|
Standard SSFNet DML attributes |
|---|
|
Reminder:
A DML schema specifies a concrete DML attribute tree by specifying key names,
attribute value types, and the allowed number of attribute instances. Conveniently,
DML schemas can be written using the DML syntax, where an attribute's value
type and number of instances are represented by a formatting expression
embedded in the DML Value of String type.
A keypath uniquely identifies an attribute in a schema tree
(either internal or leaf) by a sequence of dot-separated keynames along
the path from the root to the attribute. This is analogous to filesystem
pathnames on Unix machines.
%[SFI]count (String, Float, Integer) specify leaf attribute
value type and number of instances. A trailing ! after an integer count specifies that the count
must be exact.
Consult the
DML reference manual for information on DML syntax.
|
Standard configuration schema for class SSF.Net.Net
|
|
The general schema of a DML configuration file read by class SSF.Net.Net to instantiate a simulation
is described below, attribute by attribute. It is an expanded version of the schema file
SSFNet/examples/net.dml included in the SSFNet distribution.
A Net is a collection of Hosts, Routers, links and included Nets.
Net inclusion is a powerful construct that permits to build
very large models from "preconfigured subnetworks".
Specification of included Nets, hosts and routers must contain either a single integer ID
(id) or a range
of contiguous, ascending integer IDs (idrange) but not both.
The attribute idrange is used as a compact notation for an array of components. Relative to
the immediately enclosing Net context the id and
idrange values must all be distinct.
These integer identifiers are employed in the
NHI component addressing method
that permits to compose bigger networks from instances of Net
without the need to re-number their internal components' ids.
Each Net component may be aligned to a symbolic alignment partition
(or "timeline"). A component's alignment may differ from that of the enclosing Net.
An SSF simulator executing on a multiprocessor
machine will assign all components with the same alignment value
to the same processor (thus they share the same simulated time value),
while partitions with distinct alignment values
may execute on distinct processors to exploit the available parallelism.
If an alignment attribute is omitted, the enclosing component is assigned to
the default alignment, thus if all alignment attributes are
omitted then all simulation Entities execute
on the same processor (serial execution) even if multiple CPUs are available.
|
|
schema fragment |
explanation
|
schemas [
Net [
frequency %I<1
|
Net is the top-level attribute enclosing the entire network configuration.
Net.frequency used only in toplevel Net sets the physical unit value of the smallest
simulated time unit as
the number (long int) of dimensionless simulator "ticks" per simulated second: thus in a simulation
where at least one component requires a microsecond time resolution one would
set the value as "frequency 1000000", and the simulated time resolution
will be "one tick = one microsecond", that is the simulated time will advance
by intervals measured in multiples of microseconds. See
SSFNet Programming Guidelines for
discussion.
|
Net [
id %I
idrange [from %I1! to %I1!]
alignment %S
cidr %S
ip %S
_extends .schemas.Net
]
|
Included Net, recursively defined by inheritance keyword _extends as having the
same attribute tree as the toplevel Net (i.e., Net, host, router, link) in addition to explicitly specified attributes.
Must contain either a single integer ID
(id) (or a range
of contiguous, ascending integer IDs (idrange) but not both).
Relative to
the immediately enclosing Net context the id and
idrange values must all be distinct.
Optional attribute Net.cidr characterizes a hierarchical
partition into CIDR IP address blocks within the enclosing Net context, see CIDR section.
Optional attribute Net.ip is a VLSM (variable length subnet mask) IP
address of a network in a.b.c.d/p notation, see IP section.
|
host [
id %I
idrange [from %I1! to %I1!]
alignment %S
|
Net.host specifies a machine
with a protocol graph installed, and zero or more configured
network interfaces. (Net.router has identical attributes.)
Must contain either a single integer ID
(id) (or a range
of contiguous, ascending integer IDs (idrange) but not both).
Relative to
the immediately enclosing Net context the id and
idrange values must be unique for each host and router.
|
graph [
description %S
ProtocolSession [
name %S1!
use %S1!
]
]
|
Net.host.graph is an internal attribute specifying a list of
protocols to be configured and the names of SSF.OS classes to use.
graph.description is an optional protocol graph description string,
ProtocolSession.name is a symbolic tag, like "tcp" or "ip",
and ProtocolSession.use is the name of a class
extending SSF.OS.ProtocolSession that will be installed. Use one ProtocolSession
for each protocol.
|
interface [
id %I
idrange [from %I1! to %I1!]
ip %S
bitrate %F
latency %F
virtual %S
queue [ use %S ]
buffer %I
monitor [ use %S ]
tcpdump %S
flaky %F
]
|
Net.host.interface:
configuration of network interface such as an
Ethernet card.
