Use
the RouteTable statement to create a table of the routes that can
be used to route IP packets based on policy. A RoutingRule statement
specifies the characteristics of IP packets and references a RoutingAction
statement, which can reference one or more RouteTable statements.
The
RouteTable statement is used to create a table of static and dynamic
routes. The RouteTable statement is made up of Route entries and
DynamicRoutingParms entries. A route entry is used to create a static
route in the route table. The syntax for the route entry is compatible
with UNIX standards and similar
to the syntax for static routes in the TCP/IP profile's BEGINROUTES
block. Dynamic routes are added to the route table by OMPROUTE based
on the information provided in DynamicRoutingParms entries. IPv6
router advertisement routes are also added to the route table based
on the information that is provided in the DynamicRoutingParms entries.
Restrictions: - A limit of 255 route tables is allowed.
- Duplicate RouteTableRef parameters are not allowed within a RoutingAction
statement
The route table can be modified as follows:
- Incoming ICMP and ICMPv6 redirect packets can replace static routes,
and can also add routes to the route table.
- The OMPROUTE dynamic routing daemon can replace replaceable static
routes, and can add dynamic routes to the route table.
- IPv6 router advertisement routes can be added to the route table
based on received IPv6 router advertisements, and can replace replaceable
static routes.
- Direct host routes to dynamic XCF addresses on other TCP/IP stacks
are added when both of the following conditions are true:
- The dynamic XCF links to those stacks are active.
- DynamicXCFRoutes Yes or DynamicXCFRoutes6 Yes is specified on
the RouteTable statement.
When a RouteTable statement is updated, the route table
in the TCP/IP stack is updated. When a route entry is added, deleted,
or updated, the static routes in the route table are updated and routes
learned by way of ICMP or ICMPv6 redirect are deleted from the route
table. When a DynamicRoutingParms entry is added, deleted, or updated,
OMPROUTE updates the dynamic routes in the route table as needed,
and any IPv6 router advertisement routes in the route table are updated
as needed.
Route precedence is as follows:
- If a route exists to the destination address (a host route), it
is chosen first.
- For IPv4, if subnet, network, or supernetwork routes exist to
the destination, the route with the most specific network mask is
chosen second. The most specific network mask is the mask with the
most bits on. For IPv6, if prefix routes exist to the destination,
the route with the most specific prefix is chosen second.
- For IPv4, if the destination is a multicast destination and multicast
default routes exist, the route with the most specific multicast address
is chosen third.
- Default routes are chosen when no other route exists to a destination.
Rules: - The RouteTable statement must contain at least one Route entry
or one DynamicRoutingParms entry; otherwise, you must specify either
DynamicXCFRoutes Yes or DynamicXCFRoutes6 Yes.
- The required parameters for the route entry must be specified
in the order shown. The optional parameters can be specified in any
order.
- The parameters for the DynamicRoutingParms entry must be specified
in the order shown.
Tip: The Options parameters on the route entry
can be abbreviated using the same syntax that is used for static routes
in the TCP/IP profile's BEGINROUTES block.
Syntax
>>-RouteTable--name--------------------------------------------->
>--| Put Braces and Parameters on Separate Lines |-------------><
Put Braces and Parameters on Separate Lines
|--+-{-------------------------+--------------------------------|
+-| RouteTable Parameters |-+
'-{-------------------------'
RouteTable Parameters
.-IgnorePathMtuUpdate No-------.
|--+------------------------------+----------------------------->
'-IgnorePathMtuUpdate--+-Yes-+-'
'-No--'
.-IgnorePathMtuUpdate6 No-------.
>--+-------------------------------+---------------------------->
'-IgnorePathMtuUpdate6--+-Yes-+-'
'-No--'
.-Multipath UseGlobal----------.
>--+------------------------------+----------------------------->
'-Multipath--+-PerConnection-+-'
+-PerPacket-----+
+-Disable-------+
'-UseGlobal-----'
.-Multipath6 UseGlobal----------.
>--+-------------------------------+---------------------------->
'-Multipath6--+-PerConnection-+-'
+-PerPacket-----+
+-Disable-------+
'-UseGlobal-----'
.-DynamicXCFRoutes No-------.
