Converged networks with Fibre Channel over Ethernet and Data Center Bridging

ManualsBrandsHP ManualsComputer AccessoriesHP Virtual Connect FlexFabric-20/40 F8 Module for c-Class BladeSystem

Converged networks with Fibre Channel

over Ethernet and Data Center Bridging

Technology brief, 2

edition

Introduction ......................................................................................................................................... 2

Traditional data center topology ............................................................................................................ 2

Early attempts at converged networks ..................................................................................................... 2

Network convergence with FCoE ........................................................................................................... 3

10 Gigabit Ethernet ............................................................................................................................. 4

HP Virtual Connect Flex-10 ................................................................................................................ 4

HP Virtual Connect FlexFabric with FCoE ............................................................................................ 5

Emerging standards for network convergence ......................................................................................... 5

FCoE standard ................................................................................................................................. 5

FCoE protocol encapsulation.......................................................................................................... 5

Fibre Channel Forwarder ............................................................................................................... 6

ENode ........................................................................................................................................ 6

FCoE and Ethernet............................................................................................................................ 7

DCB standards ................................................................................................................................. 8

Priority-based Flow Control ............................................................................................................ 8

Enhanced Transmission Selection .................................................................................................. 10

Quantized Congestion Notification ............................................................................................... 12

Data Center Bridging Exchange ................................................................................................... 14

Migrating to converged fabrics............................................................................................................ 17

HP strategy ....................................................................................................................................... 19

For more information .......................................................................................................................... 20

Call to action .................................................................................................................................... 20

Summary of content (20 pages)