Must contain either a single integer ID
(id) (or a range
of contiguous, ascending integer IDs (idrange) but not both).
Relative to
the immediately enclosing host context the id and
idrange values must be unique for each interface.
An interface that is not named in any link attribute is considered "down" - that is, unconnected.
Optional Net.host.ip in a.b.c.d/p notation, see IP section.
If a host specification names multiple interfaces
with interface.idrange and one interface.ip
IP address attribute,
each interface in the range consumes successive IP addresses, using the given
IP attribute as the base address. In general, it is strongly recommended to avoid
manually assigning IP addresses and to rely on automatically assigned addresses.
Attributes bitrate (bits per second) and latency (on-card latency in seconds),
together describe the timing behavior of a stream of packets leaving the interface.
If omitted, the default bitrate is 10 Mbs = 10,485,760 bps; and the
default latency is 0.0001 sec.
If optional attribute virtual is "true",
construct a virtual interface; if omitted virtual the default is "false".
Optional attribute queue names the queue manager class to use (in String %S),
if omitted, the default queue class is SSF.Net.droptailQueue. Any addidional queue
configuration attributes should be implemented as sub-attributes of queue.
Optional attribute buffer specifies the queue buffer size in bytes
(as used by SSF.Net.droptailQueue).
If not used, the default is unlimited buffer size.
Optional attribute monitor specifies to use class named in String %S for
the queue monitor. A queue monitor must implement Java interface SSF.Net.PacketQueueMonitor.
If omitted, no monitor is installed.
Optional attribute interface.tcpdump will 'tcpdump' all passing traffic
into the named file; using class SSF.OS.binaryTcpDump.
Optional attribute interface.flaky with float value P, 0.0<=P<=1.0,
will cause interface instance to drop packets with the given probability P.
The default value is 0.0.
|
route [
dest %S1!
interface %I<1
next_hop %S<1
]
nhi_route [
dest %S1!
interface %I<1
next_hop %S<1
]
] # end of .schemas.Net.host
|
The attribute Net.host.route allows to manually specify a static route to remote network or host.
For a host attached to a LAN it is mandatory to specify the route to the gateway router
using dest default.
Either route.interface or
route.next_hop must be provided. If only route.interface
is provided, the specified interface must be on a point-to-point link, so that
the next hop is unique. If route.interface attaches to a link with two or
more other interfaces, route.next_hop is mandatory if
route.dest is not on this link.
Note two variants of static route specification, using either IP or NHI addressing.
The values of route.dest and route.next_hop
are an IP address (in a.b.c.d/p notation) of destination host or network, and
IP address of next hop, repectively.
The values of nhi_route.dest and nhi_route.next_hop
are an NHI address of destination host or router interface, or Net, and
NHI address of next hop interface, repectively.
In both variants the interface attribute is an integer
interface ID to use in the host's context.
|
router [
_extends .schemas.Net.host
]
|
Syntactically, a router is synonymous with host, but it will
usually have distinct protocols in its protocol graph,
and optional configuration attributes required by OSPF or
other routing protocols. The different roles of routers and hosts
in a simulation warrant using different names.
|
link [
attach %S2
delay %F
alignment %S
cidr %S
ip %S
]
|
Net.link
specifies a level 2 network connecting a set of attached host
and/or router network interfaces.
A link should be understood as a "link layer". It must have two or more
link.attach attributes naming the attached network interfaces.
Does not mandate what level 2 (i.e., link layer) routing
or broadcast capabilities are present, or what symmetries of interface
properties must exist (e.g., matching bitrates) -- that's left to the
interfaces involved.
link.delay, with value in seconds, represents
the contribution of the link itself to total transmission delay of an IP packet.
It does not include interface-specific sources of delay
such as NIC frame processing latencies or buffering (see Net.host.interface).
Attributes Net.link.cidr, and Net.link.ip are explained in the
section on CIDR IP address blocking and
section on IP address assignment, respectively.
It is recommended not to use them and let SSF.Net.Net automatically allocate IP addresses
unless you know what you are doing.
