Cloud Computing for Rethinking IT and Reinventing Business
In IBM’s Global Chief Information Officer Study– “The Essential CIO” – based on a poll of 3,000 global CIOs– shows that cloud computing has come of age with interest in the cloud increasing more than any other CIO priority. As I mentioned earlier, this 2010 study shows that 60 percent of organizations are ready to embrace the cloud over the next five years, an increase that is nearly double that of the 2009 study.
The cloud represents a profound evolution of IT with revolutionary implications for business and society, creating new possibilities and enabling more efficient, flexible and collaborative computing models. The cloud enables “IT without boundaries”—systems and processes that break down traditional silos and simplify access to information in order to deliver better business outcomes. Cloud computing offers organizations dramatic increases in agility and efficiency— innovation to ensure speedy, cost-effective delivery of products and services. To realize the benefits of cloud computing while overcoming the inherent challenges, organizations must take a holistic approach that spans business and innovation, users and applications, systems and networks.
IBM enables Smarter Computing managed in the cloud with a wide range of solutions. For example, IBM Tivoli Service Automation Manager enables users to request, deploy, monitor and manage cloud computing services. IBM SmartCloud Managed Backup
services provide end-to-end, cloud-based managed services to help protect business data, regardless of where it's stored. IBM SmartCloud Enterprise
is an agile cloud infrastructure as a service (IaaS) designed to provide rapid access to security-rich, enterprise-class virtual server environments, well suited for development and test activities and other dynamic workloads. IBM Smart Business Storage Cloud
offers a storage-virtualization solution designed to support storage optimization efforts. The IBM BladeCenter Foundation for Cloud
provides a pre-integrated and preloaded system with software, server, storage, networking and start-up services to help take the guesswork out of establishing a virtualized data center environment.
System networking is essential to the cloud, and our Cloud Ready Network Architecture
addresses the network-related requirements for private and public cloud computing. As enterprises and cloud providers seek to harness the considerable advantages of cloud computing, their networks must be equipped with five essential elements -- high-bandwidth/low-latency switching, convergence to Ethernet, massive virtualization for agile workloads, scalable fabric management and advanced energy efficiency. Cloud Ready Network Architecture is extending the performance and capabilities of data center networks from blade server to switch to SAN that will enable the scale-out performance and economies promised by cloud computing. Our Cloud Ready products
include the industry's first FCoE-ready 10 Gigabit Ethernet blade server switch, a high-performance 10/40GbE top-of-rack switching solution and a massive virtualization solution that supports more than 1,000 virtual ports for the cloud-ready data center.
Cloud computing is a new model of consuming and delivering IT and business services. It enables users to get what they need, as they need it—from advanced analytics and business applications to IT infrastructure and platform services, including virtual servers, storage and networks. The cloud is only as agile and extensible as the underlying network, and IBM System Networking is committed to delivering networking solutions that enable clients to harness the cloud’s transformative potential to rethink IT and reinvent business.
Harnessing the Full Potential of Virtualization
With VMworld 2011 taking place this week in Las Vegas (August 29 to September 1), the industry’s attention is riveted on virtualization. IBM provides deep virtualization expertise and an integrated portfolio of solutions to enable and support business and IT virtualization objectives. IBM approaches transformation from a holistic perspective with smarter systems, software, and services, enabling virtualization not only at the platform level, but also across operating systems and middleware.
Virtualization brings both benefits and drawbacks to the data center: it can maximize underutilized resources and minimize infrastructure spending—but add complexity and administrative overhead for the network administrator.
Today, among the biggest problems clients are facing with their virtual server and storage infrastructures is that when VMs move, the network that connects these server and storage devices is not VM aware, relying instead on conventional physical device awareness. Thus, with a network that is not VM-aware, when VMs move, network addresses, security policies and class of service policies need to be re-configured manually. Enabling the network with awareness of virtualization is what is needed.
This inability of conventional data center networks to understand the language of virtualization and see Virtual Machines (VMs) as they move from server to server and even from data center to data center poses a major impediment to the use of virtualization, particularly for production application workloads, where performance, availability, and security are bottom-line priorities. Those who manage physical and virtual infrastructures know that the network must keep pace with virtualization, and they are increasingly vocal about their desire for scalable, secure and open solutions for building a virtualization-aware network.
To fulfill the promise of smarter computing, Virtual Machines (VMs) need to be supported by an intelligent network. IBM System Networking’s switch-resident VMready® with Virtual Vision is a unique solution that enables the network to be Virtual Machine aware, so that the network can be configured, managed and secured for 1000s of virtual ports (v-ports).
