Tuesday, October 6, 2015

IBM Power Systems LC: buy on-line with faster performance, lower cost than x86 systems!

By Rich Ptak and Bill Moran 

Moore’s law is dead. Don’t remember the law[1]? You should still care because it underlies computer industry progress the last 50 years. Briefly, it states that doubling the number of components on a transistor every two years (or so), doubles performance while shrinking physical size and cost. Unfortunately, thermodynamics and physics mean this approach no longer works.

IBM is replacing the law with one that sees the future fueled by innovation across the total stack of system components, augmented by open systems collaboration. To that end, IBM has announced new IBM Linux servers, the Power Systems LC[2] lineup. It has three new systems designed for data and cognitive workloads, as well as an entry level systems at the lowest price ($6595!) yet available for Power Systems. With technology from OpenPOWER Foundation members, they run workloads faster and cheaper than x86-based systems.

The three new POWER8 Linux severs are the S812LC (entry), S822LC (Commercial computing) and the S822LC (High Performance Computing) designed specifically for clouds and clusters. They deliver performance and price advantages over x86 systems. The S812LC completes a Spark workload with about 2.3 times better performance/dollar-spent as a Xeon E5-2690 v3 System.

In the post-Moore’s law world, improvements will require the collaboration and innovation from multiple companies and institutions. IBM created the OpenPOWER Foundation[3] to encourage such efforts on Power System technology. With open access to the base Power architecture, members can innovate with their own technology and integrate improvements into Power processors.

Here’s how innovation and collaboration with OpenPOWER Foundation members delivers performance improvements. CAPI[4], a standard feature of POWER8, allows direct access to high volumes of data using Flash Memory in a NoSQL environment. Redis Labs[5] managed to reduce the number of POWER8 servers (compared with X86) needed in a 40TB NoSQL case by a factor of 13[6].

In just two years of existence, the Foundation has attracted over 150 member companies and over 35 new products. IBM, Mellanox and NVIDIA collaborated to win a $325 million super computer contract from the US Department of Energy. There are many more documented successes available.

We recommend anyone considering a server purchase evaluate the new Power Systems LC line. For many data intense applications that are run in cluster environments, you may find that these systems deliver value not available elsewhere. For those looking to test their applications before doing on-site proof of concept, a POWER8 developer cloud is available before ordering a system.

IBM recently announced LinuxONE[7] for the mainframe world; now there is the Power Systems LC to make things even more interesting. For more on new capabilities and products view the webcast at:

[1]'s_law describes the law and its history.
[2] Special configurations/pricing available for easy on-line purchasing:
[4] Coherent Accelerator Processor Interface, See CAPI and NoSQL:   
[6] In fact, 80 X86 servers were replaced with 6 Power8 servers.

Tuesday, September 29, 2015

Big Data/Analytics Performance – Driving IBM Power System Success

By Rich Ptak

It’s no secret that the action today is in Big Data and the associated Analytics! Whether for business, retail, education, government, medical, media, whatever, the focus is on data! Lots of it! Coming from every direction in every conceivable source and format. It is structured, unstructured, transactional, audio and visual flowing from IoT, mobile, social, production…to the tune of some 2.5 quintillion new bytes generated every single day.

IT is tasked with processing this raw data into the insights, wisdom, knowledge that result in new services or deliver solutions to previously impenetrable mysteries. The ultimate goal is to deliver benefits and provide value to users, clients and customers. Processing large amounts of data has been computing’s forte since their inception. BUT now the processing of data and generating results is immensely more complicated and must be delivered more quickly and economically than ever before.

IBM’s Power8 was specifically designed as a Big Data server with industry leading memory bandwidth, thread density, and cache architecture. It has the analytics tools[1], operating systems[2], databases[3] to be the System of Insight equipped to deal with the software, performance and management challenges of Big Data analysis, integration and governance.

