This guide provides information about installing, configuring, restoring, and managing the HP IP Distance Gateway (mpx110).
It is intended for system administrators responsible for installing, managing, and servicing the mpx110 and the SAN to which
it is attached.
HP Part Number: 5697-2060
Published: June 2012
Edition: 9
The information contained herein is subject to change without notice. The only warranties for HP products and services are set forth in the express
warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. HP shall
not be liable for technical or editorial errors or omissions contained herein.
Microsoft and Windows are U.S. registered trademarks of Microsoft Corporation.
Java is a U.S. trademark of Sun Microsystems, Inc.
This chapter provides a description of the mpx110 and an overview of FCIP.
mpx110 product description
The HP IP Distance Gateway (referred to as the mpx110) provides FC SAN extension over an IP
network. Used in conjunction with the P10000/3PAR, P6000/EVA, and P9000/XP storage system
families, HP Continuous Access P6000/EVA, P9000/XP, and P10000/3PAR Remote Copy
software, the mpx110 provides long-distance remote replication for disaster tolerance.
A base FCIP configuration consists of a minimum of two mpx110 gateways—one for the local site
and one for the remote site. A single mpx110 gateway is HP part number AG680A—HP IP Distance
Gateway. One mpx110 per site is required. See “FCIP overview” (page 9).
Optional equipment
HP part number AG681A—HP IP Distance Gateway Upgrade (single mpx110 gateway for
redundancy, one per site required) is available for hardware redundancy.
The following configurations implement redundant pairs of gateways:
•Redundant pairs of gateways, one long-distance link (page 12)
•Redundant pairs of gateways, two long-distance links (page 13)
•Redundant pairs of gateways, fully redundant long-distance links (page 13)
NOTE:See Configuration rules and guidelines (page 11) for additional required and optional
equipment for your configuration.
8Overview
FCIP overview
FCIP enables connectivity between geographically dispersed FC devices over an IP network. To
deploy FCIP, two mpx110 gateways are required. Each gateway is configured for FCIP and
connected to a fabric. The gateways are connected to each other through an IP network
(LAN/WAN). For more information, see Figure 1 (page 9).
Local FC devices need no additional hardware or software to access remote FC devices using the
mpx110 deployed for FCIP.
Figure 1 FCIP overview
Using FCIP to encapsulate FC packets
With FCIP, gateways transport FC frames over an IP network. From the perspective of the local
and remote fabrics, the FC devices accessed through the gateways appear to be part of one unified
fabric. This effect is possible because FC traffic is carried over the IP network in such a way that
the FC fabric and all FC devices on the fabric are unaware of the presence of the IP network.
Once configured, FCIP instances on each gateway become active and establish their connectivity
through the IP network. The FC devices in the local fabric access the FC devices in the remote
fabric using FC frames. The FC frames are encapsulated in IP packets by the local gateway and
then transmitted to the remote gateway. The remote gateway strips the IP packet data and passes
only the FC frames to the remote FC devices.
The gateways deployed for FCIP are configured to use TCP, which uses standard TCP flow control
and error recovery algorithms.
FCIP overview9
Redundant FCIP network structure example
In a high-availability FCIP configuration, such as between pairs of mpx110 gateways and two
independent IP networks that provide full redundancy, a loss of connectivity that occurs through
one of the IP networks does not result in a loss of connectivity between the fabrics. See FCIP fully
Figure 11 Basic configuration, MPX200 FCIP with remote IP Distance Gateway (mpx110)
14Configuration rules and guidelines
Figure 12 shows a configuration using the mpx110 with FCIP and B-series switches with Integrated
WAN
FCIP
LANLAN
GbEGbE
26626a
Fabric A2
Fabric A1
Local
storage system
Remote
storage system
Local FC
servers
Remote
FC servers
FC1
FC2
GE1
GE2
MGMT
IOIOI
!
HP StorageWorks
mpx100
mpx110
FC1
FC2
GE1
GE2
MGMT
IOIOI
!
HP StorageWorks
mpx100
mpx110
EX
E
Routing. This provides fabric isolation between the local and remote fabrics, enabling device access
without merging the fabrics. This can be implemented in all supported mpx110 FCIP configurations
using B-series Fibre Channel switches with Integrated Routing or B-series routers configured for
Fibre Channel routing.