Attribute Net.link.alignment in SSFNet 1.1 causes different synchronization
behavior for point-to-point links (class SSF.Net.ptpLinkLayer) and for LANs, i.e., links
with three or more attached interfaces (class SSF.Net.lanLinkLayer). The former does not
contribute to the multi-timeline synchronization, while the latter does (via the link
delay attribute).
|
randomstream [
generator %S
stream %S
reproducibility_level %S
]
|
Net.randomstream
attribute provides centralized control of the management and seeding of parallel random number
streams, and degree of sharing of random number streams by consumers of the random numbers
(i.e., RandomDistributions). This is critically important in a good quality simulation.
Random numbers are needed by TCP to randomize the offsets
of TCP timers and by other ProtocolSessions.
generator string is the name of a pseudorandom number generator distributed with SSFNet.
The names you can choose are highest-quality generators "MersenneTwister" and "Ranlux" from packages
cern.jet.random and edu.cornell.lassp.houle.RngPack distributed with SSFNet.
You can also choose name "Java" (which is java.util.Random).
stream value is an arbitrary user-defined string.
The seeds for the random number generators are formed from such a root string, concatenated with
other SSFNet-provided strings,
using the cryptographic message digest techniques (currently the MD5 function).
Changing this root string
automatically and deterministically changes the seeds in all instantiated
generators.
reproducibility_level controls how many separate instances of a random
number generator will be created in the running model, and how the output
of these generators will be shared by various consumers of random numbers.
The name "reproducibility_level" refers to the fact that with this attribute a user
may either guarantee that the result of simulation will be exactly the same regardless
of how many processors are used in a parallel simulation, or is willing to
sacrifice that strict reproducibility for sake of performance (while all generators in SSFNet are
fast, computing random numbers is not free and uses memory, and you may not want
the penalty of having thousands of generator instances in a large model.)
The available string values of reproducibility_level in the order
from most shared generators to least shared generators (one per instance of RandomDistribution) are:
- timeline:
all RandomDistributions instantiated in Entities aligned to a timeline
share the same random number stream. Different timelines have independent
and idependently seeded random number streams ( = generators). For a sequential
execution (default, when no alignment attributes are used, as in this example)
there will be just one instance of a named random number generator, whose output
is shared by all instances of RandomDistributions (as said, in this tiny model
they are buried in the TCP code).
- host:
all RandomDistributions instantiated in a Host share
the same random number stream. Different Hosts have independent
and idependently seeded random number streams ( = generators).
- protocol:
every instance of a ProtocolSession owns an independent
random number stream. Different ProtocolSessions have independent
and idependently seeded random number streams ( = generators).
- distribution:
every instance of a RandomDistributions owns an independent
random number stream. Different RandomDistributions have independent
and idependently seeded random number streams ( = generators).
See SSFNet/src/SSF/Util for the source code of random distributions. See the SSF.Net.Net source
code for how the seeds are formed and distributed to individual generators.
|
traffic [
pattern [
client %S1!
servers [
list %S
nhi %S
nhi_range [from %S1! to %S1!]
port %I1!
]
]
]
] # end of .schemas.Net
] # end of .schemas
|
Toplevel .Net.traffic attribute is used in SSFNet 1.1 and later to specify
the global traffic scenarios
for workload generation by client/server application level ProtocolSessions.
All SSFNet 1.x application-level client protocols look up this attribute
to find which servers they may connect to.
Used by classes SSF.OS.TCP.test.tcpClient, SSF.OS.UDP.test.udpStreamClient, SSF.OS.WWW.httpClient.
Within the traffic attribute, there may be an arbitrary number of pattern
subattributes.
Within the pattern attribute, there must be a single client attribute,
and one or more of servers attributes.
Each pattern specifies:
Each pattern may have many servers,
and it is up to the client implementation which servers it will choose and at which times.
The optional attribute servers.list introduced in ssfnet_1.3
can be matched by tcpClient or httpClient
attribute server_list
to narrow down a list of servers, and - more importantly - to allow
a client application to distinguish multiple servers on a single host (Note
that tcpClient and httpClient configuration does not specify the server port
number; instead a client searches the list of all servers in the
.Net.traffic attribute, and chooses the ones that match the client's
host NHI address and - if present - also match the server_list
string.
|
|
|
SSF.OS protocol DML configurations |
|---|
|
You should first read the tutorial on SSF.OS and protocol design.
Each contributed protocol (such as TCP, BGP4, OSPF, and various client/server application level protocols)
may define its own DML configuration attribute tree.