And, in discussions with users, we’ve seen that it delivers. See our blogs about customer[4] success at dealing with Big Data challenges using Power8 systems.  Whether the goal is near real-time response (1.5 microsecond Algo-Logic’s Tick-to-Trade); significant cost savings with improved performance (IBM Platinum Partner Redis Labs processes more REDIS-NoSQL transactions with faster response times with fewer CAPI-Power8 servers); or  TalkTalk[5], a UK communications service provider, updating their network and improving the service to their customers by switching to Power-CAPI powered servers. 

No industry-standard benchmark existed for Apache Spark[6] until IBM developed the SparkBench benchmark suite. The first version includes 10 benchmarks covering four use cases: Machine Learning, Graph, SQL and Streaming Spark. The results are that a wide variety of Spark workloads consistently run 2x faster on POWER8 than competitor platforms. (FACT: POWER8 with 24 processor cores runs 37% faster than Haswell with 36 processor cores.) You can get SparkBench details and results here[7]. And, if you want to make sure that the SparkBench is the REAL thing, it is available to the public here[8]. IBM recently announced LinuxONE[9] for the mainframe world, we expect more interesting information in the October 5th webcast on new capabilities and products. We’ve registered and suggest that you do so also at:

[1] Hadoop, Big Insights, DB2BLU and Spark
[2] Red Hat, SuSE and Ubuntu
[3] Oracle, DB2LUW, MariaDB, MongoDB, PostgreSQL
[6] An in-memory distributed compute engine to complete analysis on large-scale data sets up to 100X faster than current technologies. More info on Apache Spark here:


Wednesday, September 23, 2015

IBM’s LinuxONE – Penguins in the Enterprise

By Bill Moran and Rich Ptak

Ever since IBM made a huge ($1 billion) commitment to Linux, its growth on the mainframe has been spectacular. It has a 10-year compound annual growth (CGI) of 45%, as 27% of all shipped mainframe capacity is Linux specific. Linux on the mainframe has penetrated the majority of IBM’s large clients.

Now, Linux is the fastest growing OS in the broader market, and IBM plans to capture a portion of that growth for the mainframe. LinuxONE is their strategy to do so.

What’s the connection between IBM’s new LinuxONE offering and the search for a cure for the deadly disease, ALS[1]? You might say “Nothing” or “Not much,” but that’s wrong. Read on to discover how IBM’s LinuxONE platform plays an important role in the search. But, before discussing IBM’s contribution, let’s review the LinuxONE offering
IBM believes combining the best of Linux with the best of the enterprise computing can be a real winner. They expect the new LinuxONE offerings will appeal to customers already familiar with Linux but who know little about the mainframe. This reasoning appears correct to us with the potential of a very successful system.

We won’t cover all of the details about IBM’s LinuxONE. These are available in the references provided below[2]. Instead, we focus on what we consider the most significant and important features. We also want to explore some humorous side effects of the offering and some problems that IBM may have along the way.

What’s new in the strategy? What difference does it make? First. We’ll look at the new items; then we’ll cover what it all means. The new items by categories are:
1.    Expanded relations with Linux Community
2.    Brand New Cloud offerings
3.    Hardware Systems
4.    Software
5.    Pricing models
These mark a major change in mainframe direction as well as an expansion outside the traditional customer base.

See the rest of our comments at:

Tuesday, September 22, 2015

BMC Mainframe MLC Cost Management – Enabling Digital Enterprise Management

By Rich Ptak and Bill Moran

It is not often that a company understates their product’s features, while actually delivering significantly more. Until recently, we found that while BMC products expertly leverage underlying technology, all too frequently, their messaging failed to highlight all inherent significant benefits and features. Happy to say, those days appear over as BMC announces new releases for its Digital Enterprise Management[1] (DEM) portfolio. To optimize mainframes for DEM, BMC is developing new features and improving products to manage data, infrastructure and costs. BMC teams are comprehensively cataloging significant features “designed to make digital business fast, seamless and optimized from mainframe to cloud and beyond.”

Our focus in this paper are monthly licensing costs (MLC) that alone can account for 30%[2] of total mainframe costs. Let’s look at BMC Mainframe MLC Cost Management V2.0.