Figure 12 mpx110 FCIP with B-series Integrated Routing
Supported configurations15
Figure 13 shows a configuration using the mpx110 with FCIP and C-series switches with IVR. This
VSAN A1
WAN
FCIP
LANLAN
GbEGbE
26637a
Local
storage system
Remote
storage system
Local FC
servers
Remote
FC servers
FC1
FC2
GE1
GE2
MGMT
IOIOI
!
HP StorageWorks
mpx100
m p x 11 0
FC1
FC2
GE1
GE2
MGMT
IOIOI
!
HP StorageWorks
mpx100
m p x 11 0
E
E
VSAN A2
provides fabric isolation between the local and remote fabrics, allowing device access without
merging the fabrics. This can be implemented in all supported mpx110 FCIP configurations using
C-series Fibre Channel switches with IVR.
Figure 13 mpx110 IP Distance Gateway FCIP with C-series IVR
16Configuration rules and guidelines
HP Continuous Access P6000/EVA 3-site configurations
This section describes the following HP Continuous Access P6000/EVA 3-site configurations:
•HP Continuous Access EVA 3-site configuration with four gateways (page 17)
•HP Continuous Access EVA 3-site configuration with six gateways (page 17)
•HP Continuous Access EVA 3-site configuration with eight gateways (page 19)
•(page 20)
The first three configurations provide a fan-in or fan-out relationship between the sites. The fourth
configuration provides a peer-to-peer relationship between all sites.
Figure 14 (page 17) shows connectivity for three sites using four mpx110 gateways, which
implements the minimum-level and lowest-cost connectivity for a 3-site configuration. Figure 15 (page
18) shows additional connectivity and redundancy using six mpx110 gateways. Figure 16 (page
19) shows the highest level of 3-site connectivity using eight mpx110 gateways.
Figure 17 (page 20) is similar to Figure 15 (page 18), with additional connectivity to allow for
replication between Site 2 and Site 3.
The following configuration rules apply to Figure 14 (page 17) through Figure 16 (page 19)
(fan-in/fan-out relationships):
•For Site 1, Site 2 or Site 3 can function as the remote site.
•For Site 2 or Site 3, Site 1 can function as the remote site.
•Replication between Site 2 and Site 3 is not supported.
The following configuration rules apply to Figure 17 (page 20) (peer-to-peer relationship):
•For Site 1, Site 2 or Site 3 can function as the remote site.
•For Site 2, Site 1 or Site 3 can function as the remote site.
•For Site 3, Site 1 or Site 2 can function as the remote site.
3-site configuration with four mpx110 gateways
The configuration shown in Figure 14 (page 17) provides long-distance link redundancy between
all three sites.
Figure 14 HP Continuous Access P6000/EVA 3-site configuration with four mpx110 gateways
3-site configuration with six mpx110 gateways
The configuration shown in Figure 15 (page 18) provides the same long-distance link redundancy
as the configuration shown in Figure 14 (page 17), with the addition of redundant mpx110
gateways at sites 2 and 3.
HP Continuous Access P6000/EVA 3-site configurations17
Figure 15 HP Continuous Access P6000/EVA 3-site configuration with six mpx110 gateways
18Configuration rules and guidelines
3-site configuration with eight mpx110 gateways
The configuration shown in Figure 16 (page 19) provides the highest 3-site redundancy, with a
dedicated mpx110 pair for all long-distance links to all three sites.
Figure 16 HP Continuous Access P6000/EVA 3-site configuration with eight mpx110 gateways
HP Continuous Access P6000/EVA 3-site configurations19
3-site configuration with six gateways and full inter-site connectivity
Figure 17 (page 20) provides long-distance link redundancy and full connectivity between all three
sites.
Figure 17 HP Continuous Access P6000/EVA 3-site configuration with six gateways, full peer-to-peer
connectivity
20Configuration rules and guidelines
Configuration rules and guidelines
The following sections define the configuration rules for using the mpx110 gateways for FCIP.
General configuration rules
The following general configuration rules apply:
•All mpx110 configurations require a minimum of two mpx110 gateways, or one mpx110
and one MPX200 Multifunction Router with an FCIP license, one local and one remote,
connected through an IP network.
•The mpx110 gateway must connect to another mpx110 or an mpx200 Multifunction Router
with an FCIP license. HP does not support FCIP connectivity between other gateway models.