With VMready, as VMs migrate across physical hosts, so do their network attributes automatically, retaining the same ACLs, QoS and VLAN attributes. VMready will support the emerging Edge Virtual Bridging (EVB) technologies now being defined by the IEEE 802.1Qbg working group to make it easier for businesses to achieve server-network virtualization in the data center.
Our clients have been crystal clear in expressing their desire for networks that can scale, secure and migrate VMs even across long distances between data centers. And, they want to achieve this automated network change management without locking into proprietary stacks or having to rip and replace their existing virtual infrastructures. With VMready and Virtual Vision, IBM is enabling data center operators to maximize their investment in server virtualization.
If you’re attending VMworld 2011, please make sure to learn more about VMready by visiting IBM in Booth #321.
IBM BNT RackSwitch 8264 Excels in Independent Test
Third-party product testing can provide a valuable resource for determining which networking products are best suited to customer needs. Third-party tests can often be accomplished on a much more extensive level with more products tested, more test equipment used and more test parameters evaluated than any single client could accomplish on their own. Consider a new independent performance test, which validates that the IBM BNT RackSwitch G8264 10/40 Gigabit Ethernet switch demonstrates significant performance and energy efficiency advantages over other top-of-rack switches offered in the market.
In the performance test conducted by The Tolly Group, the IBM BNT RackSwitch G8264 consistently demonstrated 100-percent line-rate throughput, lower latency, and the capability to buffer more packets than similar products in the markets, while providing an additional 16 10GbE ports, or 160Gbps more capacity than all other switches tested.
In addition to the performance advantages, the IBM BNT RackSwitch G8264 consumed up fewer Watts per Gbps than all other switches tested. The IBM BNT RackSwitch G8264 also demonstrated an average of 55% better price/performance than the three other switches.
Kevin Tolly, founder of the Tolly Group had this to say, “Today’s data centers are serving content for fixed and mobile clients across large scale public and private cloud computing clusters. Video rendering, high frequency trading and oil reservoir simulations are just a few examples of the complex applications being served over computing clusters connected at 10 Gigabit Ethernet (10GbE). The IBM BNT RackSwitch G8264 is a high-density 10GbE Top-of- Rack (ToR) switch with a powerful combination of the performance, low latency and energy efficiency required for today’s consolidated networking and storage traffic as well as server virtualization.”
The Tolly Group’s test clearly demonstrates the performance and energy efficiency advantages of the IBM BNT RackSwitch G8264. We are proud that IBM System Networking’s RackSwitch products indeed excel at delivering intelligence and speed at the edge of the network, the essential access, distribution and aggregation layers where essential server and storage systems are connected to the data center network.
IBM Leads All in 40G Ethernet Shipments
I am proud to report that IBM holds the leadership position in the emerging market for 40 Gigabit Ethernet in the data center. According to research firm Dell’Oro Group, which recently began tracking the 40G Ethernet market, IBM is the market leader. This leadership position is due to the continued growth of 10G Ethernet, which necessitates 40G uplinks and aggregation switches as delivered by our innovative IBM RackSwitch G8264 and IBM RackSwitch G8316 top-of-rack switches.
In Dell'Oro's first 40G market report for the third calendar quarter of 2011, IBM RackSwitch products held a market share of more than 69%, more than 4 times greater than any other vendor.
The market for 40G Ethernet is still in its infancy. However, as data centers and cloud operators increasingly deploy servers equipped with 10G Ethernet, this drives the need for 40G upstream connectivity. The IBM RackSwitch G8264 and G8316 are ideal in this regard because they can be equipped with 10G server interconnects and 10G uplinks that can be easily upgraded to 40G as requirements demand.
The IBM RackSwitch G8264 is a 10/40G top-of-rack switch specifically designed for applications requiring the highest performance. It combines 1.28 Tbps throughput with up to 4 40G ports or up to 64 10G ports. The RackSwitch G8316 is a 40G top-of-rack aggregation switch that can be configured with either 16 40G ports or up to 64 10G ports. These switches set IBM apart and our leadership position in the emerging market for 40G Ethernet demonstrates that our innovative products are being well received in the marketplace.