First, existing customers with paid versions of release 1 will get the release free for those capabilities they have licensed. Customers who have licensed the v1 components (Cost Analyzer, Intelligent Capping, Subsystem Optimizer for DB2, Subsystem Optimizer for IMS) will get the corresponding new V2 features free. After, reviewing the details of the new release, we think you will agree this represents a real bargain.

Those unfamiliar with BMC’s Mainframe Cost Management products or considering evaluating them, should first uncover the true costs of the mainframe. We say this because many companies, wittingly or not, tend to inflate mainframe costs. They do so by including an assortment of general IT costs as mainframe costs. It just doesn’t make sense to assign distributed system air conditioning and staffing to the mainframe.

With real mainframe costs in hand, the next step is to determine the MLC charges for various IBM software products, e.g. z/OS operating system, DB2 databases, etc. This data will help decide if potential MLC cost management savings justifies an evaluation. NOTE: some BMC customers report MLC cost[3] reductions of more than 20%!

This review of BMC’s MLC Cost Management, only hits product highlights[4]. Functions of particular interest include:
  1. Cost Analysis – allows active management based on analyzing all system and cost data;
  2. Intelligent capping – to reduce costs by identifying and tracking system capacity peaks and costs drivers; 
  3. Optimizing sub-system placement – using pre-implementation “What if” testing of changes to quantify their potential cost impact to get the best results.

Let’s examine these.

First is the Cost Analyzer which breaks down the MLC charges into separate elements to identify what drives the peak elements being charged. “What-if” exercises reveal how changes affect MLC costs. A key new feature performs what-if analysis of moving batch jobs to a time when they will not impact MLC peak charges. BMC estimates a potential of 5-10% MLC cost savings from this product. BMC’s estimates could be conservative as some customers report larger savings.

Second is the Intelligent Capping, which lowers MLC costs by putting a cap or limit on the workloads running in a given LPAR. IBM offers very basic capping as a standard zOS feature – BMC’s technology uses intelligence-driven process automation to minimize overall costs. You can control how much or little of the recommended optimization to implement. BMC estimates reductions of between 5-10%. BMC has a patent-pending application for the cost-aware aspect of its intelligent capping technology.

Third is the Subsystem Optimizer, which reduces MLC by allowing customers to separate the running of certain IBM subsystems (like CICS) from other IBM subsystems it accesses (such as DB2), on to different LPARs. This can reduce billed MSUs for the subsystems, which can significantly impact MLC costs. It also has the ability to optimize cross-LPAR access to applications. BMC estimates it can reduce costs by 10-15%.

However, that’s not all. BMC Subsystem Optimizer has attractive features not strictly related to MLC cost, but which we believe will prove valuable. For example, for certain DB2 failures, it can automatically switch to another copy of DB2. This considerably increases the resilience of affected services and applications. Further, for IMS failures, Subsystem Optimizer can reroute IMS transaction requests to another IMS rather than waiting for the original IMS to restart (the default). It not only saves money but improves operational reliability by reducing the risk of an IMS subsystem failure. Both are great bonuses.

BMC began introducing these solutions about two years ago. They have been well received. We’ve mentioned a few benefits. By maintaining close contact with users, BMC is able to provide details on numerous customer experiences demonstrating product potential for direct cost savings as well as from automation. Watch for case studies to appear shortly.

BMC’s mainframe group has shown itself to be especially responsive to the concerns and efforts of mainframe users to optimize, not simply minimize, the costs and expense of mainframe operation. We highly recommend that mainframe users consider an in-depth product evaluation.

[1] BMC – Preparing Businesses for Digital Transformation,
[3] This doesn’t include some potentially significant personnel time savings. Replacing manual MLC cost management with BMC’s product can free-up the time of highly skilled, higher paid staff. A business case would include an estimate of such potential savings.
[4] A more detailed description requires discussing algorithms used by IBM to calculate MLC costs which exceeds the scope of this paper.