Linux Kernel Components : IBMhas made significant contributions to most of the major subsystems includingclustering, scalability, containers, virtualization, devicedrivers, documentation, file systems, internationalization,networking, performance, RAS, real time, security, storage, systems management, and more. Linux Toolchain : IBMhas contributed to many develop-ment tools such as GCC, GLIBC, GDB, and oprofile. Real-time Linux : Real-time Linux provides technology forrunning real-time workloads that require predictableresponse times. KVM and Xen : Open source virtualization technology thatallows users to run multiple operating systems concur-rently on the same physical box. Common Criteria Certification : Whispered by many to be impossible, IBM was a key leader driving common criteriacertification for Linux to the highest level of certification ofany general purpose operating system. Apache : The most popular Web server in the world. Eclipse : An open development platform( http://www.eclipse.org ) founded by IBMand others in2001. It has grown to be a hugely successful communityfor building, deploying, and managing software. Samba : Provides file and print services to all types ofSMB/CIFS clients. InfiniBand : A development effort to add the InfiniBandcommunication fabric stack to the Linux kernel. Community Linux Distributions: Ubuntu, Debian, Fedora,Asianux, and more
Let’s talk about what the product is. As I mentioned, it’s a system of systems. And what we - what that does for our clients is it helps unify IT across a heterogeneous environment to produce predictable service delivery models as three main components. On the left the z Enterprise 196, the world’s fastest server. It’s ideal for large data and transaction serving in mission critical applications just as Z has always been. But even more so with the performance and scalability of this platform and additional enhancements such as the new release z/OS, z/OS 1.12. It’s also the lowest cost platform for large scale Linux consolidation. And finally, it’s capable of massive scale. Over 50 billion instructions per second if you want to calculated. It’s interesting, we now are kind of switching from talking from mps, now we’re going to be talking about bps. It’s a truly massive scale. The next stage of this is something we call the unified resource management. And this really is the secret sauce. It allows us to integrate workloads running on different platforms. Now those platforms could be z/OS, they can be Linux on System z, they can be IBM Blade, specifically Power 7and System X Blades for more scale out applications. The unified resource manager gives us a way to integrate the management of the workload platforms and hardware and produce the kind of workload management normally only available in a z/OS environment. By moving the z/OS capabilities for workload management into the Firmware we’re able to provide that workload management to improve application availability and our clients’ ability to meet service level in a z/OS, a Linux on Z and a heterogeneous environment with Power 7 and X86. Really unifying that management is the key concept here. And on the right you see the chassis that allows us then to plug those into a System z environment. It’s based on the IBM Blade Center and it is an extension to the System z. It’s shipped from the factory, configured and tested with private data and management networks for high speed low (unintelligible) communication between the Blade environment on the right and the System z Enterprise on the left. It includes both general purpose Blades based on Power 7 and X86 and specific optimizers that accelerate time to insight and reduced costs. Now I will tell you to make a comment here on availability without specific dates. The system on the left, the Enterprise 196 that will GA later this summer. The extension of that with the IBM zEnterprise BladeCenter Extension (zBX) will occur in the fourth quarter of this year with Power 7 and the optimizers. And in 2011 for the inclusion of X86 applications and Blades in that environment. So it’ll be a phased environment. And that’s actually what our clients are looking for because they really want to be able to get the new System z installed and then start extending that out to their integrated environment. Let’s talk about the z - the Enterprise 196. Why do we call it a 196?