>--+---------------------------+-------------------------------->
'-DynamicXCFRoutes--+-Yes-+-'
'-No--'
.-DynamicXCFRoutes6 No-------.
>--+----------------------------+------------------------------->
'-DynamicXCFRoutes6--+-Yes-+-'
'-No--'
.---------------------------------------------------------------------------------.
| .-----------------. |
V V | |
>----+--------------------------------------------------------+----+-------------+-+-+-->
'-Route--| Destination |--| First Hop |--| Packet Size |-' '-| Options |-'
.-------------------------------------------------------.
V |
>----+---------------------------------------------------+-+----|
| .-IPV4---------. |
'-DynamicRoutingParms --intf_name--+--------------+-'
+-IPV6---------+
'-gateway_addr-'
Destination
|--+-ipaddress/prefixLength-+-----------------------------------|
+-ipaddress--------------+
+-DEFAULT6---------------+
'-DEFAULT----------------'
First Hop
|--+-gateway_addr-+--intf_name----------------------------------|
'-=------------'
Packet Size
|--MTU--+-mtu_size----+-----------------------------------------|
'-DEFAULTSIZE-'
Options
.-NOREPLaceable-. .-MAXImumretransmittime 120.00--.
|--+---------------+--+-------------------------------+--------->
+-REPLaceable---+ '-MAXImumretransmittime seconds-'
'-NOREPLaceable-'
.-MINImumretransmittime 0.50----. .-ROUNDTRIPGain 0.125-.
>--+-------------------------------+--+---------------------+--->
'-MINImumretransmittime seconds-' '-ROUNDTRIPGain value-'
.-VARIANCEGain 0.25--. .-VARIANCEMultiplier 2.00--.
>--+--------------------+--+--------------------------+--------->
'-VARIANCEGain value-' '-VARIANCEMultiplier value-'
.-DELAYAcks---.
>--+-------------+----------------------------------------------|
+-NODELAYAcks-+
'-DELAYAcks---'
Parameters
- name
- A string 1 - 8 characters in length specifying the name of this
RouteTable statement.
Restriction: Do not specify the values
EZBMAIN or ALL (in any combination of upper and lower case letters)
for the name value. The name EZBMAIN is reserved for the main route
table that is generated by the TCP/IP profile's BEGINROUTES or GATEWAY
statements. The name ALL is reserved for use with the PR modifier
of the Netstat ROUTE/-r command.
- IgnorePathMtuUpdate
- Indicates whether IPv4 ICMP Fragmentation Needed messages should
be ignored for this route table. When IPv4 path MTU discovery
is enabled (PATHMTUDISCOVERY parameter on the IPCONFIG statement in
the TCP/IP profile), IPv4 ICMP Fragmentation Needed messages are used
to lower the MTU value that is used to send data to a specific IPv4 destination.
- No
- IPv4 ICMP Fragmentation Needed messages should be processed for
this route table. This is the default.
- Yes
- IPv4 ICMP Fragmentation Needed messages should be ignored for
this route table.
You might want to ignore the path MTU update
for a policy-based route table that contains routes that are known
to support large path MTU values. If routes in another route table
are defined to the same destination or destinations that need a smaller
path MTU value, specifying IgnorePathMtuUpdate Yes ensures that a
path MTU update that is the result of sending data on a small MTU
route does not cause an update to the path MTU for the route in the
policy-based route table.
Guideline: The IgnorePathMtuUpdate
option is an advanced option. You do not need to set IgnorePathMtuUpdate
Yes. If you specify IgnorePathMtuUpdate Yes, path MTU updates are
ignored for all IPv4 routes in the route table.
- IgnorePathMtuUpdate6
- Indicates whether IPv6 ICMP Packet Too Big messages should be
ignored for this route table. IPv6 ICMP Packet Too Big messages are
used to lower the MTU value that is used to send data to a specific
IPv6 destination.
- No
- IPv6 ICMP Packet Too Big messages should be processed for this
route table. This is the default.
- Yes
- IPv6 ICMP Packet Too Big messages should be ignored for this route
table.
You might want to ignore the path MTU update for a policy-based
route table that contains routes that are known to support large path
MTU values. If routes in another route table are defined to the same
destination or destinations that need a smaller path MTU value, specifying
IgnorePathMtuUpdate6 Yes ensures that a path MTU update that is the
result of sending data on a small MTU route does not cause an update
to the path MTU for the route in the policy-based route table.
