以太网发展迎新里程碑，分析师眼中的未来之路

预计6月17日将是以太网的发展史上新的重要里程碑。因为在这一天，40Gb/s和100Gb/s的以太网标准将正式批准生效，从而结束了业界为将以太网的网速突破最高10Gb/s长达四年半的艰苦路程。

正如有些人可能知道的那样，这一旅程在刚开始时面临着困境，因为当时的“研究小组”必须决定是否要将40Gb/s的数据传送速度添加为项目目标。从外往里看，有很多人质疑，标准的进程是否有问题，而不是考虑这一争论实际反映了当时行业内正在关注的话题。

回过头看，很多人对10Gb以太网推出后的这四年感到很失望，在此期间甚至没有销售出100万端口。相反，这些人盯着在相同长的时间内1Gb以太网发货近2亿端口。这个议题变成，10Gb以太网的成功是通过它的价值或是发货量来判断，而没有看到10Gb以太网在实现可升级网络中发挥的作用——使得约2亿的1Gb以太网端口被售出。

Chris Rowen博士

所以，当我们在临近40Gb以太网/100Gb以太网标准的预期正式批准时，服务器行业正在考虑向基于10Gb以太网的服务器转移。随着聚合引擎将实现几百万台基于10Gb以太网服务器的出货以及允许10Gb以太网依靠前几代以太网的打下的基础而存活，转向40Gb和100Gb以太网看似合理。

这将要求以太网新速度经历开发的一般步骤，如OIF(光互联论坛)推出的25Gb/s通信标准，这将导致更高的密度和更低的成本解决方案。请考虑一下：10Gb以太网标准在首次推出后，用了近五年的时间进入研发其它基于10Gb以太网的解决方案。

这将需要对行业某些部分进行大笔投资。但是，这笔投资也将构成推动下一次速率增长产生的基石。这一主题目在由ITU-T和IEEE主持的研讨会的某个小组会议上讨论过。终端用户(谷歌、Verizon和阿姆斯特丹互联网交换中心)、系统供应商和部件供应商聚在一起讨论市场需要的下一代网速，同时考虑400Gb以太网或者兆兆以太网(Terabit Ethernet)的技术和经济可行性。这真的是很有意思的讨论！因此，尽管没有启动正式的项目，这些论坛为进行这些讨论提供了条件，最终实现“建立共识”。

当然，如果大家老是相安无事，人生还有什么乐趣？旅行的意义在于留下美好回忆，而不是到达目的地。

John D'Ambrosia是Force10 Networks公司CTO办公室以太网标准的主管。自1999年以来，他一直在积极参与以太网相关技术的开发。

参考原文：Viewpoint: Divining Ethernet's future

《电子工程专辑》网站版权所有，谢绝转载

{pagination}

Viewpoint: Divining Ethernet's future

The next major milestone in the history of Ethernet is anticipated for June 17. This is the date that the expected ratification of the 40Gb/s and 100 Gb/s Ethernet standard is set to occur, thus ending the industry’s four-and-a-half year journey to move Ethernet beyond its current highest speed of 10Gb/s.

As some may know, early on in that journey the effort faced a serious dilemma, as the then- "Study Group" had to decide whether it wanted to add the 40 Gb/s date rate to the project objectives. From the outside looking in, many questioned if there was an issue with the standards process, rather than considering that the debate actually reflected discussions that were occurring within the industry.

Looking back, many were disappointed that four years after the 10 Gigabit Ethernet was released, it had not yet shipped a million ports. Instead, these individuals looked at the near 200 million ports that Gigabit Ethernet had shipped in that same time frame. The conversation became one where the success of 10 Gigabit Ethernet was judged by individuals either by its value or the volume shipped with some losing sight of the role that 10 Gigabit Ethernet played in providing scalable networks that enabled the shipment of approximate 200 million ports of Gigabit Ethernet.

So as we approach the anticipated ratification of the 40 Gigabit Ethernet / 100 Gigabit Ethernet standard, the server industry is looking to the migration to 10 Gigabit Ethernet-based servers. It would seem reasonable to look at 40GbE and 100 GbE as the aggregation engine that will enable the multi-million shipment volumes for 10GbE based servers, and allow 10GbE to live up to the legacy established by prior generations of Ethernet.

This will require that the new speeds of Ethernet undergo the usual steps of development, such as the 25Gb/s signaling efforts within the OIF, that will lead to higher density and lower cost solutions. Consider the roughly five years of development work after the initial release of the 10 Gigabit Ethernet standard that went into the creation of additional 10GbE based solutions.

This will require a significant investment on the part of the industry. But that investment will also yield the building blocks that will drive the development of the next rate hike. This very topic was the discussion of a panel session at a workshop sponsored by the ITU-T and the IEEE. End users (Google, Verizon, and the Amsterdam Internet Exchange), system vendors and component vendors got together to discuss the next speed from a market need while considering the technical and economic feasibility of such a feat. 400 Gigabit Ethernet? Terabit Ethernet? Interesting discussions indeed! So while there is no official project underway, these forums provide a place for these discussions to happen, which ultimately is the enabler to "consensus building."

Of course, where would the fun be in life, if we always agreed with each other? The journey is what is memorable, not the arrival.

John D'Ambrosia is a director of Ethernet-based standards, CTO office, at Force10 Networks. He has been an active participant in the development of Ethernet-related technologies since 1999.