AirMagnet has been in the WLAN game for a lot of years. Now part of Fluke Networks, it shares company with the likes of Ekahau and the Berkeley Varitronics for putting top-quality tools in the hands of those designing and supporting Wi-Fi networks that simply must function well. Good planning means less grief when clients hit the air, and in a perfect world there's very little mystery to the operation of the business WLAN of any size. And good planning for the small business wireless environment is what AirMagnet's latest Planner utility is all about.

I routinely use AirMagnet Planner (and Ekahau Site Survey) for modeling new wireless spaces before actual deployment. Whereas my designs tend to be aimed at large buildings or outdoor areas, the occasional smaller-space design does get the same treatment. By accurately modeling the environment that your access points will work in, you can achieve impressive predictive accuracy on how many APs will be needed to provide various levels of capacity and data rates in a variety of "what-if" scenarios. Better tools--and I include AirMagnet Planner for Cisco Small Business in the category--allow for easy changing of wall types and other attenuation sources to show how things like new modular furniture might affect the RF environment of a given space.

But what about the Cisco Small Business part of the planner's name? Between Cisco's big-dollar Enterprise Wireless product set and its Linksys home-networking line lives a niche of access points and wireless routers aimed at the small-business space. Most of these devices have profiles in the AirMagnet Planner tool, making for highly accurate simulation at planning time. For example, if I'm designing a wireless network for a car dealership and want to determine how many Cisco WAP4410N access points will be needed to provide a high-throughput WLAN, I simply plop down APs on the floor plan and ask the tool to tell me how I'm doing.

Antenna characteristics and available power and channel settings are built in for each Cisco Small Business wireless component so you can compare different components virtually before settling on a design. You can even have the tool take a stab at doing AP placement work in "advisor" mode, after you designate minimum required data rate, areas where signal must be present, locations where APs cannot be placed and the like.

Given the exponential growth in the use of Wi-Fi for all sorts of business applications, I am very pleased to see a top-tier WLAN toolmaker like AirMagnet recognize the need for small businesses and those who support them to get wireless right. Despite the increasing overall user savvy that comes with more devices in more hands, system design is still an area fraught with confusion. Wireless networks can be overbuilt to the tune of wasted dollars, or underbuilt and therefore unusable. Neither situation is good, and small business tend to have less tolerance for wasted infrastructure dollars. With AirMagnet Planner for Cisco Small Business properly used for wireless design, small business won't have to worry about the design quality of their WLANs.

Disclaimer: I do not use AirMagnet Planner for Cisco Small Business, but I am a customer of AirMagnet and Ekahau, as mentioned. There are only so many enterprise-quality wireless tools out there. ]]> AirMagnet Planner for Cisco Small Business brings big-network quality to smaller wireless environments with the same excellent tool set that has been the genesis for many successful enterprise WLANs.

AirMagnet has been in the WLAN game for a lot of years. Now part of Fluke Networks, it shares company with the likes of Ekahau and the Berkeley Varitronics for putting top-quality tools in the hands of those designing and supporting Wi-Fi networks that simply must function well. Good planning means less grief when clients hit the air, and in a perfect world there's very little mystery to the operation of the business WLAN of any size. And good planning for the small business wireless environment is what AirMagnet's latest Planner utility is all about.

I routinely use AirMagnet Planner (and Ekahau Site Survey) for modeling new wireless spaces before actual deployment. Whereas my designs tend to be aimed at large buildings or outdoor areas, the occasional smaller-space design does get the same treatment. By accurately modeling the environment that your access points will work in, you can achieve impressive predictive accuracy on how many APs will be needed to provide various levels of capacity and data rates in a variety of "what-if" scenarios. Better tools--and I include AirMagnet Planner for Cisco Small Business in the category--allow for easy changing of wall types and other attenuation sources to show how things like new modular furniture might affect the RF environment of a given space.

But what about the Cisco Small Business part of the planner's name? Between Cisco's big-dollar Enterprise Wireless product set and its Linksys home-networking line lives a niche of access points and wireless routers aimed at the small-business space. Most of these devices have profiles in the AirMagnet Planner tool, making for highly accurate simulation at planning time. For example, if I'm designing a wireless network for a car dealership and want to determine how many Cisco WAP4410N access points will be needed to provide a high-throughput WLAN, I simply plop down APs on the floor plan and ask the tool to tell me how I'm doing.

Antenna characteristics and available power and channel settings are built in for each Cisco Small Business wireless component so you can compare different components virtually before settling on a design. You can even have the tool take a stab at doing AP placement work in "advisor" mode, after you designate minimum required data rate, areas where signal must be present, locations where APs cannot be placed and the like.

Given the exponential growth in the use of Wi-Fi for all sorts of business applications, I am very pleased to see a top-tier WLAN toolmaker like AirMagnet recognize the need for small businesses and those who support them to get wireless right. Despite the increasing overall user savvy that comes with more devices in more hands, system design is still an area fraught with confusion. Wireless networks can be overbuilt to the tune of wasted dollars, or underbuilt and therefore unusable. Neither situation is good, and small business tend to have less tolerance for wasted infrastructure dollars. With AirMagnet Planner for Cisco Small Business properly used for wireless design, small business won't have to worry about the design quality of their WLANs.

Disclaimer: I do not use AirMagnet Planner for Cisco Small Business, but I am a customer of AirMagnet and Ekahau, as mentioned. There are only so many enterprise-quality wireless tools out there. ]]> http://www.networkcomputing.com/data-networking-management/240001066/feed/ http://www.networkcomputing.com/unified-communications-voip/240001043 http://www.networkcomputing.com/unified-communications-voip/240001043#comments Fri, 25 May 2012 13:45:00 +0000 Esther Shein Uncategorized http://www.networkcomputing.com/unified-communications-voip/240001043 The study of 500 business and IT executives focused on unified communications as a service. "You have companies that might be familiar with what many [communications] technologies are because they've been in news or use them in their personal lives, [and] they start to think this could be great if those types of things would be available for their business--and they would be," says Seth Robinson, director of technology analysis at CompTIA, a nonprofit trade association for IT professionals and companies. "But the reality is, getting everything established and talking together and getting users to actually use the tools in the way they're meant to be used is a tricky proposition."

Managed services and cloud-based delivery can help improve unified communications by removing some of that complexity, as well as bring perspective to what a business needs, Robinson says. Communications providers also know the questions to ask that will match tools with the business strategy, such as where a business needs to be and how to get there.

"They can suggest the right suite of tools that they're familiar with and manage and maintain them, and have a good understanding of how the tools can be incorporated in the business. Certainly, as companies have aging infrastructures and look to replace [equipment] the cloud has lot of benefit," he says, adding that goes for any other IT-related service.

Companies need to look at the unified communications benefits the cloud can provide and figure out whether going that route will cost them less, create more agility and give them more features. The challenge providers have is "to build a holistic platform that considers the various tools available and the best way to utilize them," according to the study. "By providing an integrated experience tied to business processes and then adding on education, the time for communications tools to become accepted can be shortened."

Next: Managed UC Services and Budget Considerations In the final analysis, says Robinson, the decision to move to a managed service model "is really going to depend on how critical certain types of communication are to a business, and if they need things to be very reliable." If they don't think they can get assurances from a cloud provider, they may opt to keep certain aspects of unified communications on-premises, while moving secondary services off-site.

That leads to a new trend toward a hybrid model of providers offering both on-premise and off-site services. Companies will likely choose to keep voice on-site, Robinson says, but may opt to move instant messaging to the cloud, since it may be something they're experimenting with and not heavily using. However, Robinson adds that there are companies that offer VoIP as a service, "so it comes down to cost equation, backup and other issues."