IBM System Sets New Low-Latency Record for Real-time Market Data over 10GbE
In the world of High-Frequency Trading (HFT), opportunities exist only fleetingly and therefore trading solutions must run at the lowest latency to be competitive . Low-latency 10 Gigabit Ethernet has become the interconnect
of choice for HFT solutions. IBM and Mellanox have demonstrated a solution that
performs at high throughput rates and low latency to facilitate High-Frequency
I am happy to report that an IBM system has set a new record in low-latency messaging performance for Ethernet networks based on
the latest STAC-M2 benchmark test conducted by
Mellanox. The record was achieved using IBM x3550 servers running IBM’s WebSphere MQ Low Latency Messaging (LLM) technology using
Mellanox ConnectX-2 EN 10GbE NICs with RDMA over Converged Ethernet (RoCE) and the IBM BNT RackSwitch
G8264 10GbE switch.
The IBM and Mellanox solution outperformed the most recent record holder (based on a comparable configuration), achieving a 30% improvement
over results that were announced just two months ago. The benchmark results highlight an enhancement in latency reduction and scaling of messaging
performance, validating IBM and Mellanox’s performance leadership for high-performance connectivity to data centers in the financial services
STAC-M2 is a vendor neutral benchmark set by the Securities Technology Analysis Center (STAC) Benchmark Council, a consortium of
financial and trading industry participants that sets testing standards and supports performance improvements for technology used in capital markets.
This STAC benchmark demonstrates that IBM and Mellanox are addressing the performance requirements of the financial services industry by providing financial services enterprises with innovative server,
storage, software and networking solutions that meet their customers’ needs for fast transactions and value-added services.
STAC Benchmark Council members with a premium subscription can request the full STAC® Report. The report highlights are available to the
public. For more information, read IBM’s white paper about this impressive
low-latency solution for High Frequency Trading. Find more information about IBM System Networking’s low latency solutions here.
IBM’s “Blue Pods”
With standardized components available for virtually every element of IT infrastructure from server to storage and network to rack, it’s perhaps no surprise that the standardized data center or “Pod” is becoming increasingly popular. For example, IBM’s Portable Modular Data Center is a data center in 20- or 40-foot shipping container that can be located anywhere that power and chilled water can be delivered. Other “Blue Pods” include the Enterprise Modular Data Center, a standardized raised-floor data center that can be built out in modules of 5,000 to 20,000 square feet, and the Modular High-Density Zone, which can convert an existing raised-floor space into an area that can deliver more computing capacity and power density.
As organizations struggle to handle growth in existing data centers, most organizations need to install and operate high-density computing to keep up with IT demands. However, the majority of today’s data centers are not designed to support high-density computing. So, these modular data centers offer a quick-to-deploy high-density solution with exceptional energy efficiency. And these modular approaches are ideal for organizations that want affordable capacity on demand. A Portable Modular Data Center can be designed, built and drop-shipped in as little as a few weeks to any location in the world, and does not require onerous capital outlays.
An IBM Portable Modular Data Center can cost up to 30 percent less to design and build compared to custom raised floor solutions, and can have up to a 50 percent smaller footprint. These Blue Pods are open, using industry-standard 19-inch racks and can house 1,428 blade servers or 1,178 IBM iDataPlex servers per container. This high-density compute power requires equivalent networking capacity. And this is where IBM System Networking is an ideal complement. IBM System Networking RackSwitch products consume power equivalent to standard light bulbs and are designed with server-friendly airflow that matches the hot-aisle/cold-aisle designs of the modular data center. IBM BladeCenter switches consume even less power and are tightly integrated with the BladeCenter architecture. Our VMready switch-resident software provides the networking capability to equip the Pod for the challenges of virtualization.
As CIOs remake IT infrastructures into more agile and automated environments that can respond to complexity with operational flexibility, Blue Pods support change while helping to mitigate associated risks. They help to address the continuing pressure to do more, faster and better – with less. In many cases, Pods enable organizations to accelerate their preparations for increasing complexity without requiring a significant expansion of existing IT operations. And Pods address the imperative to do more, faster and better—with less. Pods enable the data center to perform as the engine of business growth. And that’s what smarter computing is all about.
Take an online tour of an IBM Portable Modular Data Center.
Interop Las Vegas is Showcase for IBM’s Collaborative Innovations
Interop is a singular event that galvanizes our industry. This year at Interop Las Vegas May 8-12, IBM System Networking will showcase the latest developments in collaborative innovations for Smarter Computing. For example, the emerging OpenFlow specification promises a greater level of intelligence in Ethernet networks through a new approach called Software-Defined Networking (SDN).