According to the CIO study conducted by IBM, interviewing 2500 CIOs in 75 countries: “ 76% of CIOs cited “implementing a virtualized computing environment” as part of their visionary plans to enhance competitiveness” [Click] Let’s learn how System z delivers better virtualization than any other platform Virtualization is architected into the system, not an “add on” feature. In fact, virtualization was pioneered by IBM on the MF over 40 years ago [Click] There is support in the hardware, with 10% of the circuits dedicated for virtualization There 2 levels of hypervisor capabilities: one at the firmware level, by PR/SM 2 , and the second at the software level via zVM Each server can be partitioned into 60 logical partitions, called LPARs, with each LPAR behaving just like a single, physically separate, mainframe And each LPAR can be further virtualized into 100s of virtual servers, all with the ability to share memory, processors and other resources, physically or virtually Note that PR/SM also has EAL5 certification, that provides assurances that many different applications running in different operating environments in different logical partitions on one z10 BC will be secure and distinct from each other. [Click] System z Deploy virtual servers in seconds, Deploying single virtual server on the newest System z server is estimated to cost just US$1K over 3 years 1 Add physical resources without taking the system down: scale out to 100s or 1000s of virtual servers inside one machine Do more with less: host more virtual servers per core and share more physical resources across all the virtual servers There is a Policy-driven workload management Extensive built-in facilities provide life-cycle management: provisioning, monitoring, security, charge-back, backup, more… zEnterprise Unified Resource Manager can manage 10x the number of virtual machines than VMWare. 3 The potential performance impact of the Linux server farm spawned from one Logical Partition is isolated from the other LPARs [Click] Distributed environment Average 10-20% utilization per server Limited per-core virtual server scalability Physical server sprawl is needed to scale Operational complexity increases as virtual server images grow VMWare, Xen, Virtual Iron focus only on x86 environments and not on cross platform hardware management. Rest is additional info: Customer Pain Points – addressed by z virtualization Underutilized IT assets Escalating energy and space costs Linear software costs per server image Staff inefficiencies managing multiple real servers Spiraling people costs Footnotes : Claims doc July 20, 2010 1 Linux for System z virtual servers can be (on z196 server): • Less than $1000 for 3 years • Less than $335 per year • Less than $1 per day IBM Study: US Enterprise Linux Server pricing, pricing may vary by country. Based on zEnterprise ELS configured with 64 IFL cores running a mixed workload averaging 47 virtual machines per core with varying degrees of activity. Includes zEnterprise hardware and z/VM virtualization software. Does not include Linux OS or middleware software. 2 PR/SM = Processor Resource/Systems Manager Evaluation Assurance Level 5 certification of IBM Processor Resource/Systems Manager (PR/SM) logical partition (LPAR) for the IBM System z10 Business Class (z10 BC) server was received on May 5, 2009. EAL5 certification provides assurances that many different applications running in different operating environments in different logical partitions on one z10 BC will be secure and distinct from each other. To view the certification report, visit BSI at http:// www.bsi.de . 3 http://www.vmware.com/pdf/vsphere4/r40/vsp_40_config_max.pdf ‘“ VM stall” the next big virtualization challenge’ – article on problems with scaling and managing virtualization that is stalling further deployment of virtualization …. http://www.networkworld.com/news/2010/060110-is-vm-stall-the-next.html?page=1
Based on IBM internal studies and customer case studies, consolidating workloads of distributed servers into a single z server can deliver: Up to 80% reduction in energy consumption 1 Up to 90% reduction in floor space usage 2 Up to 90% reduction in software license costs 3 The financial impact of consolidation onto System z can be substantial, with savings of up to 70 percent in total cost of ownership compared to distributed platforms. 4 Actual savings may vary by customer Footnotes and source notes: Claims doc – July 20, 2010 1 Reduce labor, energy , and development costs, by up to 70%, 80%, and 20% (respectively) when consolidating workloads to Linux on System z. 2 Reduce floor space by up to 90% 3 Up to 90% reduction in SW costs when consolidating distributed Oracle licenses on to Linux on System z. 1,2,3 IBM Internal Study based on Oracle SW with Nehalem 4 core running 25% utilization, consolidating databases on z196 Deployment configurations based on IBM studies and will vary based on workload characteristics. Price calculations based on publicly available US list prices, prices will vary by country. Distributed server comparison is based on IBM cost modeling of Linux on zEnterprise vs. alternative distributed servers. Given there are multiple factors in this analysis such as utilization rates, application type and local pricing, etc. savings may vary by customer. 4 ECM ‘Big Green’ Project - IBM Case: consolidate 292 Sun servers to Linux on System z10 and save 95% on power and cooling, and 97% on floor space reduction. 2, 3 (1) p16 – p18 …. not used here, but added to illustrate that cases will vary depending on conditions at customer.
Let’s go deeper into the System z to explain how it can deliver scalability, and green savings better than distributed platforms A System z server can be run up to 100% utilization compared to an average of 10-20% utilization for distributed servers It is architected and designed to share resources for maximum utilization across multiple applications Yes, you can run multiple applications, of mixed workloads, OLTP, web, data serving, on a single z server By contrast, you typically run a single app on one server in the distributed environment This sharing of resources - includes the processors, memory, network, adapters, cryptography All of which contributes to a lower power consumption per unit of work and hence contribute to significant customer savings
Press release: http://blogs.eweek.com/ibm-oracle/content/business_intelligence_/system_z_proving_to_be_a_big_catch_for_fishermen.html Fish Market Fishermen can use a &quot;touch screen&quot; installed on their boats to describe which kind of fish they are catching. If that fish is in demand by the market, the solution starts an electronic auction. If the bid price is below certain level the fish sale is redirected to non-profit company, e.g. soup kitchen. If the price is above the level, the solution automatically specifies the quantity to be distributed into boxes ready for shipping and distribution upon arrival. Solution is available on the zCloud for use by different fish markets (distributed on territory). Wine market Winemakers at up to 60 cooperative wineries are able to determine market demand for various types of wines by accessing the cloud computing-based systems, then package and ship wines demanding the highest price. MoniCA: Monitoraggio e Controllo Adattivo Logistics solution that tracks and collects data in real time from trucks (e.g. temperature, humidity, shocks and routes). Solution will be available on the zCloud to serve different logistics companies.
