Cisco Cisco MDS 9500 Series Supervisor-2 Module Libro blanco
© 2006 Cisco Systems, Inc. All rights reserved.
Important notices, privacy statements, and trademarks of Cisco Systems, Inc. can be found on cisco.com.
Page 9 of 16
Fibre Channel over IP
Over short distances, such as within a data center, SANs are typically extended over optical links with multimode optical fiber. As the distance
increases, such as within a large data center or campus, single-mode fiber or single-mode fiber with coarse wavelength-division multiplexing
(CWDM) is typical. Over metropolitan distances Dense Wave Division Multipexing (DWDM) is preferable. DWDM is also used where higher
consolidation density or aggregation of FICON, Enterprise Systems Connection (ESCON), and 10-Gigabit Ethernet data center links is required.
In contrast, Fibre Channel over IP (FCIP) can be used to extend a Fibre Channel SAN across any distance. FCIP can be used over metro and campus
distances or over intercontinental distances where IP might be the only transport available.
SAN extension using FCIP typically has many cost benefits over other SAN extension technologies. It is relatively common to have existing IP
infrastructure between data centers that can be leveraged at no incremental cost. Additionally, IP connectivity is typically available at a better price
point for long-distance links compared to pricing for optical Fibre Channel transport services.
FCIP is a means of providing a SAN extension over an IP infrastructure, enabling storage applications such as asynchronous data replication, remote
tape vaulting, and host initiator to remote pooled storage to be deployed irrespective of latency and distance. FCIP tunnels Fibre Channel frames
over an IP link, using TCP to provide a reliable transport stream with a guarantee of in-order delivery.
Cisco MDS 9000 Family switches offer a number of enhancements on top of standards-based FCIP to improve the functionality and usability
enabled by FCIP:
•
IVR is included with IP SAN extension functionality. IVR enables IP SAN extension without compromising fabric stability and reliability by
allowing for routing between SAN extensions and sites without the need to create a common merged fabric.
•
Fibre Channel traffic can be very bursty, and traditional TCP can amplify that burstiness. With traditional TCP, the network must absorb these
bursts through buffering in switches and routers. Packet drops occur when there is insufficient buffering at these intermediate points. To reduce
the probability of drops, Cisco MDS 9000 switches use traffic shaping to reduce the burstiness of the TCP traffic leaving the Gigabit Ethernet
interfaces. This is enabled through the use of a variable-rate, per-flow shaping and by controlling the TCP congestion window size.
the probability of drops, Cisco MDS 9000 switches use traffic shaping to reduce the burstiness of the TCP traffic leaving the Gigabit Ethernet
interfaces. This is enabled through the use of a variable-rate, per-flow shaping and by controlling the TCP congestion window size.
•
Various enhancements to TCP based on RFC1323, including modifications to TCP slow start and TCP retransmissions, are used within the Cisco
MDS to provide higher levels of throughput to FCIP than would otherwise be possible using standard TCP.
•
Compression (hardware based on MPS-14/2 module, software based on IPS-4/8 modules) enables higher traffic levels to be sustained through
the WAN. Compression utilizes standards-based LZS and deflate algorithms, typically achieving compression ratios of 2:1 to 3:1 with a real-
world traffic mix (as measured using industry-standard tests such as Canterbury Corpus) and up to 30:1 compression on very compressible data.
Hardware compression supports up to 150 MBps (~1400 Mbps) of compressed throughput per port; software compression is available for up to
300 Mbps of compressed throughput per port. Higher aggregate throughput can be achieved by load-balancing traffic across multiple FCIP
tunnels and Gigabit Ethernet interfaces port channeled together.
world traffic mix (as measured using industry-standard tests such as Canterbury Corpus) and up to 30:1 compression on very compressible data.
Hardware compression supports up to 150 MBps (~1400 Mbps) of compressed throughput per port; software compression is available for up to
300 Mbps of compressed throughput per port. Higher aggregate throughput can be achieved by load-balancing traffic across multiple FCIP
tunnels and Gigabit Ethernet interfaces port channeled together.
•
FCIP Write Acceleration (FCIP-WA) is a SCSI protocol spoofing mechanism designed to improve application performance by reducing
the overall service time for SCSI write I/Os and replicated write I/Os over distance. Most SCSI FCIP write I/O exchanges consist of two or
more round trips between the host initiator and the remote target array or tape. With FCIP-WA, multiple round trips are eliminated so that there
is one round trip per SCSI FCIP write I/O operation. This means that write operations are typically accelerated with FCIP-WA by a factor of 2:1,
halving the I/O service time associated with synchronous and asynchronous replication or approximately doubling the distance between data
centers for the same total actual I/O latency.
more round trips between the host initiator and the remote target array or tape. With FCIP-WA, multiple round trips are eliminated so that there
is one round trip per SCSI FCIP write I/O operation. This means that write operations are typically accelerated with FCIP-WA by a factor of 2:1,
halving the I/O service time associated with synchronous and asynchronous replication or approximately doubling the distance between data
centers for the same total actual I/O latency.
•
FCIP Tape Acceleration (FCIP-TA) is used to improve remote tape backup performance by minimizing the effect of network latency or distance
on remote tape applications. With FCIP-TA, a local Cisco MDS 9000 IPS or MPS module proxies as a tape library and a remote IPS or MPS
module (where the tape library is located) proxies as a backup server. Similar to FCIP-WA, FCIP-TA recognizes and proxies elements of
the upper-level SCSI protocol in order to minimize the number of end-to-end round trips required to transfer a unit of data and optimally make
use of the available network bandwidth. FCIP-TA achieves this by proxying the transfer ready and SCSI status responses while maintaining data
integrity in the event of a variety of error conditions. The net benefit of FCIP-TA is that tape throughput can be maintained irrespective of latency
and distance.
module (where the tape library is located) proxies as a backup server. Similar to FCIP-WA, FCIP-TA recognizes and proxies elements of
the upper-level SCSI protocol in order to minimize the number of end-to-end round trips required to transfer a unit of data and optimally make
use of the available network bandwidth. FCIP-TA achieves this by proxying the transfer ready and SCSI status responses while maintaining data
integrity in the event of a variety of error conditions. The net benefit of FCIP-TA is that tape throughput can be maintained irrespective of latency
and distance.