Guideline: The IgnorePathMtuUpdate6 option
is an advanced option. You do not need to set IgnorePathMtuUpdate6
Yes. If you specify IgnorePathMtuUpdate6 Yes, path MTU updates are
ignored for all IPv6 routes in the route table.
- Multipath
- Indicates whether the multipath routing selection algorithm is
enabled for outbound IPv4 traffic that uses this policy-based
route table.
- UseGlobal
- Use the MULTIPATH or NOMULTIPATH setting from the IPCONFIG statement
in the TCP/IP profile to determine multipath processing. This is
the default.
- PerConnection
- Enables the multipath routing selection algorithm for outbound IPv4 traffic
that uses this policy-based route table. If multiple equal-cost routes
to an IPv4 destination exist in this policy-based route table,
a round-robin algorithm is used to select a route. After a route
is selected, connection-oriented or connectionless-oriented IP packets
using the same association always use the same route, as long as that
route is active.
- PerPacket
- Enables the multipath routing selection algorithm for outbound IPv4 traffic
that uses this policy-based route table. If multiple equal-cost routes
to an IPv4 destination exist in this policy-based route table,
a round-robin algorithm is used to select a route. Connection or
connectionless oriented IP packets using the same association do not
always use the same route, but do use all possible active routes to
the destination.
The PerPacket option should not be used if IP
security is enabled on the IPCONFIG statement in the TCP/IP profile.
If Multipath PerPacket is specified for a policy-based route table
and the route table is installed in a TCP/IP stack with IPv4 security
enabled, multipath routing for IPv4 traffic is disabled.
The following message is displayed: EZD0028I IPV4 MULTIPATH
PERPACKET NOT VALID WITH IPV4 SECURITY - MULTIPATH SUPPORT DISABLED
FOR ROUTE TABLE table
The Netstat ROUTE/-r PR command
displays the MultiPath setting No(Policy) if multipath
routing for IPv4 traffic has been disabled. This occurs because IPv4 security
is enabled.
- Disable
- Disables the multipath routing selection algorithm for outbound IPv4 traffic
that uses this policy-based route table. If multiple equal-cost routes
to an IPv4 destination exist in this policy-based route table,
the first active route that is found is used to send each IP packet
to that destination.
- Multipath6
- Indicates whether the multipath routing selection algorithm is
enabled for outbound IPv6 traffic that uses this policy-based route
table.
- UseGlobal
- Uses the MULTIPATH or NOMULTIPATH setting from the IPCONFIG6 statement
in the TCP/IP profile to determine multipath processing. This is the
default.
- PerConnection
- Enables the multipath routing selection algorithm for outbound
IPv6 traffic that uses this policy-based route table. If multiple
equal-cost routes to an IPv6 destination exist in this policy-based
route table, a round-robin algorithm is used to select a route. After
a route is selected, connection-oriented or connectionless-oriented
IP packets that use the same association always use the same route,
as long as that route is active.
- PerPacket
- Enables the multipath routing selection algorithm for outbound
IPv6 traffic that uses this policy-based route table. If multiple
equal-cost routes to an IPV6 destination exist in this policy-based
route table, a round-robin algorithm is used to select a route. Connection-oriented
or connectionless-oriented IP packets that use the same association
do not always use the same route, but use all possible active routes
to the destination.
If IP security is enabled on the IPCONFIG6 statement
in the TCP/IP profile, do not use the PerPacket option. If Multipath6
PerPacket is specified for a policy-based route table and the route
table is installed in a TCP/IP stack with IPv6 security enabled, multipath
routing for IPv6 traffic is disabled. The following message is displayed: EZD0028I
IPV6 MULTIPATH PERPACKET NOT VALID WITH IPV6 SECURITY - MULTIPATH
SUPPORT DISABLED FOR ROUTE TABLE table
The
Netstat ROUTE/-r PR command displays the MultiPath6 setting No(Policy) if
multipath routing for IPv6 traffic has been disabled. This occurs
because IPv6 security is enabled.
- Disable
- Disables the multipath routing selection algorithm for outbound
IPv6 traffic that uses this policy-based route table. If multiple
equal-cost routes to an IPv6 destination exist in this policy-based
route table, the first active route that is found is used to send
each IP packet to that destination.