Focus Group Open Source 09.05.2011 Massimiliano Belardi
Massimiliano Belardi Linux su Mainframe “ 10 anni di esperienze con i clienti”
Source: Linux Kernel Development - August 2009 <ul><li>“ Linux brings open innovation to all IBM server and storage system platforms, freeing datacenters from vendor lock-in with choice and flexibility to scale your business on the fastest growing operating system in the world*”. </li></ul><ul><li>IBM is among the top Companies in Linux Kermel contribution. </li></ul><ul><ul><li>IBM holds ongoing, significant roles in a large number ofLinux-related and other open source projects, as*: </li></ul></ul><ul><ul><li>- Linux Kernel Components </li></ul></ul><ul><ul><li>- KVM and Xen </li></ul></ul><ul><ul><li>- Apache, Samba </li></ul></ul><ul><ul><li>- Common Criteria Certification </li></ul></ul><ul><ul><li>- Infiniband </li></ul></ul><ul><ul><li>-Community Linux Distributions </li></ul></ul>*Source : IBM is committed to Linux and Open Source. http://public.dhe.ibm.com/common/ssi/ecm/en/lxb03001usen/LXB03001USEN.PDF <ul><li>Linux on System z is not a special Linux </li></ul><ul><ul><li>- Pure ASCII environment! </li></ul></ul><ul><ul><li>- Does not need any other OS to run on a mainframe </li></ul></ul><ul><ul><li>- Not a replacement for any other OS on IBM System z </li></ul></ul><ul><ul><li>- Novell-SuSE and RedHat are the supported distributions </li></ul></ul>
Linux on IBM System z in 2Q2010 <ul><li>The momentum continues: </li></ul><ul><ul><li>Shipped IFL engine volumes increased 6% from 1H09 to 1H10 </li></ul></ul><ul><ul><li>Installed IFL MIPS increased 8% from 1Q10 to 2Q10 </li></ul></ul><ul><ul><li>Installed IFL MIPS increased 37% from 2Q09 to 2Q10 </li></ul></ul><ul><li>More than 3,200 applications available for Linux </li></ul><ul><ul><li>80 new ISV applications ported to Linux on System z in 1H10 </li></ul></ul><ul><li>Linux is 18% of the System z customer install base (MIPS) </li></ul><ul><li>70% of the top-100-System z-clients are running Linux on System z </li></ul><ul><ul><li>Average number of IFLs installed at top-100-clients grew from 8 IFLs in 4Q09 to 11 IFLs in 2Q10 </li></ul></ul><ul><li>Two Linux partners: Novell SUSE and Red Hat </li></ul><ul><li>Gold standard in virtualization with z/VM ® </li></ul>
IBM zEnterprise ™ System – Best-in-class systems and software technologies A “System of Systems” that unifies IT for predictable service delivery IBM zEnterprise 196 (z196) <ul><li>Unifies management of resources, extending IBM System z qualities of service end-to-end across workloads </li></ul><ul><li>Provides platform, hardware and workload management </li></ul><ul><li>Optimized to host large-scale database, transaction, and mission-critical applications </li></ul><ul><li>The most efficient platform for large-scale Linux ® consolidation </li></ul><ul><li>Capable of massive scale-up </li></ul><ul><li>New easy-to-use z/OS V1.12 </li></ul>zEnterprise Unified Resource Manager <ul><li>Selected IBM POWER7 blades and IBM System x Blades* for tens of thousands of AIX and Linux applications </li></ul><ul><li>High-performance optimizers and appliances to accelerate time to insight and reduce cost </li></ul><ul><li>Dedicated high-performance private network </li></ul>* All statements regarding IBM future direction and intent are subject to change or withdrawal without notice, and represents goals and objectives only. zEnterprise BladeCenter Extension (zBX)
<ul><li>Extended capabilities for server consolidation and IT optimization </li></ul><ul><li>Extended qualities of service </li></ul>Designed for Greater Savings, Operational Simplification and System Reliability <ul><li>I mproved total system capacity </li></ul><ul><ul><li>Up to 80 user-configurable cores </li></ul></ul><ul><li>Up to 3 TB of real memory per server </li></ul><ul><li>Increased internal networks (HiperSockets ™ ) </li></ul><ul><li>Quad-core 5.2 GHz processor </li></ul><ul><ul><li>New instructions to accelerate Java ™ code </li></ul></ul><ul><ul><li>Designed to improve processor-intensive workloads </li></ul></ul><ul><li>Ability to extend amount of addressable storage capacity </li></ul><ul><li>Improved power efficiency </li></ul><ul><li>Concurrent add/delete of I/O </li></ul><ul><li>Redundant array of independent memory (RAIM) </li></ul><ul><li>HiperSockets network traffic analyzer </li></ul><ul><li>Crypto Express3 cryptographic enhancements </li></ul>Linux on zEnterprise 196 Dramatic improvement: Improvement in performance Less cost <ul><li>With no increase in energy consumption </li></ul><ul><li>And even better performance with new software </li></ul>60% Up to for 35%