The study also found that budgets for communications/collaboration are increasing because companies want workers to be more productive, which leads to some improvement in business metrics; and that there is an issue with end-user adoption. Among the reasons, says Robinson, are the learning curve and that end users are going to use tools with which they're most familiar.

The most surprising finding from the study, he says, was the priorities people put on their communications goals. Usually, people cite improving cost savings and productivity as their typical goals, he says, but "what was surprising was that business metrics and business objectives were lower on the list of priorities. So the way we look at that is, companies are probably saying, 'We want these communications to help with productivity,' but really, they may want to shorten sales cycles or [achieve] better customer satisfaction."

Says Robinson, "To the extent that companies can make a more direct link with the communications strategy and end goal of the business metric or objective will help them see how to implement that business strategy." ]]> The study of 500 business and IT executives focused on unified communications as a service. "You have companies that might be familiar with what many [communications] technologies are because they've been in news or use them in their personal lives, [and] they start to think this could be great if those types of things would be available for their business--and they would be," says Seth Robinson, director of technology analysis at CompTIA, a nonprofit trade association for IT professionals and companies. "But the reality is, getting everything established and talking together and getting users to actually use the tools in the way they're meant to be used is a tricky proposition."

Managed services and cloud-based delivery can help improve unified communications by removing some of that complexity, as well as bring perspective to what a business needs, Robinson says. Communications providers also know the questions to ask that will match tools with the business strategy, such as where a business needs to be and how to get there.

"They can suggest the right suite of tools that they're familiar with and manage and maintain them, and have a good understanding of how the tools can be incorporated in the business. Certainly, as companies have aging infrastructures and look to replace [equipment] the cloud has lot of benefit," he says, adding that goes for any other IT-related service.

Companies need to look at the unified communications benefits the cloud can provide and figure out whether going that route will cost them less, create more agility and give them more features. The challenge providers have is "to build a holistic platform that considers the various tools available and the best way to utilize them," according to the study. "By providing an integrated experience tied to business processes and then adding on education, the time for communications tools to become accepted can be shortened."

Next: Managed UC Services and Budget Considerations In the final analysis, says Robinson, the decision to move to a managed service model "is really going to depend on how critical certain types of communication are to a business, and if they need things to be very reliable." If they don't think they can get assurances from a cloud provider, they may opt to keep certain aspects of unified communications on-premises, while moving secondary services off-site.

That leads to a new trend toward a hybrid model of providers offering both on-premise and off-site services. Companies will likely choose to keep voice on-site, Robinson says, but may opt to move instant messaging to the cloud, since it may be something they're experimenting with and not heavily using. However, Robinson adds that there are companies that offer VoIP as a service, "so it comes down to cost equation, backup and other issues."

The study also found that budgets for communications/collaboration are increasing because companies want workers to be more productive, which leads to some improvement in business metrics; and that there is an issue with end-user adoption. Among the reasons, says Robinson, are the learning curve and that end users are going to use tools with which they're most familiar.

The most surprising finding from the study, he says, was the priorities people put on their communications goals. Usually, people cite improving cost savings and productivity as their typical goals, he says, but "what was surprising was that business metrics and business objectives were lower on the list of priorities. So the way we look at that is, companies are probably saying, 'We want these communications to help with productivity,' but really, they may want to shorten sales cycles or [achieve] better customer satisfaction."

Says Robinson, "To the extent that companies can make a more direct link with the communications strategy and end goal of the business metric or objective will help them see how to implement that business strategy." ]]> http://www.networkcomputing.com/unified-communications-voip/240001043/feed/ http://www.networkcomputing.com/wireless/240000916 http://www.networkcomputing.com/wireless/240000916#comments Thu, 24 May 2012 14:23:00 +0000 Greg Ferro Uncategorized http://www.networkcomputing.com/wireless/240000916 Personally, I've found company-provided IT equipment to be substandard and of limited use. In the worst example, about a year ago I was given a 6-year-old laptop with Windows XP, a Celeron CPU and just 1 GB memory. I could run Outlook or MS Word, but not both. I gave the computer back and refused to use it. More recently, it was a brand-new Lenovo T420. I was hopeful? In practice, however, the Symantec bloatware (Antivirus, pcAnywhere, Altiris) caused the machine to reboot about once per day. This led to consistent loss of data and work, as well as frustration. I now do most of my work with a personal laptop.

These are just specific examples. I also waste a lot of time working around the limitations of full disk encryption, use of corporately mandated software, virus/malware engines, remote control, VPN clients and the bloated MS Office software that isn't optimal for everyday use. As a network engineer, I need a lot of non-standard software--PuTTy, Java clients--to function every day. It usually takes days for approvals and authorizations to happen before I can get the software installed by the corporate desktop team.

As a freelancer, I can and do make extensive use of online services such as Dropbox, GoToMeeting, Skype and Google Apps. I also use online services for automatic laptop backup. My address book is synchronized to my phone and computer, and backed up via Apple's iCloud.

These services set my expectation for BYOD and the service level I want to achieve if my devices are inside the corporate network. In fact, they set the expectation for what corporate IT should provide for all its users.

There are very few companies that will let me use these services because they simply don't conform to corporate policies for issues such as security, data retention and control. My data is distributed; they want data centralized. My information is mine; they want control of everything, since they can't differentiate between personal and company information. I want ease of use and freedom of expression; corporate IT wants ownership, restricted access with data controls, and, finally, remote wipe capabilities.

In the current method, BYOD doesn't work for me. I will never connect my devices to the corporate networks, for fear of theft of my personal data. Or that the company would wipe my personal data from all of my devices, including my phone, laptop and tablet. The company will also likely prevent me from using my most productive apps. And I'm going to loathe it for that.

I want to use tools that make me productive, not tools that suit corporations' need for cost control, asset management or data loss prevention. And that's the BYOD problem. The reality is that people want better tools and would prefer to use devices that deliver that. In fact, they're demanding it.

The tension between consumer usage and corporate purpose is a massive gulf of expectation deficit. Consumer technology such as Apple's iPhone and Google Android can be truly useful, and a delight to use.

The corporate IT of today is simply not able to deliver the promise already offered by consumer technology. The IT department will hold back productivity and prevent effective use of tools because of good corporate policies--policies that are based on sound, old-fashioned reasons rooted in law and good corporate governance.

And users are going to hate us for it. ]]> Personally, I've found company-provided IT equipment to be substandard and of limited use. In the worst example, about a year ago I was given a 6-year-old laptop with Windows XP, a Celeron CPU and just 1 GB memory. I could run Outlook or MS Word, but not both. I gave the computer back and refused to use it. More recently, it was a brand-new Lenovo T420. I was hopeful? In practice, however, the Symantec bloatware (Antivirus, pcAnywhere, Altiris) caused the machine to reboot about once per day. This led to consistent loss of data and work, as well as frustration. I now do most of my work with a personal laptop.

These are just specific examples. I also waste a lot of time working around the limitations of full disk encryption, use of corporately mandated software, virus/malware engines, remote control, VPN clients and the bloated MS Office software that isn't optimal for everyday use. As a network engineer, I need a lot of non-standard software--PuTTy, Java clients--to function every day. It usually takes days for approvals and authorizations to happen before I can get the software installed by the corporate desktop team.

As a freelancer, I can and do make extensive use of online services such as Dropbox, GoToMeeting, Skype and Google Apps. I also use online services for automatic laptop backup. My address book is synchronized to my phone and computer, and backed up via Apple's iCloud.

These services set my expectation for BYOD and the service level I want to achieve if my devices are inside the corporate network. In fact, they set the expectation for what corporate IT should provide for all its users.

