Sunday, November 14, 2010

Cavium Scales its OCTEON II Processor Line

Cavium Networks is extending its family of OCTEON II processors with four new processor lines that integrate one to ten cnMIPS64 cores. The additions expand both low and mid range options in a fully software compatible manner with other members of the OCTEON II and OCTEON Plus family, enabling enterprise, data center and service provider equipment including routers, switches, appliances, 3G/4G wireless base stations, RNCs, xGSNs, evolved packet core, services gateways, DPI equipment, storage switches and intelligent server adapters. In addition, Cavium is introducing four new, unique technologies in Octeon II to achieve lower power, deep packet inspection, virtualization and anti-counterfeiting.

The OCTEON II family combines multiple cnMIPS64 R2 cores along with L2-L7 application and security acceleration engines, virtualization features and 100Mbps - 100Gbps of connectivity. The custom designed cnMIPS™ cores implement an enhanced MIPS64 R2 instruction-set and have been architected to deliver superior performance per area along with dedicated compute and cache resources essential for deterministic, low-latency performance.

The newly introduced OCTEON II processors and target applications include:

CN60XX (1 - 2 Cores): WLAN Access points (AP), Entry-level VPN/UTM/Firewalls, Low-cost Control Plane applications, Satellite and Broadband Gateways.

CN61XX and CN62XX (2 -- 4 Cores): Enterprise Switches & Routers with Integrated Data and Control, High-performance Control Plane, UTM & Security Appliances, Printers, Small to Medium Business (SMB) and Low to Mid-range Enterprise class NAS/iSCSI appliances.

CN66XX (6 -- 10 cores): WAN Optimization, Enterprise, Data Center and Storage applications requiring 2x10GbE performance and connectivity and High-performance 3G/4G Base Stations.

These new processors incorporate a range of hardware accelerators including L2-L7 application and security acceleration engines, TCP offload, load balancing, compression and support for virtualization.

The four new technologies in OCTEON II include:

1. Power Optimizer that dynamically adjusts power in real time depending upon the application-level processing requirement, without the need to adjust CPU voltage or clock frequency. Now application developers can specify the maximum power that a program is allowed to consume on a per core basis.

2. "Hyper Finite Automata" (HFA) provides the industry's best Deep Packet Inspection solution by combining both DFA and NFA architectures into one cohesive solution. Performance is unchanged regardless of the number of rules and number of flows that are processed.

3. "EmVisor" provides unprecedented hardware support for a complete SoC virtualization. Now OCTEON II CPUs can act like a series of Virtual SoCs, each with their own virtual memory, Ethernet ports, PCIe Bus and EmVisor also provides a virtual switch between Virtual SoCs. This also enables in-field software upgrades in customer equipment.

4. "Authentik" is a technology that allows the whole multi-core processing chip to be locked; so that OEMs that entrust third parties to assemble their systems can be assured that no rogue or counterfeit copies of their systems are also being manufactured.

F5 Positions its Service Delivery Networking for Mobile Data Tsunami

F5 Networks, which specializes in layer 7 traffic management solutions, is positioning itself as a strategic control point for the growing tsunami of mobile data traffic. Specifically, F5's Service Delivery Networking (SDN) leverages its BIG-IP appliance to offer a set of integrated solutions for service and subscriber management -- such as intelligent traffic steering, large-scale IPv4 to IPv6 translation, secure DNS traffic management, traffic optimization and acceleration, and policy enforcement.

F5's goal is to enable a Full Proxy Services Architecture that uses the high performance packet processing capabilities of its BIG-IP for subscriber services ranging from DDOS protection, SSL VPNs, geolocation, rate shaping, fast caching, compression, TCP optimization, etc. When positioned behind a GGSN, F5's BIG-IP VIPRION could deliver intelligent traffic steering for policy-based, premium mobile services. It could also be used to optimize AAA infrastructure and specifically to address the problem of server overloading due to smartphone requests. In more advanced applications, F5's SDN could be used for dynamic, personalized ad insertion in mobile data sessions.

F5 has a number of ecosystem partners in the mobile arena, including Ericsson, NSN, Bridgewater Systems, bytemobile, Openwave and others.

Cisco Debuts Carrier Packet Transport with MPLS-TP

Cisco introduced its Carrier Packet Transport (CPT) System -- a metro access aggregation platform that aims to integrate the capabilities of Ethernet + TDM switching, transponders and ROADMs in order to deliver packet network efficiencies in the transport network. Essentially, CPT unifies the packet and transport layers in one compact, carrier-class platform. It integrates DWDM, OTN, Ethernet, and MPLS-TP functionality, giving service providers a P-OTS platform for next-generation transport networks.

Cisco's CPT is the first to support standards-based MPLS transport profile (MPLS-TP), which interoperates seamlessly with MPLS offered on routing products like the Cisco ASR 9000 Series Aggregation Services Routers and Cisco CRS-3 Carrier Routing System.