OpenFlow is being promoted by the Open Networking Foundation formed by six companies that own and operate some of the largest networks in the world — Deutsche Telekom, Facebook, Google, Microsoft, Verizon and Yahoo! — as well as 17 member companies, including IBM and other major equipment vendors, networking and virtualization software suppliers and chip technology providers. We are proud to be one of the inaugural members of the Open Networking Foundation, which again demonstrates IBM’s long-standing commitment to open industry standards, which are essential for the Smarter Computing that powers a Smarter Planet.
At Interop Las Vegas, IBM System Networking and NEC will show in NEC’s booth #904, for the first time publically, a proof-of-concept demo of an OpenFlow-enabled high-performance IBM BNT RackSwitch G8264 working in concert with an NEC OpenFlow controller. IBM will also participate in the InteropNet OpenFlow Lab, which will educate attendees on the principles, functions and features of OpenFlow.
IBM is also actively involving in open standards to better equip enterprise data centers for server virtualization. Edge Virtual Bridging (EVB) technologies, including the new Virtual Ethernet Port Aggregator (VEPA) technologies now being defined by the IEEE 802.1Qbg working group can make it easier for businesses to achieve server-network edge virtualization in the data center.
IBM’s Renato Recio is a driving force behind the 802.1Qbg standards, yet another of Renato’s numerous contributions to important industry standards that have dramatically simplified the way virtualized data centers operate. Renato was recently named an IBM Fellow, IBM’s most prestigious technical honor, recognizing his outstanding career of innovation that has so far resulted in 97 patents issued and 80 patents pending. I am proud to have Renato as a key member of our IBM System Networking team.
IBM System Networking’s switch-resident VMready with Virtual Vision is a unique EVB-ready solution that enables the network to be Virtual Machine aware, so that a data center network environment be configured, managed and secured for thousands of mobile Virtual Machines.
If you are attending Interop Las Vegas, May 8-12, please visit IBM in booth #621 to find out more about our leadership in collaborative innovations, such as OpenFlow and Edge Virtual Bridging.
Networks Go Flat So Data Can Go Faster
Today’s application requirements place new demands on the data center network fabric to deliver non-stop, ultra-low latency traffic flows. This traffic is increasingly “east-west” in nature to enable server-to-server communications versus the “north-south” traffic that characterizes conventional client/server and Web-based application environments. In fact, today, as much as 80 to 85 per cent of the traffic in cloud and virtualized server infrastructures – moves from server to server.
Deployment of the network fabric to serve these “east-west” requirements ideally begins at the edge of the network, close to servers, applications, users and innovation. So, today’s evolving data center architectures start with building out the fabric at the edge, connecting multiple blade servers, racks or systems, and eventually connecting multiple data centers. This is all about allowing clients to take advantage of fabric technology without expecting them to undertake an expensive and resource intensive forklift upgrade. That’s why starting at the edge makes a lot of sense.
For the intensive machine-to-machine communications required for server virtualization, cloud computing and high performance computing applications such as high-frequency trading, latency should be as close as you can possibly get to zero. There is the basic latency associated with the speed of light and the transmission medium, so as close as you can possibly get to zero defines the ideal. People in the industry have used the term the race to zero latency as a way of describing this insatiable quest for driving latency out of the system.
Clients are deploying IT infrastructure on an unprecedented scale. For example, data centers that were deploying five to six thousand servers on an annualized basis are now are routinely looking at deploying an order of magnitude more server infrastructure. So the scale at which these clients are deploying server infrastructure and storage capacity is huge. And if the network fabric that is connecting these servers and storage devices does not scale along with servers and storage, then you’ve got a problem.
While on one hand, clients are staring down the barrel of deploying an order of magnitude more servers and storage infrastructure than they ever have before, on the other hand they’re also concerned about ineffectiveness as it relates to utilization of their IT infrastructure. A big focus for enterprises is to increase the utilization of their IT infrastructure. The airline industry calls this yield management. The hospitality industry calls this occupancy rates. In the context of the data center, this is all about maximizing the utilization of IT infrastructure.
We in IBM System Networking agree with The Register’s Timothy Prickett-Morgan who in an insightful article “No More Tiers for Flat Networks,” writes, “For companies that need network traffic to move more efficiently at higher bandwidth and with lower latencies, then a leaf-spine network that has a flatter architecture, or perhaps a fat tree network inspired by supercomputers or a Clos network inspired by telecommunications, might be just the ticket.”
The definition of an ideal fabric is one that does not require clients to jump through hoops, change out servers or add unnecessary complexity. A good fabric should be one that can provide connectivity for the client’s existing infrastructure. That’s why IBM is a big proponent of standards, because we fundamentally believe that standards can bring not only the most innovative solutions to marketplace, but also deliver solutions that don’t lock clients into a particular type of technology or a specific vendor.