There are very few companies that will let me use these services because they simply don't conform to corporate policies for issues such as security, data retention and control. My data is distributed; they want data centralized. My information is mine; they want control of everything, since they can't differentiate between personal and company information. I want ease of use and freedom of expression; corporate IT wants ownership, restricted access with data controls, and, finally, remote wipe capabilities.

In the current method, BYOD doesn't work for me. I will never connect my devices to the corporate networks, for fear of theft of my personal data. Or that the company would wipe my personal data from all of my devices, including my phone, laptop and tablet. The company will also likely prevent me from using my most productive apps. And I'm going to loathe it for that.

I want to use tools that make me productive, not tools that suit corporations' need for cost control, asset management or data loss prevention. And that's the BYOD problem. The reality is that people want better tools and would prefer to use devices that deliver that. In fact, they're demanding it.

The tension between consumer usage and corporate purpose is a massive gulf of expectation deficit. Consumer technology such as Apple's iPhone and Google Android can be truly useful, and a delight to use.

The corporate IT of today is simply not able to deliver the promise already offered by consumer technology. The IT department will hold back productivity and prevent effective use of tools because of good corporate policies--policies that are based on sound, old-fashioned reasons rooted in law and good corporate governance.

And users are going to hate us for it. ]]> http://www.networkcomputing.com/wireless/240000916/feed/ http://www.networkcomputing.com/data-networking-management/240000947 http://www.networkcomputing.com/data-networking-management/240000947#comments Thu, 24 May 2012 13:52:00 +0000 Gary Hilson Uncategorized http://www.networkcomputing.com/data-networking-management/240000947 A recent report by Redwood City, Calif.-based Dell'Oro Group found shipments of 10 Gbps Ethernet controllers grew by double digits during the first quarter of 2012. The firm predicts this trend will continue during the next several quarters, says Dell'Oro analyst Sameh Boujelbene, as server manufacturers introduce new products based on Intel's "Romley" server platform.

"The two big drivers for 10GbE adoption are virtualization and big data," says Boujelbene. If enterprises are to leverage the capabilities of more powerful servers, connectivity must also be improved to avoid bottlenecks.

Boujelbene says it makes sense for enterprises to upgrade their entire ecosystem if they're already investing in servers with more processing power and 10GbE capability.

According to IDC, the worldwide Ethernet switch market reached $5.1 billion in the first quarter of 2012, representing strong growth of 7.4% year over year. Meanwhile, the worldwide router market increased a more modest 2.8% year over year in the same quarter.

Cindy Borovick, IDC's program vice president for enterprise and data center networks, says the growth is driven by the network infrastructure needs of enterprise IT as it embraces cloud technologies. While Gigabit Ethernet is still supporting a variety of applications at the network edge, 10GbE is driving incremental growth in datacenter and campus core deployments.

Borovick says the 10GbE adoption is also affected by data center consolidation and the increased server density that comes from virtualization. The servers are more robust and are handling greater workloads, she says. "We've reached the point where the network has become the bottleneck."

Bob Laliberte, a senior analyst at Enterprise Strategy Group Server, says virtualization is increasing throughput requirements, as the number of virtual machines per physical server jumps. "Enterprises are increasing virtual machine density and really driving up the throughput needs," he explains.

While server virtualization is a primary driver for adopting 10GbE, Laliberte says the costs to upgrade from 1GbE has decreased enough for enterprises to consider an end-to-end refresh, especially as they consolidate data centers and look at setting up private cloud computing environments. "If they are buying new servers, they are probably doing a refresh anyway," he says.

In addition, 10GbE has become a standard feature on servers, adds Laliberte, so it makes sense to take advantage of the capability because more powerful processors allow for more virtual machine density.

A recent survey by Enterprise Strategy Group of 280 North American IT professionals found that both current requirements and future considerations were factors in adopting 10GbE. The top five considerations were:

• 43%: Server virtualization increases throughput requirements.
• 42%: Current or anticipated data center traffic.
• 41%: Costs have decreased to an acceptable level.
• 36%: Data center consolidation is driving massive data center scale.
• 35%: Current or future implementation of a private cloud in our data center.

Data center traffic is changing, adds IDC's Borovick. "There is an explosion in the use of Ethernet to handle storage traffic. When you move to 10GbE, you can consolidate all traffic onto one network."

Meanwhile, adds Petr Jirovsky, a senior research analyst in IDC's Networking Trackers Group, 40GbE uplinks are becoming more common. "By 2020, we'll be talking about 100GbE in the core." ]]> A recent report by Redwood City, Calif.-based Dell'Oro Group found shipments of 10 Gbps Ethernet controllers grew by double digits during the first quarter of 2012. The firm predicts this trend will continue during the next several quarters, says Dell'Oro analyst Sameh Boujelbene, as server manufacturers introduce new products based on Intel's "Romley" server platform.

"The two big drivers for 10GbE adoption are virtualization and big data," says Boujelbene. If enterprises are to leverage the capabilities of more powerful servers, connectivity must also be improved to avoid bottlenecks.

Boujelbene says it makes sense for enterprises to upgrade their entire ecosystem if they're already investing in servers with more processing power and 10GbE capability.

According to IDC, the worldwide Ethernet switch market reached $5.1 billion in the first quarter of 2012, representing strong growth of 7.4% year over year. Meanwhile, the worldwide router market increased a more modest 2.8% year over year in the same quarter.

Cindy Borovick, IDC's program vice president for enterprise and data center networks, says the growth is driven by the network infrastructure needs of enterprise IT as it embraces cloud technologies. While Gigabit Ethernet is still supporting a variety of applications at the network edge, 10GbE is driving incremental growth in datacenter and campus core deployments.

Borovick says the 10GbE adoption is also affected by data center consolidation and the increased server density that comes from virtualization. The servers are more robust and are handling greater workloads, she says. "We've reached the point where the network has become the bottleneck."

Bob Laliberte, a senior analyst at Enterprise Strategy Group Server, says virtualization is increasing throughput requirements, as the number of virtual machines per physical server jumps. "Enterprises are increasing virtual machine density and really driving up the throughput needs," he explains.

While server virtualization is a primary driver for adopting 10GbE, Laliberte says the costs to upgrade from 1GbE has decreased enough for enterprises to consider an end-to-end refresh, especially as they consolidate data centers and look at setting up private cloud computing environments. "If they are buying new servers, they are probably doing a refresh anyway," he says.

In addition, 10GbE has become a standard feature on servers, adds Laliberte, so it makes sense to take advantage of the capability because more powerful processors allow for more virtual machine density.

A recent survey by Enterprise Strategy Group of 280 North American IT professionals found that both current requirements and future considerations were factors in adopting 10GbE. The top five considerations were:

• 43%: Server virtualization increases throughput requirements.
• 42%: Current or anticipated data center traffic.
• 41%: Costs have decreased to an acceptable level.
• 36%: Data center consolidation is driving massive data center scale.
• 35%: Current or future implementation of a private cloud in our data center.

Data center traffic is changing, adds IDC's Borovick. "There is an explosion in the use of Ethernet to handle storage traffic. When you move to 10GbE, you can consolidate all traffic onto one network."

Meanwhile, adds Petr Jirovsky, a senior research analyst in IDC's Networking Trackers Group, 40GbE uplinks are becoming more common. "By 2020, we'll be talking about 100GbE in the core." ]]> http://www.networkcomputing.com/data-networking-management/240000947/feed/ http://www.networkcomputing.com/private-cloud/240000907 http://www.networkcomputing.com/private-cloud/240000907#comments Wed, 23 May 2012 20:40:00 +0000 JD Sartain Uncategorized http://www.networkcomputing.com/private-cloud/240000907 EMC World, Puppet Labs announced Razor, a new open-source infrastructure management tool developed with EMC. Razor provides comprehensive automation to system administrators for the distributed enterprise and private cloud provisioning.

