Data Center Fibre Channel network architecture increasingly popular

Http://www.aliyun.com/zixun/aggregation/10145.html "> Enterprise Architecture Senior Manager Brenden Bryan said in the Government Employee Health Association (GEHA) Once the data center is completely tuned and the Fibre Channel network infrastructure infrastructure is deployed, the program becomes "Really Simple" and is no longer as complex as many other IT projects. "When I look back at the whole process and say ' I want to make a different decision ', we don't have any problems or troubles. ”

Geha, located in Missouri State Kansas City, is the second-largest health plan and dental Health program in the United States, which is dealing with reimbursement for medical expenses for federal employees, retirees and their family members. Bryan said the main reason for moving to the Fibre Channel network architecture was simplifying, consolidating, and removing legacy Fibre Channel SAN environments.

Bryan began working for Geha in August 2010. He said he was taking over a fairly typical infrastructure: a patchwork of parts produced by many manufacturers and with lots of faults. The association also wants to virtualize its mainframe environment and turn it into a distributed architecture. "What I need is an infrastructure with redundancy and a high availability," Bryan said. "Once the new infrastructure is in place and running steadily, the next step is to move all the Geha Tier 2 and Tier 3 applications to this, and eventually the Tier 1 reimbursement processing system," he said.

Geha has deployed Ethernet switches and routers produced by Brocade. It is now over a year since the end date of the six-month project. He says they provide a high-speed environment and a ratio of 20:1 virtual machines to blade server hardware.

"I can keep the minimum number of physical servers and use them as much as I can," says Bryan. They allow me to improve storage efficiency and computational efficiency. ”

Deploying a data center Fibre Channel Network architecture requires some planning. This means upgrading and replacing older switches with new switch devices because of the use of different traffic configurations in Fibre Channel networks, says Zeus Kerravala, chief analyst at ZK. "You have to rebuild your network and reconnect to the server," he said. ”

To achieve flat

Data Center Fibre Channel Network architecture is a network optimized for horizontal information flow, which is more flat and simpler than traditional networks. Kerravala explained that it is more aimed at client/server settings designed to achieve the server and network core implementation of information exchange.

In Fibre Channel network mode, traffic flows horizontally between the network and the virtual machine. "So it's more of a server connection to the server. "Fibre Channel networks are more flat and no more than two layers. By contrast, legacy networks have at least three or even more layers. Kerravala said that in the past years, the storage network has been used in this way, and now the data network needs to adopt this design method.

One factor driving the transition to Fibre Channel networks is the virtualization of half of the enterprise data center workloads in companies on the FTSE 2000 rankings. Once the data center load was implemented for virtualization, they began to focus on how to make their servers communicate with other servers and the network.

"We see this as a development in the Datacenter network architecture," said Bob Laliberte, senior analyst at Enterprise Storage Group. The driving factor is that more and more servers are connected to the server. These different servers need to talk to each other and communicate with the core of the network. These increase processing load and latency. ”

Virtualization adds complexity because it means moving a lot of things dynamically, "so network vendors are trying to simplify these complex environments." ”

When the data center cannot be extended

As foreclosures peaked in 2006, the Walz group found that their data centers could not be expanded to respond effectively to additional service growth requirements. The company is responsible for providing document management, compliance and compliance services for multiple industries. "This is hampering business growth," said Bart Falzarano, the company's chief information security officer. ”

Falzarano says the company has recruited additional IT staff to cope with these different systems and management practices, build new servers, expand networks, and increase disaster recovery services. "But managing these technical problems is very difficult, especially as we begin to try to move to a virtual environment." "he said. The company has many applications that cannot be virtualized, which must be managed separately. "These systems, storage, and networks have different entry points. We have become unable to achieve the desired goal. ”

To reduce complexity, the Walz group deployed Cisco's Unified Data Center platform in 2009, a unified Datacenter Fibre Channel architecture that integrates computing, storage, networking, and management into a single platform to automate it and make it a service that spans both physical and virtual environments. The platform is now connected to the NetApp SAN Storage flexpod platform in the company.

Falzarano recalls that prior to their use of HP's technology, when a node of the database failure, you need to call the manufacturer, the final need to take out four CPU three, and it will take four hours to troubleshoot. Until the return to normal operation, 14 hours has passed.

"Now, for the same fault, if we find a problem Blade server node, we can disconnect the SQL application and reconnect the SQL application in 4 minutes." This can also be done at the virtual management level. "he said.

It has been tracking data center performance and documenting key metrics. Falzarano said they found the density to be 8:1, which means less cabling complexity and lower cable demand. The virtualization efficiency of old technology is 4:1, the new technology has higher efficiency, reach 15:1. Teams can virtualize applications that they previously could not virtualize.

Other advantages include a 50% reduction in racks and more centralization of system management--an IT engineer can now manage 50 systems.

"We encountered a large number of hardware failure problems when using old technology." In our multiple data centers, a failure occurs every 30 days. There may not be a single failure in a year. "he said.