The data center network is going through a major transformation to support server virtualization and cloud computing, convergence of data storage, server-to-server traffic and new high-performance applications. To address these needs, the data center network fabric, the system network architecture that interconnects server devices and storage devices in a data center environment, has become a critical lynchpin of data center architecture. And the move to flat networks is enabling the fabric to help data go faster. And for today’s requirements, faster is most definitely better.
New IBM System Networking Innovations Address CIOs’ Pain Points
As I meet with CIOs around the world, the key issues they are working to address are invariably similar. Across the board, they are seeking to implement data center solutions that enable scalability, density, simplicity of management and provisioning, IT efficiency and virtualization, enhanced security, big data and analytics and low lifecycle costs. IBM System Networking is introducing a range of innovative new products that address these requirements. With these new offerings, CIOs can:
Scale SAN fabrics for private clouds. IBM System Networking is delivering new, high-performance 16Gbps Fibre Channel storage area networking (SAN) backbone, switch and management solutions designed to help enterprise clients migrate smoothly to private cloud architectures through faster data transfers, fewer links needed to accomplish the same task and fewer devices managed overall, with energy consumption seven times more efficient than competitive solutions.
Increase density and flatten networks. The new IBM Rackswitch G8316 is IBM System Networking’s new 40 Gigabit Ethernet aggregation switch with 16x 40G ports to aggregate multiple racks of servers or 64x 10G ports using breakout cables and a powerful control plane for higher performance. The IBM RackSwitch G8316 supports fat-tree networks for higher bandwidth and lower latency, provides an affordable alternative to expensive core switches, delivers predictable low latency through its single-chip design, and enables an end-to-end standards-based integrated solution using IBM switches and servers for easy management and provisioning.
Simplify management and provisioning. A unified management software solution for mixed SAN and FCoE environments, IBM Network Advisor provides greater insight for end-to-end network management to help reduce costs and simplify operations. IBM Networking Element Manager (formerly BLADEHarmony Manager) now includes better integration with Tivoli, topology discovery and support for additional managed devices.
Increase IT efficiency and virtualization. The IBM Networking Operating System (formerly BLADEOS) is a best-in-class data center network operating system for IBM RackSwitch and BladeCenter switches that incorporates the innovative VMready capability with support for the IEEE’s Edge Virtual Bridging (EVB) 802.1Qbg standard to bring scalable virtualization awareness to the data center network. With VMready, clients can confidently deploy and manage production application workloads in private clouds consisting of hundreds of physical servers and thousands of virtual machines.
Enhance security. IBM iFlow Director is a purpose-built flow distribution system that enables high-performance and low-cost BladeCenter-based appliance solutions. IBM iFlow Director provides flow-based load balancing for security appliances, such as deep packet inspection devices, firewalls and IDS/IPS systems, anti-virus and anti-spyware appliances and lawful intercept devices as well as wireless Internet mobile gateway appliances.
Power big data and analytics. IBM System Networking’s top-of-rack, blade and SAN smarter networking solutions deliver the massive amounts of bandwidth required for today’s big data systems.
Reduce lifecycle costs by increasing network control and flexibility. Using the OpenFlow-based IBM RackSwitch G8264 10/40 Gigabit Ethernet switch, user-controlled virtual networks are easy to create, optimize performance dynamically and minimize complexity. OpenFlow is a network protocol that has been developed over the past six years at Stanford University. The new OpenFlow solution enables the network administrator to easily configure and manage virtual networks that control traffic on a per-flow basis.
CIOs can count on IBM for smarter data center solutions that address their most pressing pain points. You can learn more about our exciting new IBM System Networking innovations by visiting ibm.com/networking.
OpenFlow Gets Traction at SC11
Seattle, SC11 was the scene for the unveiling of IBM's new Blue Gene/Q supercomputer project to solve the most challenging problems facing engineers and scientists, such as predicting the path of hurricanes, analyzing the ocean floor to discover oil, simulating nuclear weapons performance and decoding gene sequences. SC11 attendees could challenge IBM’s Watson supercomputer in a game of Jeopardy and see the most innovative network research projects in “programming the network” using OpenFlow.
