Hard Times May Boost Linux in Financial Services
Today Linux is the go-to operating system for high performance computing, while it continues to extend its footprint in the broader IT community. In the financial services arena, in particular, Linux is being seen as a critical technology for increasing ROI.
On Monday, at the High Performance Linux on Wall Street conference in New York, Inna Kuznetsova, director of IBM’s Linux Strategy, led a panel that discussed how Linux can be used to reduce costs and improve performance in these economically challenging times. We recently got the opportunity to ask Kuznetsova about the increasing profile of Linux for IBM customers and how the technology is enabling them to realize cost savings.
HPCwire: How have your clients IT priorities changed during this economic downturn?
Inna Kuznetsova: The priorities have certainly changed. First, we see our clients much more focused on the whole aspect of the total cost of ownership reduction. The financial crisis is putting a new lens on TCO claims. Companies prioritize projects with a clear ROI and predictable time to savings. The focus switches from enhancing functionality to cost reduction. Second, the expense control calls for higher accountability. Stimulus packages in different countries include funding for IT infrastructure but it comes with an increased scrutiny. To spend taxpayers’ money in a prudent way, reducing up front costs becomes increasingly important. Third, as we know from the experience of the previous recessions, companies often have to merge to withstand tough conditions — and that brings the issue of consolidating IT infrastructure. Being able to merge IT systems fast and keeping costs under control becomes vital for success.
HPCwire: How does Linux help address those new priorities?
Kuznetsova: Linux has unique attributes that help to improve savings. You cannot only reduce the costs with often lower rates but also eliminate CALs to avoid uprgrade penalties. Paying for a subscription instead of a license provides for a higher degree of flexibility should the customer decide to reduce resources, as often happens during an economic downturn. Standardizing on Linux reduces the number of skilled resources needed to manage multiple environments — and at the same time, a customer can select the best hardware platform for a particular workload. Also, during mergers, Linux, because it runs on the broadest set of hardware platforms, often becomes the “common denominator,” providing for a streamlined integration.
Consolidation on larger servers is one of the best ways to reduce energy consumption, office space and systems management costs, and many of our customers consolidate these days on Power Systems or mainframes running Linux. IDC recently published a study based on applying quantitative measures to such consolidation and quoted an average ROI in less than seven months and the reduction of TCO by 50 percent. Many of our financial customers follow this route — for example, the Bank of New Zealand consolidated its front office, Internet banking and tellers system on an IBM System z mainframe running Linux. It has reported a decrease in energy consumption close to 40 percent, 33 percent reduction in heat output and a need for only one administrator per 200 virtual servers.
Last but not least, customers can achieve great savings by leveraging Linux desktop in a traditional or virtual implementation. IBM offers an open-standards-based alternative to Microsoft Office, called Open Collaboration Client Solution (OCCS). It is a package of office productivity tools, including email, messaging and social collaboration software, that can work in a heterogeneous environment, with some users working on Linux and some on Windows. Users save 500-700 dollars per workspace when switching to Linux with IBM OCCS, and they can keep users on different systems collaborating with each other, using the same team rooms and tools. There are further savings associated with the virtual Linux desktop, launched last December, which can be deployed from any x86 server and provides additional savings in energy consumption, reduced need for workstation upgrades, memory, and system administration.
HPCwire: Are there instances where a proprietary OS is preferable?
Kuznetsova: There is no “one size fits all” approach in the selection of an operating system — many factors come into play. The customers have to consider what workloads and what applications they are running, what skills they have in house, what are their requirements to the reliability, availability, security, level of service, costs and resources for system administration.
An application migration is a costly process and may not be justified by savings. Furthermore, many workloads and applications may be designed in such a way that they use possible benefits that come from the combination of an operating system and hardware designed by the same vendor. Such combinations will always be better optimized for a particular platform while Linux is better optimized across vendors and platforms just by the nature of its development process.
HPCwire: In the financial services industry specifically, what is driving Linux adoption?
Kuznetsova: The financial services industry leads Linux adoption today — with the majority of our customers running the front-end operations on it and many exploring the opportunities that Linux offers for the back office. Last year I hosted a panel at the Linux on Wall Street event where we had a very interesting discussion about the use of Real Time for latency reduction and fast trade execution.
Our customers increasingly leverage Linux to select the best hardware platform for each workload. Thus, Mizuho Securities uses IBM BladeCenter with Cell broadband Engine processor to run its Exotic Derivatives Trading system. Cell has superb computational capabilities that are well suited to accelerating financial modeling calculations.
A lot of our customers consolidate operations on IBM System z running Linux, leveraging the high level of reliability, availability and security, as well as the low cost of system administration provided by mainframes. For example, the Bank of Russia went from 74 datacenters to a consolidated system on four IBM System z mainframes running Linux, reducing the per-transaction cost by 95 percent, which resulted in $400 million savings.
Other customers increasingly use Linux on System x for fast server provisioning at minimal cost. Antony Golia from Morgan Stanley will join me on stage at the Linux on Wall Street event to talk about their experience.
HPCwire: For HPC applications in particular, performance and the broader concept of productivity are the critical issues. How does Linux meet these needs?