According to Puppet Labs spokesman Ray George, Razor (which is freely available and open sourced under the Apache 2.0 license) is the first hardware provisioning tool with the responsiveness and productivity of the cloud. "Puppet can now automate every phase of the IT infrastructure lifecycle, from bare metal to deployed cloud applications," says George.

Servers shipped in bulk from manufacturers (Dell, HP, IBM, etc.) typically don't have installed operating systems--they're just chunks of metal holding CPUs, disk and memory (i.e., "bare metal"), explains George. Razor provisions the correct operating system software onto the bare metal server as the first step in building up a server to host an application--for example, a Web server or a database. This type of tool is generally used by the system administrators responsible for building a server.

"Puppet Enterprise starts at the operating-system layer and configures and manages this layer and the application layers above this one. Razor sits below PE, provisions the operating system on top of the bare metal, automatically installs PE onto the operating system, and then hands off to PE, which allows it to continue building the server," adds George.

Razor provides autodiscovered, real-time inventory data for every hardware node, eliminating inefficient, error-prone manual processes. As a result, system administrators get full insight into the latest, up-to-date status of their hardware inventory. Razor automatically selects the correct operating system image based on the autodiscovered, real-time inventory data, removing the need for manual intervention whenever there's a change in hardware configuration, explains George.

Different server hardware configurations are optimized for different applications. For example, for database servers, the system administrator usually wants a hardware server with lots of hard-drive space and, for Web servers, he wants a hardware server with a fast CPU. Previously, in order to match the operating system for database servers to the hardware server configured for database servers, system administrators were required to manually perform this function and then maintain some sort of record of this "match" in a spreadsheet.

"With Razor, the system administrator establishes rules; for example, 'only match database operating system software with server hardware that has lots of hard drive space,' and/or 'only match web servers with server hardware that has fast CPUs,'" George says. "Then, when the hardware server starts up, Razor automatically discovers the inventory of the hardware server ('looks like this hardware server has lots of hard drive space') and, based on the rules, automatically matches that hardware server with the appropriate operating system. No more manual matching and no more manual inventory and tracking. And, if the hardware changes; that is, upgrades the CPU or removes some hard drives, Razor uses the rules to determine--automatically and dynamically--whether it should select a different operating system the next time the server starts up."

Other available cloud friendly infrastructure management tools, which are complimentary to Puppet Labs include Zenoss, RightScale and Nodeable. ]]> EMC World, Puppet Labs announced Razor, a new open-source infrastructure management tool developed with EMC. Razor provides comprehensive automation to system administrators for the distributed enterprise and private cloud provisioning.

According to Puppet Labs spokesman Ray George, Razor (which is freely available and open sourced under the Apache 2.0 license) is the first hardware provisioning tool with the responsiveness and productivity of the cloud. "Puppet can now automate every phase of the IT infrastructure lifecycle, from bare metal to deployed cloud applications," says George.

Servers shipped in bulk from manufacturers (Dell, HP, IBM, etc.) typically don't have installed operating systems--they're just chunks of metal holding CPUs, disk and memory (i.e., "bare metal"), explains George. Razor provisions the correct operating system software onto the bare metal server as the first step in building up a server to host an application--for example, a Web server or a database. This type of tool is generally used by the system administrators responsible for building a server.

"Puppet Enterprise starts at the operating-system layer and configures and manages this layer and the application layers above this one. Razor sits below PE, provisions the operating system on top of the bare metal, automatically installs PE onto the operating system, and then hands off to PE, which allows it to continue building the server," adds George.

Razor provides autodiscovered, real-time inventory data for every hardware node, eliminating inefficient, error-prone manual processes. As a result, system administrators get full insight into the latest, up-to-date status of their hardware inventory. Razor automatically selects the correct operating system image based on the autodiscovered, real-time inventory data, removing the need for manual intervention whenever there's a change in hardware configuration, explains George.

Different server hardware configurations are optimized for different applications. For example, for database servers, the system administrator usually wants a hardware server with lots of hard-drive space and, for Web servers, he wants a hardware server with a fast CPU. Previously, in order to match the operating system for database servers to the hardware server configured for database servers, system administrators were required to manually perform this function and then maintain some sort of record of this "match" in a spreadsheet.

"With Razor, the system administrator establishes rules; for example, 'only match database operating system software with server hardware that has lots of hard drive space,' and/or 'only match web servers with server hardware that has fast CPUs,'" George says. "Then, when the hardware server starts up, Razor automatically discovers the inventory of the hardware server ('looks like this hardware server has lots of hard drive space') and, based on the rules, automatically matches that hardware server with the appropriate operating system. No more manual matching and no more manual inventory and tracking. And, if the hardware changes; that is, upgrades the CPU or removes some hard drives, Razor uses the rules to determine--automatically and dynamically--whether it should select a different operating system the next time the server starts up."

Other available cloud friendly infrastructure management tools, which are complimentary to Puppet Labs include Zenoss, RightScale and Nodeable. ]]> http://www.networkcomputing.com/private-cloud/240000907/feed/ http://www.networkcomputing.com/data-networking-management/240000894 http://www.networkcomputing.com/data-networking-management/240000894#comments Wed, 23 May 2012 16:30:00 +0000 Steve Wexler Uncategorized http://www.networkcomputing.com/data-networking-management/240000894 7950 XRS supports 32 Tbps capacity and 160 100 Gigabit Ethernet ports in a single system, which is five times the density of today's core routers, and also slashes power consumption by more than 66% compared with typical core routers, according to the company.

Based on Alcatel-Lucent's 400G FP3 network processor unit, the router can support both IP backbone networks and regional "metro" core networks. The company says 7950 XRS also extends its commitment to tighter integration between IP networks and optical transport networks, including support for transponder integration into the 7950 XRS at 10G, 40G and 100G speeds. It can be managed with the 5620 Service Aware Manager platform and an Optical Extension Shelf capability that integrates the 1830 Photonic Service Switch with the 7950 XRS.

Last month, Broadcom introduced a line of programmable 100GbE switch processors to address the exponential traffic growth on Internet back-haul switches. A forecast from Infonetics Research predicts a compound annual growth rate of 170% in the number of 100 Gbps ports deployed through 2016.

Cisco Systems unveiled in February 100GbE capability for its Nexus 7000 line for data center and service provider networks. By 2016, sales of 40 and 100GbE products will amount to $3 billion, reported Dell'Oro Group in its five-year forecast for the Ethernet switch market. According to Dell'Oro, the market includes Alcatel-Lucent, Cisco, Avaya, Brocade, Extreme Networks, Dell, HP, IBM and Juniper Networks.

Service providers face a serious problem as they tackle 100G, the next great inflection point in their routing, switching and optical networks, says Michael Howard, co-founder and principal analyst, carrier networks at Infonetics Research. If they keep adding more of the same equipment to their networks, they will end up multiplying space and power requirements.

All the major router players--Huawei, Cisco, Juniper, Alcatel-Lucent and even Brocade, although it's a smaller SP router player--have 100GbE blades for at least some of their routers, says Howard. "There have been two very formidable core router competitors in the market for many years (in 2011, first-place Cisco and second-place Juniper held around 80% of the market). In 1Q12, Cisco, Juniper, Huawei had 96% of the core router market."

Howard adds that he doesn't believe Alcatel-Lucent's entry will substantially change the total available market (TAM) for core routers. "Service providers invested [approximately] $3B on core routers worldwide last year, and we expect [approximately] $4B to 5B in 2016."

