No leader in data center switch industry

Despite years of innovation and millions of dollars in investment, no network vendor has established a leadership position in the data center switch fabric market.

Nearly five years, almost every manufacturer has released a new generation of data center switch and fabric. Not only do these switches have faster bandwidth, lower latency, and higher port densities, they also have a large number of data center architectures that are technologically innovative. Don't forget, back in the 2011, data center fabric is fairly well known in network technology before the software definition network subversion industry.

The data center FABRICDE is designed to allow organizations to deploy arbitrary, arbitrarily connected, large, two-tier networks to server traffic in a highly virtualized data center or cloud. Cisco has fabricpath based on Trill on its Nexus data center switch; Brocade has a trill based VCs on his vdx switch; Avaya SPB into his vena architecture; Juniper go farther and provide qfabric, he can compress the entire network into a layer, so that two-tier multipath is no longer important.

All of these technologies have adherents, but no one has won the market. Cisco is no exception. Two leading analysis companies in recent months have released a market assessment of the data center network industry--gartner's Magic Quadrant and the Forrester Research wave. No analysis has put the manufacturer in the lead position. This is a rare situation, even if you feel that these market evaluations are hype.

Juniper was frustrated by the late adoption of qfabric and frustration at the market for its high expectations. Cisco may have sold thousands of Nexus switches, but how many customers have activated the Fabricpath serial number? I have communicated with many Nexus clients, but none of them told me that they used fabricpath in production. If Cisco can get the fabric approved as soon as possible, I think I should have talked to a lot of people who support Fabricpath.

I've talked to some clients who used VCs, qfabric, brocade and juniper, but these vendors have long known that not everyone is ready for the next generation of data center networks in the marketplace. All two companies have built parks and data center switch product lines, and now they are out.

Juniper New EX9200 To some extent admits that not everyone will adopt Qfabric,juniper will ex9200--a repackaged MX router-positioned in the park and data center core platform. Brocade's more success came from the VCs and VDX switches in the data center, which recently released a new icx 6650 switch. Brocade's icx switch is part of the Enterprise Park product, but Brocade also recommends using 1.6Tbps 6650 as a top or terminal switch in the Non-fabric data center.

Even Cisco's successful Nexus products have not been able to drive old catalyst 6500 away from customer data centers. I've communicated with network engineers who are still using the classic Catalyst 6500 (not the newer Catalyst 6500-E) in the data center, and they don't intend to withdraw catalyst 6500 in the near future. These users are satisfied with the traditional data center network built in previous years.

Now the industry needs to deal with Sdn. Many SDN vendors are committed to solving problems that were originally resolved by the data center fabrics. Many network architects are not quite sure how fabrics and SDN are going to solve-are they buying this or that?

I don't think the manufacturer has missed the chance of fabric. Many network architects will consider using data center fabrics in the coming years. But all manufacturers have to prove something. The industry lows shown by Forrester and Gartner are not accidental. Each vendor must ensure that its technology meets the needs of its customers, and each vendor will prove to the customer that its technology is best.

What do you mean, better? solution needs to really address highly virtualized data center issues (low latency, east-west traffic) to be compatible with a Third-party platform (cloud architecture, management program, security); it wants to simplify the operation; it wants to get good support from vendors because it's a new technology; it's available. In recent years, some vendors have produced core switches with code vulnerabilities. Finally, these vendors need to explain how fabrics and snd exist together in the datacenter. A lot of people are already starting to do this, and there will be more discussion in the future, especially SDN's real potential is unclear.

At the same time, manufacturers need to realize that a shoe is difficult to fit all feet. Some network technicians do not need Nexus7000 or complete qfabric deployments. They prefer to stick with the catalyst 6500s or ex switch, which is fine for businesses that only need to keep exchange and SharePoint running properly.

(Responsible editor: Fumingli)