•The mpx110 gateway is supported for FCIP extension with HP P9000 and P6000 Continuous
Access (see EVA storage system rules and guidelines (page 21) and XP storage system rules
and guidelines (page 22)), and P10000/3PAR Remote Copy software.
NOTE:For current support, see SPOCK at http://www.hp.com/storage/spock. You must sign
up for an HP Passport to enable access.
Operating system and multipath support
The mpx110 gateway is supported using FCIP with all operating systems and multipath software
supported by HP. For more information, see the HP SAN Design Reference Guide, available at
http://www.hp.com/go/SDGManuals.
P6000/EVA storage system rules and guidelines
Observe the following P6000/EVA storage system rules and guidelines:
•P6350/P6300/P6550/P6500
•The mpx110 gateway configured for FCIP is supported for use with the following HP Continuous
Access P6000/EVA storage systems:
◦EVA4400/4400 with embedded switch
◦EVA4000/4100/6000/6100/8000/8100
◦EVA6400/8400
•The mpx110 gateway is supported for use in all HP-supported P6000 Continuous Access
configurations, including the standard two-fabric, five-fabric, and six-fabric configurations.
•HP P6000 Continuous Access supports RCS and non-RCS LUNs with FCIP extension.
•HP mpx110 gateway supports the minimum IP bandwidth/maximum DR groups.
Table 1 (page 22) defines the minimum IP bandwidth and maximum EVA DR groups for EVA XCS
and VCS.
Configuration rules and guidelines21
Table 1 Minimum IP bandwidth and maximum DR groups
Gateway pair
Minimum IP bandwidth and maximum DR groups
1
Single or shared IP link latency (0 to 100 ms one-way)Dual fabric latency (0 to 100 ms one-way)
IP Distance
Gateway
(mpx110)
1
1 Gb/s IP bandwidth can have up to 128 DR groups with VCS 4.x, and up to 256 DR groups with XCS.
Minimum: At least 2 Mb/s for 1 DR group
Recommended: At least 5 Mb/s for 1 to 5 DR groups
Minimum: At least 4 Mb/s for 1 DR group
Recommended: At least 10 Mb/s for 1 to 5 DR groups
P6000/EVA storage system software
The mpx110 gateway is supported with current P6000/EVA storage software applications such
as HP P6000 Continuous Access, Command View EVA, Business Copy, SSSU, and Replication
Solutions Manager.
P9000/XP storage system rules and guidelines
Observe the following P9000/XP storage system rules and guidelines:
•Supported P9000/XP models are P9500/XP24000/20000 and XP12000/10000, with
supported firmware levels. For more information, see SPOCK at http://www.hp.com/storage/
spock.
•The mpx110 gateway configured for FCIP is supported for use with P9000/XP Continuous
Access Sync, Async, and Journal.
•The mpx110 gateway is supported for use in all HP-supported P9000/XP Continuous Access
FCIP configurations. For more information, see the P9000/XP Continuous Access documentation
and the HP SAN Design Reference Guide, available at http://www.hp.com/go/SDGManuals.
•A P9000/XP storage system requires a minimum IP bandwidth of 16 Mb/s per path. The
maximum latency is 100 ms round-trip.
P9000/XP storage system software
The mpx110 gateway is supported with current versions of P9000/XP storage software applications,
such as XP Continuous Access, Command View XP, Continuous Access Journal XP, Business Copy
XP, and XP Array Manager.
Fibre Channel switch and firmware support
The mpx110 is compatible with the following Fibre Channel switches:
NOTE:For current support, see SPOCK at http://www.hp.com/storage/spock. You must
sign up for an HP Passport to enable access.
FC switch requirements
The following additional B-series, C-series, and H-series FC switch requirements must be observed:
•Local and remote mpx110 gateway pairs must be connected to the same Fibre Channel switch
product line series.
•The maximum distance between an mpx110 and a Fibre Channel switch is 300 meters at 2
Gb/s.
NOTE:The mpx110 Fibre Channel ports operate at 2 Gb/s.
22Configuration rules and guidelines
IP network requirements
HP requires that the following standards be met for the IP network:
•Supported network protocols are TCP/IP IPv4 and IPv6 Ethernet 1,000 Mb/s.
See EVA storage system rules and guidelines (page 21) and XP storage system rules and
guidelines (page 22) for minimum IP bandwidth requirements.