- DynamicXCFRoutes
- Indicates whether direct routes to IPv4 dynamic XCF addresses
on other TCP/IP stacks should be added to this route table. The same
routes are automatically generated in the main route table when IPv4 dynamic
XCF links are active. See the dynamic XCF information in z/OS Communications Server: IP Configuration
Guide for information about the dynamic XCF function
and the definitions that are automatically generated when IPCONFIG
DYNAMICXCF is specified in the TCP/IP profile.
- Yes
- Add direct routes to IPv4 dynamic XCF addresses on other
TCP/IP stacks when the dynamic XCF links are active.
- No
- Do not add direct routes to IPv4 dynamic XCF addresses
on other TCP/IP stacks. This is the default.
Rule: Duplicate routes are not allowed within
a route table. If a statically defined route is a duplicate of a
route generated by DynamicXCFRoutes Yes, the statically defined route
takes precedence.
- DynamicXCFRoutes6
- Indicates whether direct routes to IPv6 dynamic XCF addresses
on other TCP/IP stacks should be added to this route table. The same
routes are automatically generated in the main route table when IPv6
dynamic XCF links are active. See the dynamic XCF information in z/OS Communications Server: IP Configuration
Guide for information about the dynamic XCF function
and the definitions that are automatically generated when IPCONFIG6
DYNAMICXCF is specified in the TCP/IP profile.
- Yes
- Add direct routes to IPv6 dynamic XCF addresses on other TCP/IP
stacks when the dynamic XCF links are active.
- No
- Do not add direct routes to IPv6 dynamic XCF addresses on other
TCP/IP stacks. This is the default.
Rule: Duplicate routes
are not allowed within a route table. If a statically defined route
is a duplicate of a route that the DynamicXCFRoutes6 Yes generates,
the statically defined route takes precedence.
- Route
- A route entry is used to create a static route in the route table.
Restriction: Duplicate
routes are not allowed within a RouteTable statement. Duplicate routes
have the same destination and first hop specification (interface
name and gateway address).
- DynamicRoutingParms
A DynamicRoutingParms entry provides parameters for OMPROUTE
to use when generating dynamic routes for the route table and for
the stack to use when adding IPv6 router advertisement routes to the
route table.
Restriction: Duplicate and overlapping
DynamicRoutingParms values are not allowed within a RouteTable statement.
In
the following example, the DynamicRoutingParms in both Table1 and
Table2 are treated as routing policy errors because the DynamicRoutingParms
values overlap:
RouteTable Table1
{
DynamicRoutingParms Link1
DynamicRoutingParms Link1 10.1.2.3
}
RouteTable Table2
{
DynamicRoutingParms Link2 IPv6
DynamicRoutingParms Link2 FE80::1:2:3
}
In the following example, the DynamicRoutingParms
in both Table3 and Table4 are treated as Routing policy errors because
the DynamicRoutingParms values are duplicates.
RouteTable Table3
{
DynamicRoutingParms Link1
DynamicRoutingParms Link1
}
RouteTable Table4
{
DynamicRoutingParms Link2 FE80::1:2:3
DynamicRoutingParms Link2 FE80::1:2:3}
- ipaddress
- The destination address. The address must be a fully qualified
IP address.
The DEFAULT or DEFAULT6 keyword in this field specifies
a default route. The destination address can be a host, network, subnetwork,
supernetwork, or default address. For IPv6, the address cannot be
an IPv4-mapped IPv6 address in hexadecimal or dotted decimal format
or an IP address with the reserved prefix ::/96. A local address is
not valid for the destination address. Multiple routes that have an
identical destination can be specified. When multiple routes are specified,
all of them are used when multipath is enabled; otherwise, only the
first active route that is specified is used.
- prefixLength
- An integer value that represents the number of bits in the ipaddress value
that are used to determine the destination address of the route. For
an IPv4 destination, the value is in the range 1 - 32. For an IPv6
destination, the value is in the range 1 - 128.