While Alcatel-Lucent (ALU) won't grow the TAM, Howard says it's the only router company that could credibly enter this mature stage of the core router market. Service providers must trust their router vendors with their core networks, which carry the highest customer traffic loads and therefore represent the greatest risk if a failure occurs, or a service does not work properly, he says.

"We believe ALU will sell core routers and grow their revenue, at a minimum with the existing edge-router customer base," he says. "ALU's current customers will be getting hardware and software that has already been proven in the 7750/7740 edge router platforms. That is, the FP3 chipset and the same network operating system software."

Infonetics expects Alcatel-Lucent to be evaluated by some non-customers, since many operators are looking at 100G/100GE as a major inflection point, around which they are refining the design of their optical and data networks, and over time will take market share.

"It won't be easy. ALU faces a big challenge: It is hard for service providers to change or add new products and/or new vendors in any critical area, and core routing is a critical area," Howard says. "The competition among core router vendors could get fierce over the next two years."

Learn more about Research: 2012 IT Spending Priorities Survey by subscribing to Network Computing Pro Reports (free, registration required). ]]> 7950 XRS supports 32 Tbps capacity and 160 100 Gigabit Ethernet ports in a single system, which is five times the density of today's core routers, and also slashes power consumption by more than 66% compared with typical core routers, according to the company.

Based on Alcatel-Lucent's 400G FP3 network processor unit, the router can support both IP backbone networks and regional "metro" core networks. The company says 7950 XRS also extends its commitment to tighter integration between IP networks and optical transport networks, including support for transponder integration into the 7950 XRS at 10G, 40G and 100G speeds. It can be managed with the 5620 Service Aware Manager platform and an Optical Extension Shelf capability that integrates the 1830 Photonic Service Switch with the 7950 XRS.

Last month, Broadcom introduced a line of programmable 100GbE switch processors to address the exponential traffic growth on Internet back-haul switches. A forecast from Infonetics Research predicts a compound annual growth rate of 170% in the number of 100 Gbps ports deployed through 2016.

Cisco Systems unveiled in February 100GbE capability for its Nexus 7000 line for data center and service provider networks. By 2016, sales of 40 and 100GbE products will amount to $3 billion, reported Dell'Oro Group in its five-year forecast for the Ethernet switch market. According to Dell'Oro, the market includes Alcatel-Lucent, Cisco, Avaya, Brocade, Extreme Networks, Dell, HP, IBM and Juniper Networks.

Service providers face a serious problem as they tackle 100G, the next great inflection point in their routing, switching and optical networks, says Michael Howard, co-founder and principal analyst, carrier networks at Infonetics Research. If they keep adding more of the same equipment to their networks, they will end up multiplying space and power requirements.

All the major router players--Huawei, Cisco, Juniper, Alcatel-Lucent and even Brocade, although it's a smaller SP router player--have 100GbE blades for at least some of their routers, says Howard. "There have been two very formidable core router competitors in the market for many years (in 2011, first-place Cisco and second-place Juniper held around 80% of the market). In 1Q12, Cisco, Juniper, Huawei had 96% of the core router market."

Howard adds that he doesn't believe Alcatel-Lucent's entry will substantially change the total available market (TAM) for core routers. "Service providers invested [approximately] $3B on core routers worldwide last year, and we expect [approximately] $4B to 5B in 2016."

While Alcatel-Lucent (ALU) won't grow the TAM, Howard says it's the only router company that could credibly enter this mature stage of the core router market. Service providers must trust their router vendors with their core networks, which carry the highest customer traffic loads and therefore represent the greatest risk if a failure occurs, or a service does not work properly, he says.

"We believe ALU will sell core routers and grow their revenue, at a minimum with the existing edge-router customer base," he says. "ALU's current customers will be getting hardware and software that has already been proven in the 7750/7740 edge router platforms. That is, the FP3 chipset and the same network operating system software."

Infonetics expects Alcatel-Lucent to be evaluated by some non-customers, since many operators are looking at 100G/100GE as a major inflection point, around which they are refining the design of their optical and data networks, and over time will take market share.

"It won't be easy. ALU faces a big challenge: It is hard for service providers to change or add new products and/or new vendors in any critical area, and core routing is a critical area," Howard says. "The competition among core router vendors could get fierce over the next two years."

Learn more about Research: 2012 IT Spending Priorities Survey by subscribing to Network Computing Pro Reports (free, registration required). ]]> http://www.networkcomputing.com/data-networking-management/240000894/feed/ http://www.networkcomputing.com/storage-networking-management/240000867 http://www.networkcomputing.com/storage-networking-management/240000867#comments Wed, 23 May 2012 13:52:00 +0000 Mike Fratto Uncategorized http://www.networkcomputing.com/storage-networking-management/240000867 EMC World, I focused more on some of the partners that augment and enhance EMC's product lines. I didn't get enough time to do all of the labs I wanted, nor attend many sessions. But what I saw, I liked.

VCE, the joint venture among VMware, Cisco and EMC, took the opportunity to highlight new feature enhancements to its Vblock architecture through integration with EMC's VPLEX, which federates storage across multiple locations. VCE also support EMC's Avamar, Data Domain and RecoverPoint products, all of which overlap somewhat in functionality, meaning Vblock customers can choose the data protection suite that fits a particular application's needs.

Given that the products are tested and supported by VCE, if there's a problem, there's one call to make. That's exactly what VCE CEO Michael Capellas says differentiates VCE from competitors' private cloud offerings. "You should not confuse individual components which come from a single company with integration," he explained at EMC World. "Other reference architectures tell you how to build it. Just because you bought a reference architecture doesn't mean that you have integrated products. If you have a problem, you call VCE and we take ownership."

I got a chance to sit down with Eric Herzog, SVP for product management and marketing of the unified storage division at EMC, who was excited about flash storage and the possibilities that tiering offers organizations. By utilizing an average of 5% of a storage system as flash, the array can offer two to three times better performance using flash. Traditionally, adding more IOPS meant adding more disks to an array, and the progression was largely linear. Flash can substantially increase performance without adding more disks. That means less power draw, and less space to hold the drives. He says the biggest hurdle he sees is educating storage admins both in organizations and EMC's channel partners that even though flash drives are far more expensive per unit than disk drives, to achieve equivalent performance, judicious use of flash plus disk can significantly reduce the overall system cost because you need far fewer disks. I suspect you'll be hearing a lot more about flash from EMC in the coming year.

Two EMC Partners to Watch

Many of EMC's partners were on display at EMC World--two caught my eye. Infineta makes a WAN optimizer that it claims competes on data reduction with the likes of Riverbed and Silver Peak, using in-memory deduplication and at 10 Gb speeds. Built using merchant silicon and an FPGA, Infineta uses RAM to deduplicate blocks as small as 8 bytes and stores the dictionary in 64 GB RAM. Since block sizes are so small, representatives claim, they can deduplicate within the same packet, as well as across packets. Infineta is focusing primarily on replication traffic from backups and SRDF, rather than on file traffic. The company is in the process of getting its equipment qualified by EMC, which will open the door to being an approved component of the SAN. Its story sounds good, but, of course, I can't vouch for Infineta's performance claims. If your current WAN optimization isn't cutting it for your replication needs, Infineta is worth investigating (and tell me what you find!).

What's a tech show in 2012 without some software-defined networking (SDN) goodness? I happened upon Vello Systems, which makes an SDN controller and optical switch that underpins it Remote Data Replication Manager (RDRM). The manager applies network policies via OpenFlow to replication data between data centers and can support, for example, EMC's SRDF and RecoverPoint, as well as VMware's Site Recovery Manager. Vello's RDRM seems targeted toward cloud providers, but any organization running fiber plant could look at Vello for WAN traffic enforcement.

