Tuesday, January 22, 2013
Virtualization: Virtual, Cloud Workload Management: 10 Clues It's Time for a System Overhaul
With the continued evolution of virtualized IT systems, the computer industry has made great strides in IT service delivery in the data center. Yet, the same claim can't be made in the management realm because most infrastructure and operations teams tend to cling to traditional or antiquated approaches to workload management that cannot scale to meet the increasing ebb-and-flow demands of virtual data centers. The impact? Stalled virtualization initiatives, higher costs and overburdened operational teams, to name a few. To keep pace with virtualization's evolution, most enterprises at some point will need to implement a new approach to IT management. With the ever-changing environment that is virtualization, enterprises will need to employ a workload management package that enables automation of configurations and workload placement to ensure performance in the most efficient way. The result will be reduced risk, maximized infrastructure investment as well as the elimination of manual and time-consuming operator intervention, which helps big-time on the bottom line. Our main resource for this slideshow is Shmuel Kliger, CTO and founder of VMTurbo, based in Burlington, Mass., which focuses on tools for preventing cloud system problems. Here are 10 tips that show you need an alternative to your current virtual and cloud workload management system.
Cisco, Arista Unveil Low-Latency Ethernet Networking Switches
The networking competitors are rolling out switches that are aimed at such businesses as high-frequency financial trading, where speed is crucial.
Cisco Systems and Arista Networks are rolling out new low-latency networking switches that are aimed at such industries as high-frequency financial trading, where a premium is put on moving data between servers as fast as possible.
The networking competitors are looking to differentiate themselves in a growing part of the market that is being driven by demands for greater networking performance to handle workloads found in such environments as high-performance computing (HPC) and big data.
With the launch of their Nexus 3548 10 Gigabit Ethernet switch Sept. 19, Cisco officials also are introducing the company's new Algorithm Boost-or Algo Boost-silicon chip that they said will drive down the latency to as little as 190 nanoseconds, and will offer performance advantages over competing switches of as much as 60 percent. Current fast switches in the industry have a latency of more than 380ns, they said.
Click Here!
In addition, Cisco also offers a feature that the officials are calling the warp switch port analyzer (SPAN), which will get stock market data delivered to financial trading servers in as little as 50ns.
"This is the new frontier in low-latency networking," Rajan Panchanathan, director of product management, told eWEEK.
The financial services market is an important one to Cisco, according to Paul Perez, CTO of the vendor's Data Center Group, and one that Arista-which is run by several former Cisco executives, including President and CEO Jayshree Ullal-also is targeting with its technology.
"That business [means] billions of dollars in revenue to Cisco," Perez said in an interview with eWEEK. "Financial services is a key vertical for us."
It's also an industry that is known to be an early adopter of new technologies as it looks for products that will help increase the performance and lower the latency in their trading platforms, and those technologies tend to eventually find their way into enterprise infrastructures.
The Algo Boost ASIC technology helps drive not only the low latency in the 1U (1.75-inch) Nexus 3548 switch, but also makes it easier to monitor and manage the switch. The chip gives users greater visibility into how the switch is performing, and offers the Precision Time Protocol feature, which helps firms keep their infrastructures synchronized, bring correlate network events and keep in compliance with regulatory requirements.
Active Buffer Monitoring warns users of potential bottlenecks in the networks that could hurt performance, and Intelligent Traffic Mirroring offers filtering and time-stamping of captured traffic to give users an immediate and deeper view into the network.
The Nexus 3548 switch running in the warp mode can hit latency numbers as low as 190ns in environments that require small or medium Layer 2 or 3 scaling. For those with greater scalability needs, the switch can offer latency of 250ns. The switch also includes Cisco's Hitless Network Address Translation (NAT) feature, which enables traders to connect to trading venues without increasing the latency.
Cisco's Panchanathan said that while the Nexus 3548 will be the first to offer Algo Boost, the chip technology will find its way into future generations of all Nexus switches. The Nexus 3548 will start showing up in companies' data centers by the end of the year, according to Cisco officials.