•For mpx110 IP data—Local and remote pairs are supported for up to 100 ms of IP network
delay one-way, or 200 ms round-trip for HP Continuous Access P6000. HP requires dedicated
IP bandwidth (see Table 2 (page 24)). For P9000 Continuous Access, the maximum distance
and delay is based on the replication mode. See “P9000/XP and VA storage system rules”
in the HP SAN Design Reference Guide.
•For mpx110 IP management—LAN and WAN are supported.
Configuration rules and guidelines23
Table 2 Network requirements for the mpx110 with XCS and VCS
DescriptionSpecification
Bandwidth
1
Must be dedicated to the HP Continuous Access storage replication function.
There is no support for dynamic pacing of the gateway.
Maximum number of DR groups
See Table 1 (page 22) for minimum supported bandwidth based on the
average packet-loss ratio and one-way intersite latencies.
1,500 bytesMTU of the IP network
Maximum latency
1
P6000/EVA: 100 ms one-way or 200 ms round-tripP9000/XP: 50 ms one-way
or 100 ms round-trip
Average packet-loss ratio
2
Low-loss network: 0.0012% average over 24 hoursHigh-loss network: 0.2%
average over 24 hours; must not exceed 0.5% for more than 5 minutes in a
2-hour window
3
1
Pre-existing restriction
2
A high packet-loss ratio indicates the need to retransmit data across the intersite link. Each retransmission delays
Must not exceed 10 ms over 24 hoursLatency jitter
transmissions queued behind the current packet, thus increasing the time to complete pending transactions.
3
Latency jitter is the difference between the minimum and maximum values, and indicates how stable or predictable the
network delay. The greater the jitter, the greater the variance in the delay, which lowers the performance predictability.
24Configuration rules and guidelines
IP performance tuning
The mpx110 supports Fibre Channel service at transmission rates of 1 Gb/s or 2 Gb/s with a
maximum frame size of 2,148 bytes. It supports Ethernet service at transmission rates of 1,000 or
100 Mb/s with an MTU size between 1,000 and 9,000 bytes (jumbo frames). Related performance
characteristics include the following:
Consider the physical distance between the mpx110 gateways. This is usually measured in RTT.
The RTT can be anywhere from less than 1 millisecond to as great as 200 milliseconds.
Bandwidth per route
Bandwidth is a measure of the volume of data that can be transmitted at a given transmission rate.
WAN data rates range from 1.5 Mb/s (T1) to greater than 600 Mb/s (OC-12). The proper
mpx110 bandwidth setting is determined based on the bandwidth available for each FCIP route,
irrespective of the total bandwidth and physical speed of the link.
To determine the proper mpx110 bandwidth setting, start with the total bandwidth of the WAN
link. Adjust this number based on the guaranteed allocated FCIP bandwidth, then further adjust
this number if the number of FCIP routes configured is greater than one. For example, if the WAN
link is 45 Mb/s, and 15 Mb/s is allocated to network traffic, the remaining 30 Mb/s is available
for FCIP. If in this example the mpx110 is configured for two routes, based on the 30 Mb/s total
bandwidth available for FCIP, you would set the mpx110 bandwidth parameter to 15, the available
bandwidth for each FCIP route.
HP P9000 and P6000 Continuous Access replication solutions require dedicated bandwidth for
the intersite link. If other applications share the intersite link, some method of QoS must be used
to ensure that the replication application has uncontested access to the allocated bandwidth.
NOTE:Setting the bandwidth per route parameter higher than the actual bandwidth available
for each route results in a decrease in performance; the optimal setting matches the bandwidth
per route setting to the actual bandwidth available to each FCIP route.
Latency
Latency is the amount of time a packet takes to traverse the network from source to destination.
MTU/Jumbo frames
For MTU size there are 3 settings, Normal (1500 bytes), Jumbo (9000 bytes) and Other, where
you are then prompted for a value between 1000 and 9000.
•Normal: Typically MTU should be set to the default of 1500; rarely do WAN networks support
MTU sizes greater than 1500.
•Jumbo - Jumbo frames can enhance the IP performance of the mpx110. Before enabling jumbo
frames, ensure that all switches in the IP path are configured for jumbo frames.
IP performance tuning25
NOTE:Jumbo frames are not supported for use with HP P9000 or P6000 Continuous Access.
•Other: Allows you to configure the MTU size to a value between 1000 and 9000 bytes.