If you're still at EMC World, take advantage of EMC's Hands-on Labs. There are 27 labs available. They're pretty well done; I ran through the vCloud Director lab in about an hour. There were several others I wanted to complete, but other obligations distracted me.

All in all, EMC put together a good show for its customers and partners. There were a bunch of sessions given by EMC staff and partners, plenty of food and seating if you needed to catch your breath, and a show floor that, while small, showcased both EMC and partner products and services.

]]> EMC World, I focused more on some of the partners that augment and enhance EMC's product lines. I didn't get enough time to do all of the labs I wanted, nor attend many sessions. But what I saw, I liked.

VCE, the joint venture among VMware, Cisco and EMC, took the opportunity to highlight new feature enhancements to its Vblock architecture through integration with EMC's VPLEX, which federates storage across multiple locations. VCE also support EMC's Avamar, Data Domain and RecoverPoint products, all of which overlap somewhat in functionality, meaning Vblock customers can choose the data protection suite that fits a particular application's needs.

Given that the products are tested and supported by VCE, if there's a problem, there's one call to make. That's exactly what VCE CEO Michael Capellas says differentiates VCE from competitors' private cloud offerings. "You should not confuse individual components which come from a single company with integration," he explained at EMC World. "Other reference architectures tell you how to build it. Just because you bought a reference architecture doesn't mean that you have integrated products. If you have a problem, you call VCE and we take ownership."

I got a chance to sit down with Eric Herzog, SVP for product management and marketing of the unified storage division at EMC, who was excited about flash storage and the possibilities that tiering offers organizations. By utilizing an average of 5% of a storage system as flash, the array can offer two to three times better performance using flash. Traditionally, adding more IOPS meant adding more disks to an array, and the progression was largely linear. Flash can substantially increase performance without adding more disks. That means less power draw, and less space to hold the drives. He says the biggest hurdle he sees is educating storage admins both in organizations and EMC's channel partners that even though flash drives are far more expensive per unit than disk drives, to achieve equivalent performance, judicious use of flash plus disk can significantly reduce the overall system cost because you need far fewer disks. I suspect you'll be hearing a lot more about flash from EMC in the coming year.

Two EMC Partners to Watch

Many of EMC's partners were on display at EMC World--two caught my eye. Infineta makes a WAN optimizer that it claims competes on data reduction with the likes of Riverbed and Silver Peak, using in-memory deduplication and at 10 Gb speeds. Built using merchant silicon and an FPGA, Infineta uses RAM to deduplicate blocks as small as 8 bytes and stores the dictionary in 64 GB RAM. Since block sizes are so small, representatives claim, they can deduplicate within the same packet, as well as across packets. Infineta is focusing primarily on replication traffic from backups and SRDF, rather than on file traffic. The company is in the process of getting its equipment qualified by EMC, which will open the door to being an approved component of the SAN. Its story sounds good, but, of course, I can't vouch for Infineta's performance claims. If your current WAN optimization isn't cutting it for your replication needs, Infineta is worth investigating (and tell me what you find!).

What's a tech show in 2012 without some software-defined networking (SDN) goodness? I happened upon Vello Systems, which makes an SDN controller and optical switch that underpins it Remote Data Replication Manager (RDRM). The manager applies network policies via OpenFlow to replication data between data centers and can support, for example, EMC's SRDF and RecoverPoint, as well as VMware's Site Recovery Manager. Vello's RDRM seems targeted toward cloud providers, but any organization running fiber plant could look at Vello for WAN traffic enforcement.

If you're still at EMC World, take advantage of EMC's Hands-on Labs. There are 27 labs available. They're pretty well done; I ran through the vCloud Director lab in about an hour. There were several others I wanted to complete, but other obligations distracted me.

All in all, EMC put together a good show for its customers and partners. There were a bunch of sessions given by EMC staff and partners, plenty of food and seating if you needed to catch your breath, and a show floor that, while small, showcased both EMC and partner products and services.

]]> http://www.networkcomputing.com/storage-networking-management/240000867/feed/ http://www.networkcomputing.com/data-networking-management/240000835 http://www.networkcomputing.com/data-networking-management/240000835#comments Tue, 22 May 2012 19:15:00 +0000 Greg Ferro Uncategorized http://www.networkcomputing.com/data-networking-management/240000835 Technical debt isn't just a term that can be used by overpaid management consultants in a wordy and overpriced analyst report. It's a valuable concept that engineers can use to explain a complex idea in a sound bite to management.

Modern IT infrastructures commonly ask us to do more with less. There are many ways to achieve this: delay a planned upgrade, or overload the memory on your VMware server just a bit more. Maybe another shot at reworking the quality-of-service configuration on the WAN link could delay an upgrade for a few more months? What about hanging onto those end-of-life firewalls for another year? What about skipping the maintenance contract on those load balancers that haven't had a problem in the last three years? Consider delaying that new hire? Or cancel the training budget?

In the right situation, all of these decisions might make perfect sense. Equally, each of these decisions incurs a debt--a very real, tangible technical debt.

Cancelling the training budget will result in both disappointed staff and skill reduction. Smart solutions are more likely to come from people who have the confidence to make decisions, which can be arrived at utilizing knowledge gained from training courses. Similarly, delaying a new hire increases stress on your team and affects morale in the long term.

Those end-of-life firewalls are likely to have security vulnerabilities, in addition to performance limits. Failing to replace them can result in performance bottlenecks, weird behavior and increased risk of a security breach.

Reworking the quality-of-service configuration is time-intensive and complex, as it's a tricky technical problem that's more art than science. Hundreds of hours of valuable BAU time can be poured into attempting to improve bandwidth. Is this the best use of an engineer's time?

Squeezing another server into the VMware leads to degraded performance and more time wasted. Delaying that project means productivity gains aren't realized.

All of these decisions lead to a technical debt. A debt that has to be paid back--eventually. It's not obvious that a delayed software upgrade will improve performance and remove bugs until you need that performance, or the device crashes because of known bugs.

As a network architect, a key part of my role is to recognize these debts and communicate them to management and program directors as a real but intangible cost. And that's hard. In a world where the bottom lines rules, spelling out the value of doing it right the first time doesn't sell, but I find that pitching the concept of technical debt forces consideration of operational cost and impact in addition to the simple, up-front capex.

It's great idea to have when you're facing the boss. Use it.

Hat tip to Bob Plankers for reminding me of the term last month. ]]> Technical debt isn't just a term that can be used by overpaid management consultants in a wordy and overpriced analyst report. It's a valuable concept that engineers can use to explain a complex idea in a sound bite to management.

Modern IT infrastructures commonly ask us to do more with less. There are many ways to achieve this: delay a planned upgrade, or overload the memory on your VMware server just a bit more. Maybe another shot at reworking the quality-of-service configuration on the WAN link could delay an upgrade for a few more months? What about hanging onto those end-of-life firewalls for another year? What about skipping the maintenance contract on those load balancers that haven't had a problem in the last three years? Consider delaying that new hire? Or cancel the training budget?

In the right situation, all of these decisions might make perfect sense. Equally, each of these decisions incurs a debt--a very real, tangible technical debt.

Cancelling the training budget will result in both disappointed staff and skill reduction. Smart solutions are more likely to come from people who have the confidence to make decisions, which can be arrived at utilizing knowledge gained from training courses. Similarly, delaying a new hire increases stress on your team and affects morale in the long term.

Those end-of-life firewalls are likely to have security vulnerabilities, in addition to performance limits. Failing to replace them can result in performance bottlenecks, weird behavior and increased risk of a security breach.

Reworking the quality-of-service configuration is time-intensive and complex, as it's a tricky technical problem that's more art than science. Hundreds of hours of valuable BAU time can be poured into attempting to improve bandwidth. Is this the best use of an engineer's time?