The Cisco switch will compete with Juniper Networks' QFX3500 and Mellenox's SX 1036. It also will compete with switches from Arista, which on Sept. 19 also rolled out a new offering, the 7150 Series that officials said will offer support for software-defined networks (SDNs) and fit well with infrastructures that include network-wide virtualization environments.
Unlike Cisco's approach with its own Algo Boost ASIC technology, Arista will continue embracing off-the-shelf chips from Intel, which last year bought Ethernet chip maker Fulcrum. In a statement, Diane Bryant, vice president and general manager of Intel's Datacenter Group, said the combination of Arista's switch technology and its Arista EOS operating system with Intel's Ethernet Switch FM6000 chip offers "a common platform" upon which "SDN functionality can be achieved."
SDNs are aimed at decoupling the data center workloads from the physical network, and moving intelligence in the network-such as directing traffic to minimizing latency to security-from switches and routers to a software-based controller. SDNs reduce the need for expensive, complex switches and routers, and instead let enterprises buy simpler and less expensive networking equipment. In addition, networks become more dynamic and easier to manage.
"The vision is a data center where all of the IT resources are virtualized but have a level of fluidity to them so the resources can be migrated easily to whatever application or business service needs them based on business policy," Zeus Kerravala, principal analyst at ZK Research, said in a Sept. 18 blog post in Network World.
Arista has been among the top proponents of SDN, and company officials said the new 7150 Series switches push that idea forward, with their easy interoperability with SDN controllers. The switches make it easier for businesses to adopt network-wide virtualization, virtual machine mobility and network services, they said.
"Arista has combined a flexible forwarding data path with the Arista EOS (Extensible Operating System) to deliver breakthrough latency, power, density and advanced SDN features in a compact 1U form factor" Arista Chairman and Chief Development Officer Andy Bechtolsheim said in a statement. "The Arista 7150 is truly the first next-generation SDN switch for virtualized data centers."
The new switches offer up to 64 wire-speed 1 GbE and 10 GbE ports or sixteen 40 GbE ports, and support VXLAN tunnels at wire speed and the easy movement of workloads between physical and virtual machines. The 40 GbE ports offer latency of 350ns for Layer 2 and 3 forwarding. Features include NAT, IEEE 1588 Precision Time Protocol and congestion management.
The wire-speed NAT capabilities reduce the forwarding delay in HPC and financial trading environments by hundreds of microseconds, and latency analyzer features offer such capabilities as congestion monitoring and analysis to monitor application performance.
The 7150 Series switches can be ordered now and will start shipping in the fourth quarter, starting at $12,995.
Cisco Systems and Arista Networks are rolling out new low-latency networking switches that are aimed at such industries as high-frequency financial trading, where a premium is put on moving data between servers as fast as possible.
The networking competitors are looking to differentiate themselves in a growing part of the market that is being driven by demands for greater networking performance to handle workloads found in such environments as high-performance computing (HPC) and big data.
With the launch of their Nexus 3548 10 Gigabit Ethernet switch Sept. 19, Cisco officials also are introducing the company's new Algorithm Boost-or Algo Boost-silicon chip that they said will drive down the latency to as little as 190 nanoseconds, and will offer performance advantages over competing switches of as much as 60 percent. Current fast switches in the industry have a latency of more than 380ns, they said.
Click Here!
In addition, Cisco also offers a feature that the officials are calling the warp switch port analyzer (SPAN), which will get stock market data delivered to financial trading servers in as little as 50ns.
"This is the new frontier in low-latency networking," Rajan Panchanathan, director of product management, told eWEEK.
The financial services market is an important one to Cisco, according to Paul Perez, CTO of the vendor's Data Center Group, and one that Arista-which is run by several former Cisco executives, including President and CEO Jayshree Ullal-also is targeting with its technology.
"That business [means] billions of dollars in revenue to Cisco," Perez said in an interview with eWEEK. "Financial services is a key vertical for us."