PAGE 1
Converged networks with Fibre Channel over Ethernet and Data Center Bridging Technology brief, 2nd edition Introduction ......................................................................................................................................... 2 Traditional data center topology ............................................................................................................ 2 Early attempts at converged networks .....................................................................
PAGE 2
Introduction Using application-specific networks for data, management, and storage is complex and costly. Network convergence is a more economical solution that simplifies your data center management by partially or completely consolidating all block-based storage and Ethernet-based data communications networks onto a single fabric. Any network topology constructed with one or more switched network nodes is a fabric. Converged networks consolidate two or more network types onto a single fabric.
PAGE 3
IB-centric protocols/networks to the native FC storage devices in data centers. High Performance Computing (HPC) environments that have adopted IB as the standard transport network use SRP and iSER protocols. Figure 1.
PAGE 4
and CEE equivalent terms. We use the term DCB to refer to an Ethernet infrastructure that implements at least the minimum set of DCB standards to carry FCoE protocols. A traffic class (TC) is a traffic management element. DCB enhances low-level Ethernet protocols to send different traffic classes to their appropriate destinations. It also supports lossless behavior for selected TCs, for example, those that carry block storage data. FCoE with DCB tries to mimic the lightweight nature of native FC protocols.
PAGE 5
HP Virtual Connect FlexFabric with FCoE Now that we have achieved an acceptable level of LAN convergence with Flex-10 technology, the next logical step is to add LAN/SAN convergence technology. Virtual Connect FlexFabric broadens Virtual Connect Flex-10 technology to provide solutions for converging different network protocols. We plan to deliver the FlexFabric vision by converging technology, management tools, and partner product portfolios into a virtualized fabric for the data center.
PAGE 6
Figure 3. Illustration of an FCoE frame Header Start of Frame End of Frame Header Payload Frame Check Layer 2 encapsulation provides several advantages to FCoE over previous converged network implementations: • Because devices use existing FC logic, FCoE devices use existing FC driver models for the new converged network adapters. • We can easily implement FCoE in switches because the logic necessary to convert between FCoE and FC is simple.
PAGE 7
The ENode (Figure 4) consists of these components: • FCoE Controller uses FCoE Initialization Protocol (FIP) to discover the SAN fabrics through the FCFs and provisions the virtual N_Ports (VN_Ports) and FCoE Link End Points (LEPs). • FCoE LEPs convert FC frames to FCoE frames on the transmit side, and convert FCoE frames to FC frames on the receive side. There is one LEP for each VN_Port established in the ENode.
PAGE 8
TCs, the existing mechanisms can cause Quality of Service (QoS) issues, limit the ability to scale a network, and affect performance. DCB standards DCB is not just the name for a set of new standards the IEEE is developing. It is a term often used for Ethernet designed to carry multiple TCs, some with lossless behavior. You can think of DCB-enabled Ethernet as applying the DCB standards to IEEE 802.3 Ethernet standards to create a new set of products to implement this improved version of Ethernet.
PAGE 9
frames of varying priority or TCs: If one queue becomes full, the device must send a pause frame to the other side of the link. This pauses all traffic, regardless of TC/priority. Supporting lossless behavior of block storage protocols on legacy Ethernet networks requires using legacy pause frames. However, this forces all traffic to be lossless on that link. The most bursty or bandwidth driven TCs dictate the behaviors of all TCs.
PAGE 10
Figure 6 illustrates the same scenario up to the point where the receiving node needs to send a pause frame. A PPP frame dictates pausing the red TC. The pause takes advantage of the class of service fields to restrict the pause to only classes of traffic that have nearly full queues. The transmitting station stops sending red traffic; the latency-sensitive green traffic continues to flow properly. Figure 6. PFC-based flow control Receive queues in a DCB Ethernet device will have high and low watermarks.
PAGE 11
for all but the very highest priority traffic classes. This includes converged networks that handle block storage traffic using a layer 2 encapsulation protocol, like FCoE. One common misunderstanding about many modern Ethernet devices, particularly Ethernet switches, is that they already have bandwidth control and traffic shaping capabilities that support layer 2 protocols like FCoE.
PAGE 12
Figure 7. Example of an Enhanced Transmission Selection (ETS) configuration Also in Figure 7, note that TC2 has priority queues 2 and 3. The ETS standard suggests that frames transmit from TC2 queues in strict priority order. In this example, the device sends any frames in the queue for priority 3 before any frames in the queue for priority 2. Again, the standard leaves the implementation of scheduling for these intra-TC queues to device vendors.
PAGE 13
use QCN in conjunction with PFC to provide lossless operation with smooth congestion management across large DCB-enabled networks. QCN uses a special new tag that allows sources of traffic, for example CNAs, to identify traffic flows to all interconnect devices in a QCN-enabled DCB network. QCN defines two specific points in a network that implement the QCN protocol, congestion points and reaction points.
PAGE 14
notification messages to the server CNAs. The switch selects the server CAN by statistically sampling the congested queue. The congestion notification occurs dynamically by sending higher feedback quanta to CNAs producing the most traffic and lower feedback quanta to sources producing less traffic. As a result, the CNAs throttle down their transmit rates on congested traffic flows.
PAGE 15
NOTE Link Layer Discovery Protocol (LLDP), IEEE 802.1AB, defines a protocol and a set of managed objects that can be used for discovering the physical topology and connection end-point information from adjacent devices in 802 LANs and MANs. The protocol is not restricted from running on non-802 media. Table 2.
PAGE 16
Figure 9. DCBX static parameter exchange CNA parameters switch parameters X The DCBX protocol’s strength lies in its ability to perform dynamic negotiation using attributes called recommended and willingness. CNAs and switches using DCBX can advertise their willingness to adopt parameter settings from their link partner. In the example shown in Figure 10, a CNA communicates the initial exchange of ETS and PFC information and willingness to consider parameters from the switch.
PAGE 17
Figure 10. DCBX dynamic negotiation X CNA willingness Switch recommended DCB - C CNA new parameters Migrating to converged fabrics In a one-hop architecture, converged traffic goes from a server to a switch that splits it to Ethernet and Fibre Channel. In two-hop architecture, converged traffic goes to a second switch before the split. The more switch hops in a DCB-enabled network, the more difficult it is to keep the network operating at peak efficiency while minimizing congestion.
PAGE 18
This is the first phase of migration to converged fabrics. CNAs will connect to converged fabric access switches that support DCB-enabled Ethernet, legacy Ethernet, and legacy FC. The CNAs will provide converged connectivity between servers and the first hop switch before disaggregating the traffic to the legacy LAN and SAN infrastructure. Figure 12 compares traditional deployment to the first phase of converged network deployment. Figure 12.
PAGE 19
Figure 13. Converged network deployment, phases 2 and 3 HP strategy We believe that the transition to DCB/FCoE can be graceful. It need not disrupt existing network infrastructures if you first deploy at the server-to-network edge and then migrate farther into the network. With this approach, you will gain the immediate benefit of reduced cable and adapter hardware with the least amount of disruption to the overall network architecture.
PAGE 20
For more information Resource description Web address HP Multifunction Networking Products http://h18004.www1.hp.com/products/servers/proliantadvantage/networking.html HP ProLiant networking http://h18004.www1.hp.com/products/servers/networking/in dex-nic.html Ethernet network adapters “Server-to-network edge technologies: converged networks and virtual I/O” technology brief http://h20000.www2.hp.com/bc/docs/support/SupportManu al/c02044591/c02044591.