OpenFlow allows the implementation of software-defined networking to enable significant innovation in High Performance Computing, which is highly reliant on network infrastructure. At SC11, the SCInet Research Sandbox (SRS)
gave researchers access to over 100 Gigabits per second of capacity to demonstrate the promise of OpenFlow on a software-programmable testbed network running on the SCinet infrastructure. The SC11 SRS will feature for the first time a 10 Gigabit Ethernet, multi-vendor OpenFlow network testbed to provide OpenFlow capabilities for wide area networking. I am pleased to report that our OpenFlow-enabled IBM RackSwitch G8264
will play a major role in this landmark demonstration of OpenFlow capabilities.
As part of the SRS, Indiana University (IU) deployed a 100 Gigabit Ethernet network for its high-speed Lustre WAN between the IU data center in Indianapolis and the convention center in Seattle utilizing OpenFlow technology for path selection and OpenFlow-based IBM RackSwitch G8264s. IU’s Global Research Network Operations Center at Indiana University has extensive network expertise and is the home of the recently announced Network Development and Deployment Initiative based on OpenFlow. IU deployed two Lustre filesystems at the ends of a 100Gb network connecting Bloomington, Indiana and the SC11 show floor. The IU demo executed real-world scientific applications that will saturate this 100Gb link. At the saturation point, application traffic will be dynamically routed over an alternative network using OpenFlow, to tune traffic based on need, priority and capacity.
The SRS utilized IBM RackSwitch G8264s in multiple other OpenFlow demos including load balancing, dynamic circuits, 802.1ag fault management, and the Steroid OpenFlow Service (SOS) for end-to-end application throughput over long-range networks.
For a complete list of OpenFlow demos in SCInet, visit: SC11/SRS.org
Also at SC11, IBM System Networking showcased our new smarter networking solutions including products our IBM System Networking RackSwitch G8316, a 40 Gigabit Ethernet (GbE) aggregation switch optimized for High Performance Computing and other applications requiring high bandwidth and low latency. The IBM System Storage SAN768B-2 and SAN384B-2 fabric backbones are among the industry's newest Fibre Channel switching infrastructure, providing reliable, scalable, high-performance foundations for private cloud storage and highly virtualized environments. The IBM System Storage SAN48B-5 SAN switch is designed to meet the demands of hyper-scale, private cloud storage environments by delivering 16 Gbps Fibre Channel technology and capabilities that support highly virtualized environments.
SC11 was a singular gathering of the High Performance Computing community, and it was a privilege for the IBM System Networking team to have the opportunity to meet with so many innovators in the engineering and science community.
OpenFlow – Next-Generation Networking for a Smarter Planet
We are seeing dramatic shifts as our planet becomes smarter. Every aspect of life is benefiting from the instrumentation, interconnection and the infusion of intelligence into the systems of the world. Networking is no exception, and the emerging OpenFlow specification promises a greater level of intelligence in Ethernet networks through a new approach called Software-Defined Networking (SDN).
OpenFlow is a network protocol that has been developed over the past six years at Stanford University. It was initially created for researchers and universities as a tool to allow experimentation with new protocols and is now showing great promise for today’s highly virtualized enterprise and cloud computing networks.
IBM is an industry leader in network virtualization technology and cloud computing through IBM System Networking's VMready product and is actively participating in the standardization of virtualization in networks through our leadership contribution to the IEEE 802.1Qbg standard. IBM is actively involved in OpenFlow technology.
OpenFlow is being promoted by the Open Networking Foundation formed by six companies that own and operate some of the largest networks in the world — Deutsche Telekom, Facebook, Google, Microsoft, Verizon and Yahoo! — as well as 17 member companies, including IBM and other major equipment vendors, networking and virtualization software suppliers and chip technology providers. OpenFlow is a disruptive technology that enables customers to choose their networking hardware and software separately in order to design best-of-breed data center solutions.
There are many applications for OpenFlow in modern networks. For example:
- A network administrator could create on-demand “express lanes” for voice and data traffic that are time-sensitive.
- Software could also be used to combine several fiber optic links into a larger virtual pipe for temporarily handing a particularly heavy flow of traffic, and then have the channels automatically separate again when the data rush is over.
- Service providers could use OpenFlow to help build a Smarter Planet by offering remote services such as home security or energy management.
- In cloud computing environments, OpenFlow improves scalability and enables resources to be shared efficiently among different services in response to the number of users.
OpenFlow revolutionizes the monolithic model of conventional network devices by managing the flow tables on network switches, routers and access points independently of the device’s software. It removes much of the network decision making, often called the “control plane” from network devices into an external controller that can be implemented using standard server technology.