It's also an industry that is known to be an early adopter of new technologies as it looks for products that will help increase the performance and lower the latency in their trading platforms, and those technologies tend to eventually find their way into enterprise infrastructures.
The Algo Boost ASIC technology helps drive not only the low latency in the 1U (1.75-inch) Nexus 3548 switch, but also makes it easier to monitor and manage the switch. The chip gives users greater visibility into how the switch is performing, and offers the Precision Time Protocol feature, which helps firms keep their infrastructures synchronized, bring correlate network events and keep in compliance with regulatory requirements.
Active Buffer Monitoring warns users of potential bottlenecks in the networks that could hurt performance, and Intelligent Traffic Mirroring offers filtering and time-stamping of captured traffic to give users an immediate and deeper view into the network.
The Nexus 3548 switch running in the warp mode can hit latency numbers as low as 190ns in environments that require small or medium Layer 2 or 3 scaling. For those with greater scalability needs, the switch can offer latency of 250ns. The switch also includes Cisco's Hitless Network Address Translation (NAT) feature, which enables traders to connect to trading venues without increasing the latency.
Cisco's Panchanathan said that while the Nexus 3548 will be the first to offer Algo Boost, the chip technology will find its way into future generations of all Nexus switches. The Nexus 3548 will start showing up in companies' data centers by the end of the year, according to Cisco officials.
The Cisco switch will compete with Juniper Networks' QFX3500 and Mellenox's SX 1036. It also will compete with switches from Arista, which on Sept. 19 also rolled out a new offering, the 7150 Series that officials said will offer support for software-defined networks (SDNs) and fit well with infrastructures that include network-wide virtualization environments.
Unlike Cisco's approach with its own Algo Boost ASIC technology, Arista will continue embracing off-the-shelf chips from Intel, which last year bought Ethernet chip maker Fulcrum. In a statement, Diane Bryant, vice president and general manager of Intel's Datacenter Group, said the combination of Arista's switch technology and its Arista EOS operating system with Intel's Ethernet Switch FM6000 chip offers "a common platform" upon which "SDN functionality can be achieved."
SDNs are aimed at decoupling the data center workloads from the physical network, and moving intelligence in the network-such as directing traffic to minimizing latency to security-from switches and routers to a software-based controller. SDNs reduce the need for expensive, complex switches and routers, and instead let enterprises buy simpler and less expensive networking equipment. In addition, networks become more dynamic and easier to manage.
"The vision is a data center where all of the IT resources are virtualized but have a level of fluidity to them so the resources can be migrated easily to whatever application or business service needs them based on business policy," Zeus Kerravala, principal analyst at ZK Research, said in a Sept. 18 blog post in Network World.
Arista has been among the top proponents of SDN, and company officials said the new 7150 Series switches push that idea forward, with their easy interoperability with SDN controllers. The switches make it easier for businesses to adopt network-wide virtualization, virtual machine mobility and network services, they said.
"Arista has combined a flexible forwarding data path with the Arista EOS (Extensible Operating System) to deliver breakthrough latency, power, density and advanced SDN features in a compact 1U form factor" Arista Chairman and Chief Development Officer Andy Bechtolsheim said in a statement. "The Arista 7150 is truly the first next-generation SDN switch for virtualized data centers."
The new switches offer up to 64 wire-speed 1 GbE and 10 GbE ports or sixteen 40 GbE ports, and support VXLAN tunnels at wire speed and the easy movement of workloads between physical and virtual machines. The 40 GbE ports offer latency of 350ns for Layer 2 and 3 forwarding. Features include NAT, IEEE 1588 Precision Time Protocol and congestion management.
The wire-speed NAT capabilities reduce the forwarding delay in HPC and financial trading environments by hundreds of microseconds, and latency analyzer features offer such capabilities as congestion monitoring and analysis to monitor application performance.
The 7150 Series switches can be ordered now and will start shipping in the fourth quarter, starting at $12,995.