- gateway_addr
- On a route entry, the gateway_addr value
is the host IP address of a gateway or router that you can
reach directly, and that forwards packets for the destination network
or host. The value must be either a fully qualified address or an
equal sign (=), which indicates that the messages are routed directly
to destinations on that network or directly to that host. A local
address is not valid for the gateway address. The equal sign is not
supported for a default route entry. For IPv6, the address cannot
be an IPv4-mapped IPv6 address in hexadecimal or dotted decimal format
or an IP address with the reserved prefix ::/96.
On a DynamicRoutingParms
entry, the gateway_addr value is one of
the following values:
IPv4
Indicates that the intf_name value
that is specified on this DynamicRoutingParms entry is an IPv4 interface.
This is the default.
IPv6
Indicates that the intf_name value
that is specified on this DynamicRoutingParms entry is an IPv6 interface.
- The host IP address of a gateway or router that you can reach
directly, and that forwards packets for the destination network or
host.
It must be a fully qualified address. A local address is
not valid for the gateway address. If an IPv6 address is specified,
it must be a link-local address. OMPROUTE uses the IP address and
the intf_name value to select dynamic routes to
be included in this route table. If the gateway_addr value
is an IPv6 address, the value is used with the intf_name value
to determine which IPv6 router advertisement routes are added to this
route table.
- intf_name
- The name of an interface as defined on the LINK or INTERFACE statement
in the TCP/IP profile.
Restriction: Loopback and VIPA links
are not allowed.
On a route entry, the intf_name value
is the name of the interface through which packets are sent to the
specified destination. If an intf_name value
is specified that is not defined in the TCP/IP profile, the route
is created but is not usable until that interface value is defined
in the TCP/IP profile.
Tip: Routes that are configured
for an undefined interface name are flagged as invalid on a Netstat
ROUTE/-r PR display. The flags field includes the letter I.
On
a DynamicRoutingParms entry, the intf_name value
is the name of an interface through which packets can be sent. OMPROUTE
uses the intf_name value to select dynamic
routes to be included in this route table. If intf_name is
the name of an IPv6 interface, the value is used to determine which
IPv6 router advertisement routes are added to this route table. If gateway_addr has
been specified, then intf_name is used in
combination with the gateway_addr value.
If an interface name is specified that is not defined in the TCP/IP
profile, no dynamic routes or IPv6 router advertisement routes are
created until the interface name is defined in the TCP/IP profile.
- MTU mtu_size
The maximum transmission unit (MTU) in bytes for the destination.
This value can be up to 65535. The keyword DEFAULTSIZE in this field
requests that TCP/IP supply a default value of 576 for IPv4 routes
and 1280 for IPv6 routes.
See Figure 1 for more information about
the largest MTU value that each IPv4 link type supports.
See Table 1 for
more information about the largest MTU value that each IPv4 interface
type supports.
See Table 2 for
more information about the largest MTU value that each IPv6 interface
type supports.
Packet size considerations - The largest mtu_size value that z/OS® Communications Server can
handle varies for different networks. For example, although the largest
packet size for the Ethernet protocol is 1500 bytes, the largest packet
size for the 802.3 protocol is 1492 bytes.
- The actual packet size is determined by the total network connection.
- If a locally attached host has a packet size smaller than your
packet size, transfers to that host use the smaller size.
- The TCP maximum segment size for the 3172 Interconnect Controller
Program is 4096. Any packet specifications over 4096 are limited by
this restriction. For example, if you specified the packet size 4352,
the resulting packet size would still only be 4096 plus the header
size, for a total packet size of 4132.
- Large packets can be fragmented by intervening gateways for IPv4
only. Fragmenting and reassembling packets is expensive because of
high bandwidth use and CPU time. Packets sent through gateways to
other networks should use the default size, DEFAULTSIZE, unless one
of the following conditions is true:
- All intervening gateways and networks are known to accept larger
packets.
- IPv4 Path MTU discovery is enabled by using PATHMTUDISCOVERY on
the IPCONFIG statement, which results in the TCP/IP stack dynamically
learning the maximum MTU for the total network connection. For IPv6,
Path MTU discovery is always enabled.
- If this is a CLAW link, the mtu_size cannot
exceed the write_size specified on the corresponding
DEVICE statement.
- You cannot specify an MTU value that is smaller than the default
MTU size. For IPv4, the default MTU value is 576 and for IPv6 it is
1280.