Protocol configuration attributes are read by a class extending ProtocolSession by its method
public void config(Configuration cfg) throws configException
that overloads the corresponding method of the SSF.OS.ProtocolSession.
In contrast to the rather stable SSF.Net.Net configuration, users keep adding new protocols and new versions of already distributed
protocols. Their DML configuration files may change more often, so consult the latest releases for
up-to-date DML configuration files.
Here we include the SSFNet release 1.1 DML configuration
file standards for SSF.OS.TCP, SSF.OS.UDP, SSF.OS.BGP, SSF.OS.OSPF, and generic network traffic DML
specifications that are implemented for various client/server implementations included in SSFNet 1.1 release.
The protocol DML configurations may appear in various places in user-defined DML files, depending on
whether each host has a distinct protocol configuration (then there would be one protocol configuration
attribute per host); or a group of hosts have the same protocol configuration (then
usually, but not necessarily, they appear under the toplevel .dictionary attribute), see the examples included
in the SSFNet release.
|
|
Standard configuration schema for class SSF.OS.IP
|
|
schema fragment |
explanation
|
ProtocolSession [for ip use SSF.OS.IP
debug %S
tiebreaker [
use %S
]
monitor [
use %S
]
]
|
The value of optional attribute debug is "true" or "false", if "true"
print verbose diagnostics; default is "false"
Attribute tiebreaker.use names the class implementing the Java interface SSF.Net.RouteTieBreaker
that should be loaded to choose among routes of equal cost.
If tiebreaker is omitted, the default is a simple hash on src and
dest addresses provided in an anonymous inner class in SSF.Net.RadixTreeRoutingTable.
Optional attribute monitor specifies to use class named in String %S for
the packet or flow monitor. A packet monitor must implement Java interface SSF.OS.IpMonitor.
If omitted, no monitor is installed.
|
|
|
Standard configuration schema for class SSF.OS.TCP.tcpSessionMaster
|
|
schema fragment |
explanation
|
ProtocolSession [name tcp use SSF.OS.TCP.tcpSessionMaster
tcpinit[
ISS %I # initial sequence number
MSS %I # maximum segment size
RcvWndSize %I # receive buffer size in units of MSS
SendWndSize %I # maximum send window size in units of MSS
SendBufferSize %I # send buffer size in units of MSS
MaxRexmitTimes %I # maximum number of retransmission times
TCP_SLOW_INTERVAL %F # granularity of TCP slow timer, sec
TCP_FAST_INTERVAL %F # granularity of TCP fast(delay-ack), sec
MSL %F # maximum segment lifetime, sec
MaxIdleTime %F # maximum idle time, sec
delayed_ack %F # delayed ack option, true or false
fast_recovery %S # use fast recovery algorithm, true/false
show_report %S # print summary connection report, true/false
]
debug %S # if true, dump verbose TCP diagnostics to files
# for session & host (see below), true/false
# dump filename prefixes - actual filenames end with "_hostID_flowID.out"
# for session info, and with "_hostID.out" for host info
rttdump %S # rtt dumpfile prefix (session)
cwnddump %S # cwnd dumpfile prefix (session)
rexdump %S # rexmit timer dumpfile prefix (session)
eventdump %S # dumpfile prefix for all events (session)
con_count %S # dumpfile prefix for number of connections (host)
rto_count %S # dumpfile prefix for timeout info (host)
]
|
See
SSF.OS.TCP documentation and validation pages for
detailed explanations.
|
|
|
Standard configuration schema for class SSF.OS.UDP.udpSessionMaster
|
|
schema fragment |
explanation
|
ProtocolSession [name udp use SSF.OS.UDP.udpSessionMaster
udpinit[
max_datagram_size %I # maximum UDP datagram payload size (bytes)
debug %S # verbose UDP session diagnostics, true/false
]
]
|
See method config() in class SSF.OS.UDP.udpSessionMaster.
|
|
|
Standard configuration schema for class SSF.OS.BGP4.BGPSession
|
|
schema fragment |
explanation
|
ProtocolSession [ name bgp use SSF.OS.BGP4.BGPSession
# Choice to configure BGP "automatically", using default rules.
# If omitted, the value of autoconfig is 'true'.
# If the value of autoconfig is `true', then absolutely no other
# configuration of the BGP protocol session may be done
# (any other attributes will be ignored).
# On the other hand, if the value of autoconfig is `false',
# then every other mandatory attribute must be explicitly
# configured, or else an error will be flagged.
autoconfig %S # (true|false)
# Two timers are set on a per-BGP speaker basis:
connretry_time %I1! # Connect Retry Interval, seconds
min_as_orig_time %I1! # Minimum AS Origination Interval, sec
# whether this BGP speaker is a route reflector
reflector %S #(true|false)
# any number of neighbors
neighbor [
# The `as' attribute must be the NHI prefix address of this
# neighbor's AS, i.e. the global id of the Net which is
# the AS boundary for this neighbor. If it is identical
# to the NHI prefix of the local BGP speaker's AS, then
# it is interpreted as an internal neighbor.
# Otherwise, interpreted as an external neighbor.
as %S1!