Cisco Ending ADC Business, Ceding Market to F5, Citrix, Riverbed
Cisco is ending development of its ACE application delivery controller, exiting a market in which it has seen its share decline as F5 and Citrix have gained.
Cisco Systems, which for more than a year has been undergoing a transformation that includes shedding underperforming businesses, reportedly will no longer develop its ACE load-balancer technology, apparently ceding the market to rivals such as F5 Networks and Citrix Systems.
JMP Securities analyst Erik Suppiger wrote in a research note late last week that Cisco officials had been telling salespeople to stop promoting the Application Control Engine (ACE) 30 load-balancer module for new deployments, though the networking giant will continue to sell and service the product for existing customers.
"We believe Cisco's de-emphasis on selling the ACE for new deployments is a more recent development that reflects more of a shift away from the application delivery controller market," Suppiger wrote. "Cisco has been losing market share in the ADC market for several years, and we believe this will accelerate the company's share erosion."
Cisco Systems, which for more than a year has been undergoing a transformation that includes shedding underperforming businesses, reportedly will no longer develop its ACE load-balancer technology, apparently ceding the market to rivals such as F5 Networks and Citrix Systems.
JMP Securities analyst Erik Suppiger wrote in a research note late last week that Cisco officials had been telling salespeople to stop promoting the Application Control Engine (ACE) 30 load-balancer module for new deployments, though the networking giant will continue to sell and service the product for existing customers.
"We believe Cisco's de-emphasis on selling the ACE for new deployments is a more recent development that reflects more of a shift away from the application delivery controller market," Suppiger wrote. "Cisco has been losing market share in the ADC market for several years, and we believe this will accelerate the company's share erosion."
Dell Wyse Updates Its Cloud VDI Lineup
Dell and its new in-house virtual desktop supplier, Wyse, on Oct. 9 introduced some new and upgraded cloud client computing packages optimized expressly for VMware shops.
The company made the announcement at VMworld Europe 2012, which continues through Oct. 11 in Barcelona, Spain.
Virtual desktop implementations have come a long, long way in terms of quality and ease of deployment in the last couple of years after spending more than a decade in the doldrums.
With its new offerings, Dell is expanding its support for VMware View and its PC over Internet Protocol (PCoIP) on its zero and thin clients, desktop as a service (DaaS) with PCoIP support and new channel partner solutions to Europe.
VMware Adds Motorola to List of Smartphone Makers Using Horizon Mobile
VMware said it is working with Motorola Mobility to ship smartphones running VMware’s Horizon Mobile service through Softbank Telecom, a wireless carrier in Japan.
Horizon Mobile is VMware’s solution to the problem of how companies can accommodate employees who want to bring their own devices to work, but let the company secure corporate data on them. Horizon works by running a work phone as a virtual instance on the employee’s personal smartphone.
The Motorola Razr 201M will be the first Motorola smartphone to be enabled for Horizon Mobile, which will be offered as a managed service with a trial period beginning Dec. 10 in Japan, VMware said.
At VMworld 2012 in August, VMware announced that Horizon would be available on Samsung and LG smartphones and that its product road map calls for the service to be added in the near future to Apple iPhones and iPads running the iOS operating system.
Red Hat Virtualization Suite Now Integrates With RH Storage
Red Hat, the world's largest independent open-source software maker, Dec. 5 updated its virtualization and cloud platforms so that they plug into each other, scale, network and handle storage more efficiently.
Strangely, Red Hat Enterprise Virtualization's previous KVM hypervisor iterations did not integrate directly with the company's own Red Hat Storage server software. Workarounds always were required.
The new release, RHEV 3.1, solves that problem, thanks to the GlusterFS cloud-ready distributed storage system Red Hat obtained from its October 2011 acquisition of Gluster Inc.
CloudForms 1.1, Red Hat's open hybrid cloud management platform, now includes updates and new language enablement.