The platform is being introduced in a six-rack unit (Cisco CPT 600) and a two-rack unit (Cisco CPT 200) model, along with a satellite device (Cisco CPT 50). The modular CPT 600 P-OTS platform scales up to 352 Gigabit Ethernet ports, while the CPT 200 offers up to 176 Gigabit Ethernet ports.

Cisco estimates its CPT system will achieve a 60 percent reduction in rack space by combining multiple interface layers into one. This directly translates into a 65 percent power savings.

The Cisco CPT system is currently in field trials with several service providers and is scheduled to be available in the first quarter of 2011.

Cisco also introduced a multivendor, Premier Integrated Network Management Experience (PRIME) software package that manages MPLS and MPLS-TP end-to-end, dramatically simplifying operations and consequently reducing operational costs. It is available as a graphical user interface on a single domain to configure, provision and manage optical and MPLS connections. It is also offered as an autonomous model that enables separate management of MPLS from that of wavelengths.

NTIA Seeks to Free up 115 MHz of Spectrum

The National Telecommunications and Information Administration (NTIA) is recommending that 115 MHz of spectrum be reallocated for wireless broadband service within the next five years.

As part of the announcement, NTIA released two complementary reports detailing the effort to nearly double commercial wireless spectrum: a Ten-Year Plan and Timetable, as well as a Fast Track Evaluation identifying the 115 megahertz of spectrum to be made available within five years.

In the Fast Track Evaluation report, NTIA examined four spectrum bands for potential reallocation within five years: 1) 1675-1710 MHz, 2) 1755-1780 MHz, 3) 3500-3650 MHz, and 4) 4200-4220 MHz and 4380-4400 MHz. The report recommends that various portions of these bands totaling 115 megahertz be made available for wireless broadband use within five years, contingent upon the allocation of resources for necessary reallocation activities.

Specifically, NTIA recommends reallocating 1695-1710 MHz, currently used for dissemination of severe weather information and alerts via satellites operated by Commerce's National Oceanic and Atmospheric Administration, and 3550-3650 MHz, which would be shared with Department of Defense radar systems mainly on ships. NTIA recommends some geographic limitations on the availability of these bands to prevent harmful interference to government facilities in the 1695-1710 MHz band and to the proposed commercial services in the 3550-3650 MHz band.

Of the 2,200 megahertz of candidate spectrum that the Ten-Year Plan and Timetable identify, 28 percent is allocated exclusively for Federal use at present, 35 percent is allocated exclusively for commercial use, and 37 percent is shared by Federal and commercial users. The 2,200 megahertz includes 280 megahertz of commercial spectrum that the FCC recommended in its National Broadband Plan be made available for mobile broadband use within five years.

"We have identified the first steps for unlocking a significant amount of the spectrum needed to reach the President's 10-year goal," Department of Commerce Secretary Gary Locke said. "In implementing this initiative, the Obama administration is seeking to catalyze innovation and private sector investment. This complex undertaking is critical to enhance America's economic competitiveness at home and abroad and promises to improve our quality of life."http://www.ntia.doc.gov

AT&T to Offer Samsung Galaxy Tab Pay-As-You-Go Monthly Data Plans

AT&T announced two pay-as-you-go data plan options for the Samsung Galaxy Tab: a $14.99 pay-as-you-go plan for 30 days with up to 250 MB of usage and a $25 pay-as-you-go plan for 30 days with up to 2 GB of usage. If customers use all the data included in their plan before 30 days, they can purchase another pay-as-you go plan. For additional convenience, these plans can automatically renew every 30 days, when customers pay for these plans with a credit card or debit card.

AT&T data plan subscribers also get access to AT&T's entire national Wi-Fi network at no additional charge.

AT&T will also sell the Galaxy tab more than 2,200 AT&T company owned stores and online beginning November 21.

Clearwire and Sprint Launch 4G in Sacramento

Clearwire has launched its WiMAX network in Sacramento, California. Regionally, service extends out to Davis to the West, to Elk Grove in the South, to Roseville/Folsom out to the East and Natomas to the North. Sprint also announced 4G commercial service availability in Sacramento.

Facebook Messaging: SMS + IM + Email + FB

Facebook unveiled its new messaging platform aimed at unifying SMS, instant messaging, email and Facebook updates. The new service, which will be rolled out to the company's 500 million subscribers, will let users send messages from within Facebook to any of the outbound channels. Incoming messages from any channels can also be seen from within Facebook and users will have the option of sharing with friends or keeping messages sorted by conversation with the single person sending you the messages.

EMC to Acquire Isilon for Big Data Storage

EMC agreed to acquire Isilon Systems, which develops "scale-out" network attached storage (NAS), for $33.85 per share in cash -- which equates to approximately $2.25 billion, net of Isilon's existing cash balance.