System z’s Extreme Virtualization Built into the architecture not an “add on” feature LPAR – Up to 60 Logical partitions <ul><li>System z </li></ul><ul><li>Deploy virtual servers in seconds </li></ul><ul><li>Highly granular resource sharing (<1%) </li></ul><ul><li>Add physical resources without taking system down, scale out to 1000s of virtual servers </li></ul><ul><li>Do more with less : More virtual servers per core, Share more physical resources across servers </li></ul><ul><li>Extensive virtual server life-cycle management </li></ul><ul><li>Hardware-enforced isolation </li></ul><ul><li>Distributed Platforms </li></ul><ul><li>Limited per-core virtual server scalability </li></ul><ul><li>Physical server sprawl is needed to scale </li></ul><ul><li>Operational complexity increases as virtual server images grow </li></ul><ul><li>VMWare, Xen, Virtual Iron focus only on x86, not cross platform hardware management. </li></ul>PR/SM ™ z/VM – 100s of virtual servers – Shared Memory
<ul><li>Increasing Customer Value </li></ul><ul><li>Up to 80% reduction in energy cost 2 </li></ul><ul><li>Up to 70% reduction in labor cost 2 </li></ul><ul><li>Up to 90% reduction in software costs 3 </li></ul>z196 Delivers Impressive IFL Scalability Enhancements <ul><li>Consolidate more virtual Linux servers per core - per z/VM LPAR - per z196 </li></ul><ul><li>Require even fewer physical servers, fewer network devices, fewer switches, less disk space, less energy, and less floor space </li></ul><ul><li>Spend even less on software license fees with extreme levels of resource sharing and workload consolidation </li></ul><ul><li>Increase the productivity of your IT staff as they manage even more server images on a single zEnterprise 196 server </li></ul>1 Client results will vary by types of workloads, technology level of consolidated servers, utilization factor, and other implementation requirements. Savings will vary by client. 2 When consolidating distributed environments on to Linux on System z. IBM Internal Study based on Oracle SW with Nehalem 4 core running 25% utilization. Results may vary. 3 When consolidating distributed Oracle licenses on to Linux on System z. IBM Internal Study based on Oracle SW with Nehalem 4 core running 25% utilization. Results may vary. Run 50 or more virtual servers per core 1 Potentially host 1000s of Linux server images using z/VM 1
Maximizing Utilization of Resources 1 Source: gomainframe.com Joe Clabby <ul><li>Up to 100% server utilization compared to 10-20% distributed server utilization 1 </li></ul><ul><li>Shared everything infrastructure allows for maximum utilization of resources </li></ul><ul><ul><li>CPU, Memory, Network, Adapters, Cryptography, Devices </li></ul></ul>Moderate distributed servers Up to 100% utilized System z server Multiple applications on one physical System z server Typically single application per physical server
Universita di Bari Solve community challenges Solution Edition for Cloud Computing Press release: http://www-03.ibm.com/press/us/en/pressrelease/32051.wss Innovative Cloud Solutions BENEFITS to Clients Universita di Bari, established in 1924, is developing cloud-based solutions for a consortium of companies and universities from five regions of southern Italy. Fish Market Electronic fish auction for fishermen while on boats Wine Market Support for 60 wineries to determine demand and get best market price MoniCA Logistics solution tracks and collects data real time Cloud computing allows multiple organizations to tap into heavy-duty computing power at minimal cost. It lowers the barrier for local businesses to benefit from this technology.
Massimiliano Belardi System z Sales Specialist – IBM Italy Mobile: +39 335 7248414 [email_address] http://it.linkedin.com/in/massimilianobelardi