- REPLACEABLE
- Indicates that the static route can be replaced by OMPROUTE or
IPv6 router advertisements if a dynamic route to the same destination
is discovered. This parameter can be abbreviated to REPL.
Restrictions: - Only one type (replaceable or nonreplaceable) of static route
can be defined to the same destination. If multiple route entries
are specified to the same destination and the REPLACEABLE or NOREPLACEABLE
setting is not the same, it is considered to be a Routing policy error.
- Do not define replaceable static routes to destination addresses
that correspond to dynamic VIPAs for which the TCP/IP stack is a sysplex
distributor target. This is not validated by Policy Agent.
Tip: You can use the Netstat ROUTE/-r PR ALL RSTAT
command to display all replaceable static routes currently configured
in policy-based routing tables.
- NOREPLACEABLE
- Indicates that the static route cannot be replaced by dynamic
routes. The static route is always used to reach the destination,
regardless of any information that dynamic routes might be available.
This is the default behavior. This parameter can be abbreviated NOREPL.
Restriction: Only
one type (replaceable or nonreplaceable) of static route can be defined
to the same destination. If multiple route entries are specified
to the same destination and the REPLACEABLE or NOREPLACEABLE setting
is not the same, it is considered to be a Routing policy error.
Retransmission parameter considerations
The parameters listed in this topic affect
the TCP retransmit algorithms. When TCP packets are not acknowledged,
TCP begins to retransmit these packets at certain time intervals.
If these packets are not acknowledged after a specified number of
retransmits, TCP aborts the connection. The time interval
between retransmissions increases by approximately twice the previous
interval until the packets are acknowledged or the connection times
out.
The time intervals between retransmissions and the number
of times that packets are retransmitted before the connection times
out differs for initial connection establishment and for data packets
. For initial connection establishment, the initial time interval
is set at approximately 3 seconds and the SYN packet is retransmitted
5 times before the connection is timed out. Data packets use a smoothed
Round Trip Time (RTT) as the initial time interval, and data packets
are retransmitted 15 times before the connection is timed out. All
of the remaining parameters listed in this topic affect the data packet
retransmission algorithm. Only the MINIMUMRETRANSMITTIME parameter
affects the initial connection establishment.
Tip: A
new route lookup is performed after every two retransmissions for
a data packet. For more information about the route lookup process,
see
Route selection algorithm in
z/OS Communications Server: IP Configuration
Guide. Be careful when you design networks with
firewalls. A firewall in an alternate routing path can generate a
RESET packet for the rerouted data packets, which causes TCP to abort
the connection.
The retransmission parameters enable system
administrators who are familiar with TCP/IP transmission performance
to alter the flow of TCP/IP data packets and acknowledgments. Under
normal circumstances, the following occurs:
- TCP typically waits to receive two packets before sending one
ACK to acknowledge the data within them.
- When TCP sends a packet, it waits for an acknowledgment. If it
times out before getting an acknowledgment, it resends the packet.
Use the following parameters to adjust the retransmission time-out
calculations; slower transmission times prevent packets from being
resent as quickly:
- MAXIMUMRETRANSMITTIME
- MINIMUMRETRANSMITTIME
- ROUNDTRIPGAIN
- VARIANCEGAIN
- VARIANCEMULTIPLIER
- DELAYACKS
- NODELAYACKS
TCP uses these values in an algorithm called the TCP Retransmission
Timeout Calculation, which is described in RFC 793. When you use this
calculation, the following occurs:
- A smoothed round trip time (SRTT) and variance (VAR) is updated
from the individual RTT derived from each packet acknowledgement.
- The retransmit time for a new packet is set to twice (approximately)
the current SRTT value plus the VAR value.
- Each time a packet is retransmitted, the retransmit time value
is doubled.
- The actual interval time used for the initial packet and each
retransmission is the retransmit time calculated previously, but limited
by the configured MINIMUMRETRANSMITTIME and MAXIMUMRETRANSMITTIME
values.
- DELAYACKS | NODELAYACKS
- Controls transmission of acknowledgments when a packet is received
with the PUSH bit on in the TCP header.