At Interop Las Vegas, May 8-12, 2011, IBM System Networking and NEC will be showing, for the first time publically, a proof-of-concept demo of an OpenFlow-enabled IBM BNT RackSwitch G8264 working in concert with an NEC OpenFlow controller. IBM will also participate in the InteropNet OpenFlow Lab, which will educate attendees on the principles, functions and features of OpenFlow. The lab will demonstrate OpenFlow in different scenarios, including loop free networking, dynamic load balancing across multiple links and quality of service for VoIP.
We are proud to be one of the inaugural members of the Open Networking Foundation, which again demonstrates IBM’s long-standing commitment to making a difference for customers with open industry standards, which are essential for the Smarter Computing that powers a Smarter Planet.
System Networking and Data Center Efficiency Redefined
The value proposition for IBM System Networking is to provide the essential network connectivity solutions under the IBM brand to connect servers to servers, servers to storage and storage to storage. IBM System Networking offers a compelling alternative for customers seeking more efficient data centers with the greatest business value and lowest total cost of ownership for their data center networks.
System networking plays a critical role in customers’ server and storage buying decisions. Today, many customers are deploying IT infrastructure on an unprecedented scale – data centers are expanding from 5,000 to 50,000 and 100,000 servers. For such scale-out architectures, the system network plays a critical role. Consider density; if it requires three data centers to house 5,000 servers, how many data centers will it take to house 50,000 servers? The answer better not be 30! So, the system network must deliver the high-density networking required to support highly consolidated and massively virtualized data center infrastructures.
If you are a CIO undertaking an order of magnitude increase in infrastructure, you want to increase utilization through virtualization, which requires the system network to be virtualization aware. And of course, the system network is vital to the security of this infrastructure.
If you have an order of magnitude increase in infrastructure, much of the functionality required to solve deployment and management issues can reside on data center switches implemented within the system network.
As companies take their businesses online, rapid and accurate business intelligence becomes ever more critical, which requires the system network for fast transport of information to and from analytic engines.
If you are employing an order of magnitude more infrastructure, total cost of ownership is important, and companies spend 15 to 20% of their investment in infrastructure on the network.
To address CIO’s key strategic issues of scaling, density, utilization/virtualization, security, data management and cost ownership, system networking is the common thread. IBM is an incredibly reputable server and storage vendor, and you can see the critical role the network plays.
The Cloud-Ready System Network
Today’s competitive business needs are driving the development of cloud-based data centers that are more cost-effective, agile, and scalable than ever before. Cloud computing places higher demands on the system network in areas such as speed, flexibility, virtualization, cost-effective operation and scalability. To meet the technical and business requirements of cloud computing, the networking layer of a cloud must offer high bandwidth and low latency, converged communications and storage, agile networks for virtual machine mobility, massive scalability and manageability and advanced energy efficiency.
The essential attributes of a cloud network include terabit scalability, predictable low latency, non-blocking throughput and high-speed interconnects using 1/10GbE and the emerging 40/100GbE. For example, the new IBM BNT RackSwitch G8264 delivers throughput of up 1.28 Terabits per second and with its single-chip architecture, low latency is delivered across all port combinations. To bring even more bandwidth to the cloud, the RackSwitch G8264 is among the industry’s first top-of-rack switches with 40GbE interconnects.
One of the main advantages of cloud computing is on-demand access to resources, and virtualization plays a key role in providing those resources. IBM System Networking's VMready network virtualization software enables cloud computing infrastructures with mobile, active virtual machines. Cloud computing users can gain even greater advantages from mobile virtual machines when they can be moved securely and with predictable performance not only within a cloud, but over greater distances to connect multiple clouds. Movement between clouds enables applications such as disaster recovery and data replication.
Today, cloud computing environments are deploying IT infrastructure on an unprecedented scale – data centers are expanding from 5,000 to 50,000 and 100,000 servers. For such scale-out architectures, the system network must deliver the high-density networking required to support highly consolidated and massively virtualized data center infrastructures. Today’s “flat” network topologies enabled through standards such as TRILL (Transparent Interconnection of Lots of Links) are key to this scalability.
Power and cooling are some of the biggest expenses of operating a cloud data center. A common estimate for data center cooling and distribution costs is two watts for every single watt consumed by data center equipment and networking gear is 10 to 15% of the entire infrastructure —so choosing the most energy-efficient network components is essential. As clouds grow to thousands of servers and beyond, per-component power savings are magnified into hundreds of thousands of kilowatts.