Key New Features in RHEV 3.1
RHEV 3.1 now offers increased scalability for guest virtual machines, provides support for up to 160 logical CPUs and up to 2TB of memory per virtual machine, and can update its KVM hypervisor to support the latest x86-type chips. It shares the same base KVM hypervisor technology as Red Hat Enterprise Linux and maintains common ABI compatibility.
RHEV 3.1 sports an updated user interface, an improved cross-platform Web administration portal, updated reporting dashboard, and new networking capabilities. The incorporation of a technology preview of storage live migration adds the flexibility to migrate virtual machine disk files between storage domains without having to power down the virtual machine, Stein said.
Finally, RHEV 3.1 also expands its localization enablement with support for English, French, Spanish, Simplified Chinese and Japanese, enabling the platform to be used even more widely around the globe.
Red Hat positions its virtualization suite as the only end-to-end enterprise open source virtualization infrastructure on the market. The platform's Kernel-based Virtual Machine (KVM) hypervisor currently holds 19 of the 27 published SPECvirt_sc2010 performance benchmarks, Red Hat Vice President of Cloud Computing Brian Stein said during a press conference webcast.
These include the best 2-socket and 4-socket scores and the only published 8-socket scores, he said.
Priced Below Competitors
Red Hat claims that it prices its virtualization platform 50 to 70 percent less than competitors such as VMware (ESX and vSphere) and Citrix XenServer.
Virtualization Will Keep Driving Storage Market in 2013: 10 Reasons Why
The just-ended 2012 was clearly a year of innovation for the storage industry. Companies in the sector had to innovate because of increased growth in enterprise and personal data, larger application workloads and demand for greater performance. This growth put greater pressure on storage platforms to work fast and efficiently. NAND flash, automated tiering software and advanced networking capabilities have helped enterprises handle the ever-larger storage volumes. Although these advancements have played a pivotal role in the revitalization of the storage industry, the biggest factor for the changing face of storage is virtualization. IDC predicts that in 2013, 69 percent of workloads will be virtualized. The wide adoption of virtualization and the transition to the software-defined data center are creating a new order in the storage business. How will this continue to evolve in 2013 and beyond? To find out, eWEEK worked with Ed Lee, architect for new-gen storage provider Tintri, who shared 10 insights on trends to watch for in 2013. Prior to Tintri, Lee was the principal systems architect at data domain (now owned by EMC) and was an original member of the Berkeley RAID team.
Apache Hadoop-Based Hortonworks Sandbox Launches
Hortonworks, a leading contributor to Apache Hadoop, has announced a new offering to help enterprises quickly get the hang of working with Hadoop and other projects in the Hadoop big data ecosystem.
Presented as a free 2GB download, the new Hortonworks Sandbox is a self-contained virtual machine with Apache Hadoop preconfigured. It is essentially a personal, portable and stand-alone Hadoop environment with a set of hands-on, step-by-step tutorials that enable users to learn and explore Hadoop on their own.
Hortonworks says its new offering takes users from zero to big data in 15 minutes. In any case, it is an on-ramp for users interested in learning, evaluating or using Apache Hadoop in the enterprise. The Hortonworks Sandbox addresses the gap between people who want to learn Hadoop and the complexity of setting up a small Hadoop cluster with an integrated environment where users, whether Hadoop novices or experts, can access demos, videos and multilevel tutorials. The Hortonworks Sandbox is built using the new Hortonworks Data Platform (HDP) 1.2, the company's open source platform powered by Apache Hadoop that delivers high-scale data processing in a manner that is easy for an enterprise to operate, Hortonworks officials said.
Apache Hadoop is an open-source software framework that supports data-intensive distributed applications. It supports the running of applications on large clusters of commodity hardware. Hadoop enables applications to be divided into many small fragments of work, each of which may be executed on any node in the cluster. As a key technology in the effort to wrangle big data, Hadoop enables applications to work with thousands of computation-independent computers and petabytes of data.
Subscribe to:
Comments (Atom)