- NODELAYACKS
- Specifies that an acknowledgment is returned immediately when
a packet is received with the PUSH bit on in the TCP header. The
NODELAYACKS parameter on the BEGINROUTES, GATEWAY, and RouteTable
statements affects only the connections that use this route. Specifying
NODELAYACKS on the TCP/IP stack BEGINROUTES or GATEWAY profile statements,
or on the Policy Agent RouteTable statement, overrides the specification
of the DELAYACKS parameter on the TCP/IP stack PORT, PORTRANGE, and
TCPCONFIG profile statements.
- DELAYACKS
- Delays transmission of acknowledgments when a packet is received
with the PUSH bit on in the TCP header. The DELAYACKS parameter on
the BEGINROUTES, GATEWAY, and RouteTable statements affects only the
connections that use this route. This is the default, but you can
override the default by specifying the NODELAYACKS parameter on the
TCP/IP stack PORT, PORTRANGE, or TCPCONFIG profile statements.
- MAXIMUMRETRANSMITTIME
- Limits the TCP retransmission interval. Decreasing this value
might decrease the total time it takes a connection to timeout. Specifying
MAXIMUMRETRANSMITTIME assures that the interval time never exceeds
the specified limit. The minimum value that can be specified for MAXIMUMRETRANSMITTIME
is 0. The maximum is 999.990. The default is 120 seconds. This parameter
does not affect initial connection retransmission.
- MINIMUMRETRANSMITTIME
- Sets a minimum retransmit interval. Increasing this value might
increase the amount of time it takes for TCP to time out a connection.
The minimum value that can be specified for MINIMUMRETRANSMITTIME
is 0. The maximum is 99.990. The default is 0.5 (500 milliseconds).
- ROUNDTRIPGAIN
- This value is the percentage of the latest Round Trip Time (RTT)
to be applied to the smoothed RTT average. The higher this value,
the more influence the latest packet RTT has on the average. The minimum
value that can be specified for ROUNDTRIPGAIN is 0. The maximum value
is 1.0. The default is 0.125. This parameter does not affect initial
connection retransmission.
- VARIANCEGAIN
- This value is the percentage of the latest RTT variance from the
RTT average to be applied to the RTT variance average. The higher
this value, the more influence the latest packet's RTT has on the
variance average. The minimum value that can be specified for VARIANCEGAIN
is 0. The maximum value is 1.0. The default is 0.25. This parameter
does not affect initial connection retransmission.
- VARIANCEMULTIPLIER
- This value is multiplied against the RTT variance in calculating
the retransmission interval. The higher this value, the more affect
variation in RTT has on calculating the retransmission interval. The
minimum value that can be specified for VARIANCEMULTIPLIER is 0. The
maximum value is 99.990. The default is 2. This parameter does not
affect initial connection retransmission.
Retransmission parameters
Use the ROUNDTRIPGAIN,
VARIANCEGAIN, and VARIANCEMULTIPLIER parameters to instruct TCP how
heavily to weigh the most recent behavior of the network versus the
long term behavior for updating the SRTT and VAR values. If you specify
smaller values for these parameters, TCP attempts to correct for congestion
only if the congestion is sustained. With larger values, TCP corrects
for congestion more quickly, and the system is more sensitive to variations
in network performance. Use the default values (unless your retransmission
rate is too high).
Use DELAYACKS to delay the acknowledgments
so that they can be combined with data sent to the foreign host.
Results: - If a HOME entry or INTERFACE is deleted from the TCP/IP profile,
all routes for the associated interface become unusable. The routes
remain in the route table and become usable again if the HOME entry
or INTERFACE is added back to the profile.
- If an interface becomes inactive, then all routes that are associated
with that interface are marked inactive by the stack.
- If an interface becomes active, then all static routes that are
associated with that interface are marked active by the stack.
Rules: - There is no limit to the number of equal-cost multipath routes
that can be associated with a single destination.
- Multicast routes can be specified using a host specification.
You can also specify multicast network routes or prefix routes.
- A valid host address must contain a nonzero value in the host
portion of the address. The host portion of an IPv4 address cannot
be all ones, which is considered the broadcast address.
- On an IPv4 route entry, the destination IP address can be either
a network or a host IPv4 address, and the gateway_addr value
must be a host IPv4 address.
- On an IPv6 route entry, the destination IP address can be either
a prefix or host IPv6 address, and the gateway_addr value
must be a host IPv6 address.