Squeezing another server into the VMware leads to degraded performance and more time wasted. Delaying that project means productivity gains aren't realized.

All of these decisions lead to a technical debt. A debt that has to be paid back--eventually. It's not obvious that a delayed software upgrade will improve performance and remove bugs until you need that performance, or the device crashes because of known bugs.

As a network architect, a key part of my role is to recognize these debts and communicate them to management and program directors as a real but intangible cost. And that's hard. In a world where the bottom lines rules, spelling out the value of doing it right the first time doesn't sell, but I find that pitching the concept of technical debt forces consideration of operational cost and impact in addition to the simple, up-front capex.

It's great idea to have when you're facing the boss. Use it.

Hat tip to Bob Plankers for reminding me of the term last month. ]]> http://www.networkcomputing.com/data-networking-management/240000835/feed/ http://www.networkcomputing.com/data-networking-management/240000809 http://www.networkcomputing.com/data-networking-management/240000809#comments Tue, 22 May 2012 15:34:00 +0000 Mike Fratto Uncategorized http://www.networkcomputing.com/data-networking-management/240000809 Brocade doesn't look at SDN and OpenFlow the same as other networking vendors. In its view of the network, the lowest tier is the network fabric, which is composed of the traditional L2/L3 switching and routing technologies in use today. Routing protocols like BGP and OSPF, forwarding technologies like spanning tree, TRILL and link aggregation work well and are well understood by network administrators. The data center bridging protocols make Ethernet lossless. On top of that, robust fabric is the network virtualization layer made up of overlay technologies such as VXLAN and NVGRE, which interconnect virtual machines via encapsulated tunnels, or even MPLS, which is often used in WANs but is making an entrance into large data centers, providing a much-needed network layer that suits virtualized data centers well. SDN sits within the network services layer and provides the network operator--which, for Brocade, means carriers, cloud providers and the largest enterprises--with the ability to apply value-added services on top of the network.

At the top layer are cloud environments like OpenStack, CloudStack, VMware vCloud and Microsoft's Systems Center. It's this top layer where the cloud architecture communicates with the network via APIs that allow features like traffic separation for multi-tenancy, traffic grooming and service-level agreement management to occur.

While many consider OpenFlow a network fabric layer protocol, Brocade positions it as a network service layer. That's because its customers, carriers and cloud providers are already using existing L2/L3 protocols to build out the majority of their networks, and are looking at OpenFlow to provide a differentiated service to their customers who need special treatment. Robust networks can be built using existing technologies, but they're still best-effort, and network issues affect all traffic. Using OpenFlow on specific ports, VLANs or MPLS trunks, Brocade lets providers carve up their networks for specific customers while letting the rest of the network run dynamically. This hybrid-switch mode also allows companies to deploy OpenFlow in increments, without affecting the entire network.

Brocade's position is just that: a position. What it means for potential customers is they can run their traditional network alongside an OpenFlow network and move traffic between networks without having to exit and enter the switch--OpenFlow becomes a switch feature. The SDN component is where the cloud layer can treat the network as a service, similar to storage or compute. Brocade is already moving toward enabling services at L4-7 via its ADX controller, which can manage traffic flows, using Brocade OpenScript, and send them to an OpenFlow or traditional network. The ADX integration is a critical component in an environment where there may be many applications residing behind a single address, and those applications require differentiated services based on the content.

Naturally, a hybrid traditional switch and OpenFlow switch can complicate troubleshooting and problem resolution. Brocade has built-in instrumentation in the CLI that allow administrators to view the OpenFlow flow tables, show interfaces and manage how flows are processed through the switch, similar to the tools used in traditional networking.

Brocade's claim of being the first vendor to provide OpenFlow in hybrid mode is inaccurate. NEC's ProgrammableFlow switches are hybrid-capable, as is HP's switches, though HP does not yet fully support OpenFlow 1.0, according to its latest release notes for switch software K.15.06.5008. However, supporting OpenFlow on its largest routers and switches is a huge feat. Since OpenFlow support doesn't require new Ethernet framing like TRILL or Brocade's VCS did, the software upgrade means Brocade can support the 100 Gb interfaces in its NetIron routers and switches. As new versions of OpenFlow are stabilized, a software update is all that will be needed to upgrade. ]]> Brocade doesn't look at SDN and OpenFlow the same as other networking vendors. In its view of the network, the lowest tier is the network fabric, which is composed of the traditional L2/L3 switching and routing technologies in use today. Routing protocols like BGP and OSPF, forwarding technologies like spanning tree, TRILL and link aggregation work well and are well understood by network administrators. The data center bridging protocols make Ethernet lossless. On top of that, robust fabric is the network virtualization layer made up of overlay technologies such as VXLAN and NVGRE, which interconnect virtual machines via encapsulated tunnels, or even MPLS, which is often used in WANs but is making an entrance into large data centers, providing a much-needed network layer that suits virtualized data centers well. SDN sits within the network services layer and provides the network operator--which, for Brocade, means carriers, cloud providers and the largest enterprises--with the ability to apply value-added services on top of the network.

At the top layer are cloud environments like OpenStack, CloudStack, VMware vCloud and Microsoft's Systems Center. It's this top layer where the cloud architecture communicates with the network via APIs that allow features like traffic separation for multi-tenancy, traffic grooming and service-level agreement management to occur.

While many consider OpenFlow a network fabric layer protocol, Brocade positions it as a network service layer. That's because its customers, carriers and cloud providers are already using existing L2/L3 protocols to build out the majority of their networks, and are looking at OpenFlow to provide a differentiated service to their customers who need special treatment. Robust networks can be built using existing technologies, but they're still best-effort, and network issues affect all traffic. Using OpenFlow on specific ports, VLANs or MPLS trunks, Brocade lets providers carve up their networks for specific customers while letting the rest of the network run dynamically. This hybrid-switch mode also allows companies to deploy OpenFlow in increments, without affecting the entire network.

Brocade's position is just that: a position. What it means for potential customers is they can run their traditional network alongside an OpenFlow network and move traffic between networks without having to exit and enter the switch--OpenFlow becomes a switch feature. The SDN component is where the cloud layer can treat the network as a service, similar to storage or compute. Brocade is already moving toward enabling services at L4-7 via its ADX controller, which can manage traffic flows, using Brocade OpenScript, and send them to an OpenFlow or traditional network. The ADX integration is a critical component in an environment where there may be many applications residing behind a single address, and those applications require differentiated services based on the content.

Naturally, a hybrid traditional switch and OpenFlow switch can complicate troubleshooting and problem resolution. Brocade has built-in instrumentation in the CLI that allow administrators to view the OpenFlow flow tables, show interfaces and manage how flows are processed through the switch, similar to the tools used in traditional networking.

Brocade's claim of being the first vendor to provide OpenFlow in hybrid mode is inaccurate. NEC's ProgrammableFlow switches are hybrid-capable, as is HP's switches, though HP does not yet fully support OpenFlow 1.0, according to its latest release notes for switch software K.15.06.5008. However, supporting OpenFlow on its largest routers and switches is a huge feat. Since OpenFlow support doesn't require new Ethernet framing like TRILL or Brocade's VCS did, the software upgrade means Brocade can support the 100 Gb interfaces in its NetIron routers and switches. As new versions of OpenFlow are stabilized, a software update is all that will be needed to upgrade. ]]> http://www.networkcomputing.com/data-networking-management/240000809/feed/ http://www.networkcomputing.com/data-networking-management/240000692 http://www.networkcomputing.com/data-networking-management/240000692#comments Mon, 21 May 2012 13:16:00 +0000 Jeff Loughridge Uncategorized http://www.networkcomputing.com/data-networking-management/240000692 This may sound familiar: Your business relies on a 15-year-old networking technology. A newer version or a functional replacement has been available for five years. The older version is beginning to show its age in its ability to meet technology and business requirements. You find yourself in a situation in which:

1. Engineers who understand the technology are more difficult to find, and thus more expensive.
2. Support costs are rising as the vendor focuses on newer technologies.
3. Network hardware is approaching end-of-life status.
4. The technology barely meets today's performance/throughput requirements.
5. Internal supports costs are rising.