Isilon, which is based in Seattle, offers a highly scalable, low-cost storage infrastructure solution for managing "Big Data", such as the storage needs for new generation applications in the life sciences (e.g. gene sequencing), media and entertainment (e.g. online streaming), and oil and gas (e.g. seismic interpretation). Isilon's scale-out NAS systems are designed to begin small and scale quickly and non-disruptively up to 10 petabytes in size, with extremely high levels of performance and availability.

EMC said its Atmos object storage provides the perfect complement to Isilon for massive globally distributed environments and object access to data for usages like Web 2.0 applications.

Joe Tucci, Chairman and CEO, EMC Corporation, said, "The unmistakable waves of cloud computing and 'Big Data' are upon us. Customers are looking for new ways to store, protect, secure and add intelligence to the vast amounts of information they will accumulate over the next decade. EMC, in combination with Isilon, sits at the intersection of these trends with leading products, solutions and services to help customers get the absolute most out of what cloud computing has to offer."

Juniper's T4000 Driven by 45nm Trio Silicon

James E. Carroll

Juniper Networks unveiled its next generation T4000 core router featuring 16-slots operating at 240 Gbps and capable of supporting high-densities of 10G connections, as well as 40 and 100GE interfaces. The new platform, which continues Juniper's Junos operating system and retains existing interfaces from its T640 and T1600 core routers, addresses the scalability challenges of service provider backbone and major Internet peering points. The T4000 packs four terabits of line rate throughput in a half rack chassis. It is driven by the company's second generation ASICs, which are implemented using 45nm technology and capable of delivering 250 million packet-per-second forwarding performance per slot -- this is double the forwarding capacity of Cisco's latest CRS-3 router, according to the company.

Juniper's Trio silicon was first unveiled in October 2009 for use in its MX series edge routers. The first generation was implemented in 60nm technology and optimized for highly granular packet processing. The second generation features the smaller and more energy-efficient 45nm technology and is optimized for the high-throughput requirements of core routing. By comparison, the T1600 uses 90nm silicon.

Key design criteria for the T4000 are investment protection and evolution of existing routing, said Luc Ceuppens, vice president of product marketing, Juniper Networks. Existing T640 and T1600 routers are in-service upgradeable to T4000 status by swapping out their power entry modules, switch interface boards, fans and adding a new face panel. The upgrade process can be completed in about 90 minutes without bringing down the system -- "like changing a car's engine while the car continues in motion." The system is also designed with High Availability hardware and component level redundancy for routing engines, CBs and SIBs.

A fully loaded T4000 system can support up to 192 ports of 10GbE, or 16 ports of 100GbE interfaces.

Like the previous generations of the T-series, the T4000 has multichassis capability designed into the architecture. The multi-chassis support, which will come after the initial release, will tie together up to four T4000s using the company's TX Matrix Plus.

Juniper said its T4000 also breaks new ground in power efficiency in terms of watt per bit processed -- 2.7 watts of power per Gigabit of processed traffic -- which is more than three times as efficient as the T1600.

The T4000 is being shown at the Supercomputing 2010 show in New Orleans. Commercial availability is expected in the second half of 2011.

Currently, there are about 6,000 Juniper T-series routers in deployment around the world.
  • Juniper Networks first unveiled its T1600 core router in June 2007.

Brocade Delivers Virtual Cluster Switching

Brocade introduced the first two products in a new line of 10 GigE data center switched specifically build for highly virtualized and cloud-optimized data centers. The switches eliminate the need for Spanning Tree Protocol (STP), collapsing the access and aggregation networking layers to create a flat, multipath, deterministic network. Brocade believes this approach is ideal for virtualized environments because it provides virtual machines (VMs) a greater sphere of mobility, increases network utilization, creates more resilient networks and simplifies the management of data center networks.

Brocade is initially introducing one- or two-rack-unit (1U or 2U) switches based on its sixth generation Ethernet switching ASIC and virtual cluster switching (VCS) technology. The switches have the ability to scale from 16 to 60 ports with pay-as-you-grow Ports on Demand (POD) licensing. The switches provide full 10 gigabits per second (Gbps) wire-speed performance from any port to any port, with latency of 600 nanoseconds (ns). The fabric uses low-latency, lossless, 10 Gbps Data Center Bridging (DCB) technology and is capable of running all types of data and storage traffic --including traditional IP, iSCSI, CIFS, NFS and Fibre Channel over Ethernet (FCoE).

"Today's infrastructure is not sufficient to handle new applications, and not flexible enough for a world where applications are mobile. It's clear that data center networks need to be upgraded to increase performance, reduce latency and eliminate downtime. They also need to be designed specifically to support highly virtualized and cloud-optimized data centers," said Dave Stevens, Brocade CTO. "We are delivering upon our Brocade One unified network strategy and vision by dramatically simplifying networks and increasing scalability without adding complexity. Today, the Brocade VDX 6720 switches, featuring Brocade VCS technology, deliver true cloud-optimized networking for the first time."