Encryption products on the WAN link often add some number of additional bytes to each
packet, so it may be necessary to decrease the mpx110 MTU size setting to between 1450
and 1200. This accommodates the additional bytes, while maintaining a total MTU size of
1500 or less. Keeping the total MTU size to a maximum of 1500 ensures a single Fibre
Channel frame (2112 bytes) will fit within two Ethernet packets, resulting in optimal
performance.
Compression
The mpx110 integrates a software compression option. Enable compression for IP fabrics with an
RTT greater than or equal to 50 ms, or guaranteed WAN bandwidth of less than or equal to 45
Mb/s. See TCP window size recommendations (page 28) for compression options for specific
network rates and RTT.
TCP window size/scaling performance tuning
The mpx110 performance is maximized when properly configured. Knowing the RTT (distance)
between mpx110 gateways and the WAN effective data rate (connection type) allows the gateways
to be tuned for optimal performance. See (page 26).
Modifying the window size and scaling factor
The mpx110 window size can be set to a maximum of 32 KB. The scaling factor is used as a
multiplier to increase the window size above 32 KB. Modify the window size and scaling factor
in the mpx110 gateway pairs based on the WAN RTT and link speed. To determine the appropriate
window size setting, use the pre-populated tables or a formula.
If the recommended TCP window size scale factor for a given RTT and WAN link speed is not
shown in Table 4 (page 28) through Table 10 (page 31), use the following formula:
(RTT (ms) x link rate (Kb/s) x (1 byte/8 bits) = minimum window size (MWS)
Then, convert the MWS to a recommended scale factor by dividing it by the default window size
(32,768 bytes). Use Table 3 (page 27) to determine the scale factor.
26Configuration rules and guidelines
Table 3 TCP window size scale factors
Scale factorMWS scale result
00 to 2
12 to 4
24 to 8
38 to 16
416 to 32
532 to 64
664 to 128
7128 or greater
IP performance tuning27
TCP window size recommendations
Table 4 (page 28) through Table 10 (page 31) provide TCP window scaling factor and window
size settings for specific WAN environments.
NOTE:The TCP window size recommendations listed in Table 4 (page 28) through Table 10 (page
31) are based on low-loss networks (0.0012% average packet-loss ratio over 24 hours). For
higher-loss, longer-latency networks, you should reduce the recommended window size and scaling
factor by one setting to compensate for the increased number of packet retransmissions.
Table 4 T1 / DS-1: 1.554 Mb/s
Round-trip time
size (bytes)
Scaling factorTotal window
Compression
recommendations
ON164 K200
ON032 K100
ON032 K50
ON032 K25
ON032 K20
ON032 K15
ON032 K10
ON032 K5
ON032 K2.5
ON032 K1 or less
28Configuration rules and guidelines
Table 5 T3 / DS-3: 45 Mb/s
Round-trip time
size (bytes)
Table 6 DS-5: 400 Mb/s
Round-trip time
size (bytes)
Scaling factorTotal window
Scaling factorTotal window
Compression
recommendations
ON62 MB200
ON51 MB100
ON3256 K50
ON2128 K25
ON2128 K20
ON164 K15
ON164 K10
ON032 K5
ON032 K2.5
ON032 K1 or less
Compression
recommendations
ON74 MB200
ON74 MB100
ON74 MB50
OFF62 MB25
OFF51 MB20
OFF51 MB15
OFF4512 K10
OFF3256 K5
OFF2128 K2.5
OFF164 K1 or less
IP performance tuning29
Table 7 OC-1: 50 Mb/s
Round-trip time
size (bytes)
Table 8 100-BASE T: 100 Mb/s
Round-trip time
size (bytes)
Scaling factorTotal window
Scaling factorTotal window
Compression
recommendations
ON62 MB200
ON51 MB100
ON4512 K50
ON2128 K25
ON2128 K20
ON or OFF2128 K15
OFF164 K10
OFF032 K5
OFF032 K2.5
OFF032 K1 or less
Compression
recommendations
ON74 MB200
ON62 MB100
ON51 MB50
ON4512 K25
ON3256 K20
ON or OFF3256 K15
OFF2128 K10
OFF164 K5
OFF032 K2.5
OFF032 K1 or less
30Configuration rules and guidelines
Loading...
+ 123 hidden pages
You need points to download manuals.
1 point = 1 manual.
You can buy points or you can get point for every manual you upload.