The various "G"s in the data centers of enterprises and cloud-level networks abound, and it is time to figure out their performance.
The rise of cloud computing and the expansion of data centers are driving the latest Ethernet speed upgrades, and cloud-based big data has increased the workload of operators. To meet this need, data centers are scaled by adding bandwidth capabilities in parallel with existing infrastructure. The rapid growth of 25G and 100G Ethernet deployments is expected to be evidence of this trend.
In order to handle the ever-increasing data load, the industry's largest long-haul cloud enterprises have joined the data center operators of their core networks to move to the 100G Ethernet architecture. However, most operators believe that 100G or even 40G is a bit excessive for server connections, because its workload only needs to be gradually improved on 10G networks. That's why although 40G and 100G Ethernet have been introduced, 25G and 50G Ethernet are still one of the reasons for the common choice within the data center. Below we will briefly explain why 25G is more suitable for these applications than 40G.
Several of the latest Ethernet bandwidth technologies are not designed to set a new high speed, but more to push such network protocols into neighboring markets, especially the data center market. Below we will introduce the specific reasons by introducing 25G, 50G and 100G respectively.
25G
The official IEEE 802.3 draft standard for 25G Ethernet will eventually be completed in 2016 (see Figure 1), which is targeted at servers in cloud data centers. This is a relatively short timeframe due to the reusability of 10G and 100G Ethernet components.
Figure 1: Development plan for the Ethernet standard
40G and 100G already exist, but why use 25G? This is confusing for some operators. The answer lies in the architectural and performance requirements. The existing 100G standard network system consists of four links, each with a bandwidth of 25 Gbps. This four-to-one ratio is equivalent to connecting servers to 25G switches and then merging them into 100G uplinks, helping network operators to more easily expand their data centers.
Similarly, 40G Ethernet is composed of four 10G Ethernet links. However, according to Ethernet Alliance President John D'Ambrosia, many data centers have adopted servers with more than 10G. That's why multiple chip vendors have offered 25G serial/deserializers. This not only makes bandwidth aggregation for 25G, 50G, and 100G Ethernet more convenient, but also reduces costs due to volume.
50G
Although there is still a period of time to implement the IEEE standard for 50G Ethernet (approximately 2018 to 2020), many industry alliances expect products to begin appearing in 2016. Similar to 25G technology, 50G Ethernet technology will be the next high-speed solution for connecting servers and data centers. According to data from analyst firm Dell'Oro, servers and high-performance flash storage systems will need more than 25G in the next few years.
To help deliver these accelerated Ethernet technologies faster, the 25G/50G Ethernet Alliance has eliminated the royalty for 25G and 50G Ethernet specifications and is open to all data center ecosystem vendors.
Reusing the 25G components of the existing 100G network can reduce the implementation cost of 50G. For example, the cost structure of 25G wiring is the same as that of 10G, but its performance is 2.5 times. Similarly, the cost of 50G is half the cost of 40G, but the performance can be increased by 25%.
100G
For long-distance carrier networks ranging from hundreds of kilometers to tens of thousands of kilometers, the deployment of 100G Ethernet will continue to grow.
But according to information from a new industry consortium, the 100G architecture will be another excellent market alternative. The 100GCLR4 Alliance, led by Intel and AristaNetwork, believes that 100G is ideal for connecting to large "very large" data centers ranging from 100 meters to 2 kilometers.
Other companies are also looking for alternative 100G implementations for data centers. Avago Technologies has joined the CWDM4MSA Industry Alliance, which is designed to define a common specification for low-cost 100G optical interfaces within 2 km of data center applications. As network infrastructures transition to 100G data rates, data centers will require long-haul, high-density 100G embedded optical connections. The MSA provides four 25G single mode fiber (SMF) link channels using coarse wavelength division multiplexing (CWDM) technology. Similarly, the OpenOpTIcsMSA organization sponsored by Ranovus and Mellanox Technologies will also focus on developing data centers that support 2km and 100G.
In the past, the increase in speed drove the development of most network components. Today, dealing with massive amounts of data flowing through the cloud requires companies to find a balance between speed and reuse technology to find a cost-effective solution.
Xinxiang Mina Import & Export Co., Ltd. , https://www.mina-motor.cn