engineered system

The Evolution and Future of Analytics

We live in a world where things around us are ever changing.

Measurement metrics are just in time, predictive and need a lot of augmented intelligence; however, we're developing more complex mind analytics when it comes to buying patterns.

This new type of analytics can give us insight into how the customer feels and what he or she experiences.

Oracle's Machine Learning & Advanced Analytics 12.2 & Oracle Data Miner 4.2 New Features

Thus, the availability of smart information will emerge.

In the future, you may walk into a store and find one or all of the below, which can be built as solutions:

a) A robot welcoming you and taking over to interact with you using connected back end and analytics.

b) Natural language or human analytics that can automatically read your mood to ultimately improve customer satisfaction.

c) Historical data about you as a customer to help up sell or cross sell products based on your interests.

d) Automatic analysis about what you're doing to bring near real-time context of data; this will enable the retailer to build a mobile based intuitive presence or no billing architecture.

e) A personal assistant model to better serve you as a customer, empowering retailers to provide solutions to unsure customers.

f) IN product or things analytics to provide information about the product that makes things intelligent through RFID, intelligent tagging, sensors etc.

g) Discounts/coupons based on mixing historical buying patterns; post purchase analytics.

h) Interactive dashboards that make augmented decisions about a few areas based on reviews; this would take expert reviews, phone calls, product management and more into account.

i) A store platform of grammar, syntax, semantics and data science grammar to create recurring patterns, challenges and build new solutions which are continuous in nature.

Based on the above, let's dive into different types of analytics available on the market. We'll look at how they will blend and intersect to develop more augmented applications for the future.

Insights into Real-world Data Management Challenges

1) Historical Analytics

This is the traditional analytics of business intelligence focused on analyzing stored data and reporting. We would build repositories and create analyses and dashboards for historical data. Solutions would include Oracle Business Intelligence.

2) Current Analytics 

Here the analytics is measurement over current process. For example, we would measure the effectiveness of a process as it happens (business activity monitoring) using a stream that processes arriving data and analyzes it in real-time.

3) Enterprise Performance Management

Here the objective is to focus on projections/what-if analysis with the current data and make projections for the future. An example would be a Hyperion or an EPM based solution which could help derive and plan reporting as projections. EPM today is also available as a cloud service.

4) Predictive Analytics

With the Big Data market growing, and with unstructured data adding parameters of velocity, variety and volume, the data world is moving on to more predictive analytics, with a blended mix of data. There is one world of data in the hadoop world and another in the classical data warehouse world. We can mix and match and do Big Data analytics.

Predictive analytics is more of a compass-like decision making with data analysis patterns. Oracle has an end-to-end Big Data solution from DW, Hadoop and analytics that can help develop predictive solutions.

MSE 2017 - Customer Voted Session: Rocketing Your Knowledge Management Success with Analytics

5) Prescriptive Analytics

To extend the predictive analytics, we would also develop systems to make decisions once we have the prediction; i.e. sending emails and connecting systems as the patterns are detected. This is the basics of building more heuristics systems to make decisions about arrived patterns.

6) Machine Analytics 

Every device and machine is going to generate data. Machine analytics is a blended form of data that can be embedded into the standard source to enhance and improve the overall data pattern. Oracle provides IOT CS as a solution to connect, analyze and integrate data from various machines and enrich new applications like  ERP, CRM and more.

Oracle Analytics and Big Data: Unleash the Value

7) AI Based Analytics

AI or deep learning is the next gen analytics pattern where we can train the systems or any entity to think and then embed the analytics pattern in the solution.

8) IORT / Robotics Analytics

With Robots / Bots and personal assistant complementing solutions, there are a lot of patterns of thinking and execution distributed to multiple systems. IORT or robotics analytics is a new branch that will focus on how we can analyze the pattern from semi thinking devices.

9) Data Science as Service 

A new branch where the analysis goes deeper in terms of algorithms and storage and is also more domain-driven. Even though data science is used as one branch in analytics, you will see a lot of analytics development. Data scientists who specialize in identifying patterns will go a long way to build patterns that are more replicable.

10) Integrated Analytics

In the future, we can form an integrated view of the above. This could be ONE IDE and you would derive patterns based on business need and use case. Today, we have a fragmented set of tools to manage analytics and it would slowly get integrated into one view.
Oracle has solution at different levels; most of them are also available as a cloud service (Software as a Service, Platform as a Service).

MSE 2017 - Advanced Analytics for Developers

It's imperative to build the right mix of solutions for the right problem and integrate these solutions.

  • Historical perspective you would use --> Business Intelligence 
  • Current processing  -->  Streaming (event processing) and Business Activity Monitoring
  • Enterprise performance management  --> Hyperion
  • Heterogeneous source of data and also large analysis of data --> Big Data Solution
  • Predictive and Prescriptive analytics --> R language and Advanced Analytics
  • Machine related --> IOT Solutions and Cloud Service

Oracle Architectural Elements Enabling R for Big Data

Oracle University provides competency solutions for all the above and empowers you with skill development and well-respected certifications that validate your expertise:

  • Big Data Analytics training
  • BI Data Analytics training
  • Hyperion training
  • Cloud PAAS Platform for Analytics and BI training

More Information:













Oracle Visual Analytics






0 reacties:

Post a Comment