I've worked with clients who see that the legacy networking technology should be ushered out. While they may recognize the problem, the dollars required to fund a transition effort are removed from the budget year after year. Why? The pain may not yet be acute, and other items in the budget may appear much more pressing. However, the delay means that once the migration does take place, it will be more costly and complex.

These companies may not have internal staff members who can make the case for migrating from older technology. The complexity of the technology migration effort may be so daunting that no one is willing to risk his upward mobility to advocate it, lest it exceed the budget and miss the targeted completion date.

In many cases, my advice to clients follows a networking law that I call the Law of Holes: When you're in a hole, stop digging. Sounds simple, right?

My law is based on a cap-and-grow strategy: Cap your spending on the old technology. Invest in the technology that represents the future. When an abrupt stop in spending isn't possible, severely restrict it, preferably backed by a mandate that any spending on the legacy technology must be approved by the CIO. This signals to the organization that leadership stands behind the technology migration effort and sees its success as crucial.

Next: The Cap-and-Grow PhilosophyFor cap and grow to be viable, the two iterations of the technology must be able to operate simultaneously. There is no requirement for a flag day in which a complete migration must be performed. Side-by-side operation could be enabled by a translation mechanism, a gateway that unites the two technologies, or a segmentation in which the technologies operate independently.

For example, enterprises have long relied on time division multiplexing (TDM) circuits for WAN connections. Fractional T1s, T1s, DS3s and OC12 circuits are examples of TDM circuits.

TDM WAN has numerous disadvantages: TDM provisioning is slow--the time to obtain or upgrade a TDM circuit is often measured in months. Upgrade increments aren't granular. The next upgrade for a DS3 circuit (45 Mbps) is an OC3 (155 Mbps). Using multiple DS3 may be an option for bandwidth in between 45 and 155 Mbps, although discrete circuits can have repercussions on load balancing.

While TDM may dominate your enterprise WAN circuits, Ethernet is the technology of choice in the LAN. The network builds in the last eight years have resulted in a sizeable Ethernet footprint for the U.S.'s top metropolitan areas. Ethernet is quickly emerging as the preferred WAN access technology. Ethernet cards for your WAN-facing routers are inexpensive. These circuits can be upgraded in small increments, and these upgrades usually involve only a few keystrokes by the provider. No more waiting months for bandwidth upgrades.

How might a cap-and-grow strategy work in an enterprise moving to Ethernet? For any upgrades to circuits to headquarters, data centers and other heavy-bandwidth sites, the business can switch to Ethernet access. TDM and Ethernet access circuits can be used simultaneously for WAN access, so there are no compatibility concerns. Branch sites can also move to Ethernet access in metropolitan areas, where the prices for Ethernet access have fallen rapidly.

I mentioned exceptions in my description of cap and grow. If you have branches with low-bandwidth needs, you may want to stick with TDM access, as the price-per-meg of Ethernet will likely exceed TDM for speeds under 6 Mbps. Don't forget to evaluate business cable service. Many of my clients are very satisfied with a business cable service. The handoff is Ethernet--just as it is in your home for cable Internet--so it can be very similar to Ethernet access from a provider such as the local exchange carrier.

My Law of Holes has many applications in IT--think IPv4 to IPv6, TDM telephony to VoIP and traditional IT operation to cloud computing. If you recognize that a migration is needed, advocate for a technology migration. Show your organization why it's time to stop digging. ]]> This may sound familiar: Your business relies on a 15-year-old networking technology. A newer version or a functional replacement has been available for five years. The older version is beginning to show its age in its ability to meet technology and business requirements. You find yourself in a situation in which:

1. Engineers who understand the technology are more difficult to find, and thus more expensive.
2. Support costs are rising as the vendor focuses on newer technologies.
3. Network hardware is approaching end-of-life status.
4. The technology barely meets today's performance/throughput requirements.
5. Internal supports costs are rising.

I've worked with clients who see that the legacy networking technology should be ushered out. While they may recognize the problem, the dollars required to fund a transition effort are removed from the budget year after year. Why? The pain may not yet be acute, and other items in the budget may appear much more pressing. However, the delay means that once the migration does take place, it will be more costly and complex.

These companies may not have internal staff members who can make the case for migrating from older technology. The complexity of the technology migration effort may be so daunting that no one is willing to risk his upward mobility to advocate it, lest it exceed the budget and miss the targeted completion date.

In many cases, my advice to clients follows a networking law that I call the Law of Holes: When you're in a hole, stop digging. Sounds simple, right?

My law is based on a cap-and-grow strategy: Cap your spending on the old technology. Invest in the technology that represents the future. When an abrupt stop in spending isn't possible, severely restrict it, preferably backed by a mandate that any spending on the legacy technology must be approved by the CIO. This signals to the organization that leadership stands behind the technology migration effort and sees its success as crucial.

Next: The Cap-and-Grow PhilosophyFor cap and grow to be viable, the two iterations of the technology must be able to operate simultaneously. There is no requirement for a flag day in which a complete migration must be performed. Side-by-side operation could be enabled by a translation mechanism, a gateway that unites the two technologies, or a segmentation in which the technologies operate independently.

For example, enterprises have long relied on time division multiplexing (TDM) circuits for WAN connections. Fractional T1s, T1s, DS3s and OC12 circuits are examples of TDM circuits.

TDM WAN has numerous disadvantages: TDM provisioning is slow--the time to obtain or upgrade a TDM circuit is often measured in months. Upgrade increments aren't granular. The next upgrade for a DS3 circuit (45 Mbps) is an OC3 (155 Mbps). Using multiple DS3 may be an option for bandwidth in between 45 and 155 Mbps, although discrete circuits can have repercussions on load balancing.

While TDM may dominate your enterprise WAN circuits, Ethernet is the technology of choice in the LAN. The network builds in the last eight years have resulted in a sizeable Ethernet footprint for the U.S.'s top metropolitan areas. Ethernet is quickly emerging as the preferred WAN access technology. Ethernet cards for your WAN-facing routers are inexpensive. These circuits can be upgraded in small increments, and these upgrades usually involve only a few keystrokes by the provider. No more waiting months for bandwidth upgrades.

How might a cap-and-grow strategy work in an enterprise moving to Ethernet? For any upgrades to circuits to headquarters, data centers and other heavy-bandwidth sites, the business can switch to Ethernet access. TDM and Ethernet access circuits can be used simultaneously for WAN access, so there are no compatibility concerns. Branch sites can also move to Ethernet access in metropolitan areas, where the prices for Ethernet access have fallen rapidly.

I mentioned exceptions in my description of cap and grow. If you have branches with low-bandwidth needs, you may want to stick with TDM access, as the price-per-meg of Ethernet will likely exceed TDM for speeds under 6 Mbps. Don't forget to evaluate business cable service. Many of my clients are very satisfied with a business cable service. The handoff is Ethernet--just as it is in your home for cable Internet--so it can be very similar to Ethernet access from a provider such as the local exchange carrier.

My Law of Holes has many applications in IT--think IPv4 to IPv6, TDM telephony to VoIP and traditional IT operation to cloud computing. If you recognize that a migration is needed, advocate for a technology migration. Show your organization why it's time to stop digging. ]]> http://www.networkcomputing.com/data-networking-management/240000692/feed/