From a business perspective, the system network architecture for today’s cloud computing applications must support incremental deployment that does not require razing an existing facility and building a new one. And as new pieces of the cloud are fitted into place, IT managers don’t want to be locked into a single-vendor implementation for any aspect of the solution. They want the freedom to select best-in-class hardware and software components, and they want to deploy new capacity as rapidly as possible. For example, IBM has extended Tivoli's virtualization management capabilities so that it can provision and deploy hundreds or thousands of virtual machines an hour for large-scale enterprise cloud implementations.
What Makes A Good Network Fabric?
As I wrote in my earlier blog post, the data center network is going through a major transformation to support server virtualization and cloud computing, convergence of data storage, application-to-application traffic and new high-performance applications. To address these needs, the data center network fabric, the system network architecture that interconnects server devices and storage devices in a data center environment, has become a critical lynchpin of data center architecture. I am frequently asked the question, “What makes a good network fabric?” The term “fabric” has nearly as many definitions and permutations as “cloud,” so I would like to suggest the top 10 attributes that a data center network fabric should possess.
- Open and Interoperable – Modern data center networking is best accomplished when it is standards-based, when multiple vendors equipment can co-exist and interoperate, and clients can choose between multiple vendors’ wares without paying a pricing penalty or needing to rip-and-replace to meet growth needs and implement next-generation approaches. Standards-based Ethernet is a fabric essential.
- High Speed – To meet the performance needs of big data, cloud computing and workload-optimized systems, data centers are increasingly implementing 10 Gigabit Ethernet on the server and in the access and aggregation layers, which is driving interest in 40 Gigabit Ethernet with 100 Gigabit Ethernet on the horizon.
- Low Latency - Applications such as high-frequency trading require the lowest possible latency. The race to zero latency is enabled by ultra-low-latency switches that deterministic with the same connection speeds across every port combination.
- Loss-less – To meet the needs of machine-to-machine applications and converged data and storage networking, Ethernet networks must be loss-less. Network equipment supports the Data Center Bridging (DCB) standards to ensure loss-less operation.
- Flat – Clos and fat tree network designs are becoming increasingly prevalent for the flow-based, non-blocking, shortest path network fabrics required in highly virtualized and cloud data centers and for converged data and storage traffic. With standards such as Transparent Interconnection of Lots of Links (TRILL) and Shortest Path Bridging (SPB) on the horizon and other alternatives for flat networks requiring proprietary implementations, network architects favor existing approaches to meshed networking such as Virtual Link Aggregation (vLAG) to maximize network efficiency, reduce congestion and address Spanning Tree limitations by creating active/active Layer 2 network paths for load balancing and redundancy.
- Virtualization Aware – To ensure that network security, performance and access policies move dynamically when live virtual machines migrate from server to server anywhere in the virtualized data center or private cloud, the Edge Virtual Bridging (EVB) 802.1Qbg standard specifies the interaction between virtual switching environments in a hypervisor and the first layer of the physical switching infrastructure. The EVB 802.1Qbg standard addresses the conventional vSwitch’s lack of network management, monitoring and security.
- Scalable – Linearly scalable is a requisite of network fabric design, which can be achieved by non-blocking, non-oversubscribed topologies implemented using top-of-rack switches.
- User Control – The emerging OpenFlow specification enables network infrastructure providers to deliver open virtual networking systems that that are easy for users to control, optimize performance dynamically and minimize complexity. OpenFlow-based networks enable the network administrator to easily configure and manage virtual networks that control traffic on a per-flow basis. The administrator can easily program an OpenFlow controller to create or delete multiple independent virtual networks and related policies that span the system network without having to deal with the complexities of the underlying physical network and protocol details.
- Unified Management – Network devices that make up the fabric should be managed, configured and provisioned as if they were a single logical device, including the ability to can track virtual machines by switch or IP address, and pre-provision network characteristics for VMs.
- Low Lifecycle Costs – And finally, per-switch costs should be affordable so networks can scale on an incremental basis, the need for expensive chassis switches should be minimized, and low power requirements and cooling efficiencies should enable low energy costs particularly for massive networks fabrics that interconnect thousands of server and storage systems.
We have designed IBM System Networking’s Unified Fabric Architecture, our fast, smart, interoperable and proven data center Ethernet network fabric, to deliver across all 10 of these attributes. Unified Fabric Architecture along with the ability to replicate pre-provisioned racks and integrated systems, enables clients that have already invested billions of dollars in their data centers to take advantage of the best innovations in the industry and achieve the lowest possible cost of ownership for their IT infrastructure.