Kuznetsova: Linux offers exceptional performance and productivity for HPC. It is enough to mention that over 80 percent of TOP500 supercomputers today are powered by Linux, including the Los Alamos system called Roadrunner – an IBM BladeCenter QS22 cluster — that tops the list. At IBM we invest in supporting Linux as a tier one system on all our server platforms. It is much more than just ensuring that Linux can run on a server — it means performing additional development, tuning and testing work to provide for a high level of reliability, availability and security (RAS). Some vendors claim that Linux is not yet ready for prime-time. Linux on IBM platforms supports business-critical operations with a highest level of RAS requirements today.
HPCwire: If Linux had never been developed, what would the landscape look like for your clients today?
Kuznetsova: Last December we celebrated the 10th anniversary of IBM’s committment to Linux. During this time we became one of the largest users of Linux — from running it on over 25,000 employee desktops to consolidating 3,900 servers in our IT infrastructure using Linux on mainframes. We became one of the largest contributors to the Linux community, being the third largest provider of changes to the Linux kernel over the last three years, having several hundred developers employed in the Linux Technology Center and participating in over one hundred open source projects. And we became one of the largest providers of Linux solutions — Linux is supported as a tier one system on all IBM server plaforms and we have over 500 software products for Linux in the market. It is difficult to imagine a different course of history and I would like to believe that there would have been a different project leveraging the intellectual power of a broad community to develop a system optimized for all hardware platforms and offering customers an open standards alternative today.
Feeds by Topic
- Developer Tools
Feeds by Industry
November 20, 2014
- Penguin Computing Announces New Collaboration with AMD
- Altair Partners with South Africa’s Centre for HPC
- TACC Receives Honor in 2014 HPCwire Awards
- UMich and UVic Utilizing SanDisk Products to Fuel Supercomputing Project
- Altair’s PBS Works Recognized in HPCwire Awards
- OpenMP ARB Releases TR3
- InfiniBand Interconnect Technology Used in 45 Percent of TOP500 Supercomputers
November 19, 2014
- DDN Honored with Six HPCwire Awards
- Cray Adds Cloudera Enterprise to Urika-XA System
- OpenMP ARB Announces Code Conversion Challenge
- Bright Computing to Demonstrate Latest Software at SC14
- Mellanox Announces ConnectX-3 Pro Adapter Card with VPI Technology
- K Computer Recognized in Class 1 and 2 of HPC Challenge Awards
- Allinea DDT for the ARMv8 Platform Released
- Mellanox and Saratoga Speed Contribute to Work at LLNL
- OpenMP ARB Releases Technical Report on Tools API for Performance Analysis
November 18, 2014
- AMAX to Introduce ServMax OCS to HPC Community
- Seagate Unveils Apache Hadoop on Lustre Connector
- Quanta Recognized with HPCwire Award
- Numascale, Supermicro, and AMD Announce World’s Largest Shared Memory System to Date
Most Read Features
- The Exascale Revolution
- CORAL Signals New Dawn for Exascale Ambitions
- No Slowdown in Sight for Supercomputer Maker Cray
- The Strange, Shifting Shape of the Supercomputing Market
- Why the Top 500 Standstill Won’t Last Forever
- CERN Details OpenStack Journey
- NextScale Nodes Crest 1 Teraflops, Await Knights Landing
- Jülich Tag Teams With IBM, Nvidia On Data Centric Computing
- Supercomputing Wrap: Top Stories from SC14
- New Architectures for HPC with Amazon Web Services
- More Features…
Most Read Short Takes
- Poland Steps Up Supercomputing Game
- 50 Computer Science Schools That Are Changing the World
- CERN Researchers Explore x86 Alternatives
- Supercomputer Models Human Brain Activity
- VMware Reports HPC Progress
- DOE HPC: A Change Is Coming
- From Quantified Self to Personalized Medicine
- LANL Showcases Supercomputing History
- DARPA Targets Autocomplete for Programmers
- Is the City of the Future Finally Here?
- More Short Takes…
Most Read Off The Wire
- Intel Announces Several New and Enhanced HPC Technologies
- ORNL to Acquire New Supercomputer
- NVIDIA Unveils Tesla K80 Dual-GPU Accelerator
- AMD Awarded $32M for HPC Research
- DOE to Build Supercomputers for ORNL and LLNL
- HPCwire Reveals Winners of the 2014 Readers’ and Editors’ Choice Awards at SC14
- Dell PowerEdge C4130 Server Utilizing Mellanox FDR InfiniBand Adapters
- Dell Announces New HPC Offerings
- Sandia Releases API to Help Standardize Supercomputer Power
- Brookhaven Lab Embarks Upon New Computational Science Initiative
- More Off The Wire…
- Read more…
- Read more…
- Read more…
- Read more…
- Read more…
- Read more…
- 11/6/14 | Nimbis Services | For manufacturers, HPC cloud is primed and ready to bring to the small-to medium-sized enterprises the very tools that keep industry giants on Read more…
- 9/26/14 | Altair | The University of Nottingham chose Altair’s PBS Works suite as the integrated workload management solution for their HPC system comprising over Read more…
- Read more…
- Read more…