# For an external neighbor, the `address' attribute must
# be a NHI interface address relative to the neighbor's AS.
# For an internal neighbor it must be an NHI *host* address
# relative to the local AS.
address %S1!
# This is the NHI interface address, relative to the local
# AS, which the neighbor should use when sending messages
# to the local BGP speaker. Often an internal peer will use
# the address of a virtual (loopback) interface, while
# an external peer will typically use the address of a
# physical interface on a point-to-point link.
use_return_address %S1!
hold_time %I1! # Hold Timer Interval, seconds
keep_alive_time %I1! # Keep Alive Timer Interval, seconds
min_adver_time %I1! # Minimum Route Advertisement
# Interval, seconds
# The ibgp attribute is only meaningful if this is an internal
# neighbor and if the local BGP speaker is itself a route
# reflector. If that is the case, then this attribute indicates
# that this neighbor is either a reflector or client in the same
# IBGP cluster as the local BGP speaker.
ibgp %S # (reflector|client)
# policy specifications:
# exactly one inbound route filter:
infilter [
# any number of clauses (none => deny every route)
clause [
precedence %I1! # all clauses in the same rule must have
# unique values,
# starting at 1 and increasing incrementally
predicate [ # always exactly 1 pr clause
# any number of atoms (none => match any route)
atom [
attribute %S1! # (origin|as_path|nhi_path|next_hop|med|
# local_pref|atomic_agg|aggregator|
# community|originator_id| cluster_list)
matcher %S1! # regular expr. for matching the attribute
]
]
action [ # always exactly 1 pr clause
# primary action to permit or deny a route
primary %S1! # (permit|deny)
# zero or more atoms
atom [ # actions for path attribute manipulation
attribute %S1!: # (origin|as_path|next_hop|med|local_pref|
# atomic_agg|aggregator|
# community|originator_id|cluster_list)
type %S1! # (set|append|prepend) action to perform
value %S # one or more values used by the action
]
]
] # end of clause
] # end of bgp infilter rules
# exactly one outbound route filter:
outfilter [
... same attributes as for infilter,
... omitted for brevity
]
] # end of neighbor
] # end of ProtocolSession
|
Class SSF.OS.BGP4.BGPSession:
Each individual BGP protocol session (BGP speaker) must be configured
separately. That is, network attributes which apply to multiple BGP
speakers simultaneously (such as IBGP clusters) must be configured
individually at each speaker in such a way that they collectively
define the network attribute, just as done in the real world. The
configuration of an individual BGP speaker takes place within a
ProtocolSession attribute (refer to the SSFNet schema,
.schemas.Net.host.graph.ProtocolSession). Such a ProtocolSession
attribute must have a subattribute name with a value of bgp as well
as a subattribute use with a value of SSF.OS.BGP4.BGPSession.
autoconfig tells whether to configure BGP "automatically", using default rules. The
default value of this attribute, should it be omitted, is 'true'. This
greatly simplifies the DML code of models for which autoconfiguration
is used. If the value of autoconfig is `true', then absolutely no other
configuration of the BGP protocol session may be done (any other
attributes will be ignored). On the other hand, if the value of
autoconfig is `false', then every other attribute must be explicitly
configured, or else an error will be flagged.
See the documentation included in the package SSF.OS.BGP4 for
aditional explanation and latest additions/revisions.
|
Net [ # top-level Net enclosing the whole model
...
Net [
AS_status boundary
....
]
bgpoptions [
# Prints messages associated with the given
# validation test. This attribute is intended
# only for the DML files of validation tests.
validation_test %I
# Minimum wait time, in seconds, at the beginning
# of the simulation during which each BGP protocol
# session is inactive.
base_startup_wait %F
# The upper bound on a range of time, in seconds,
# to be used for adding jitter to the startup
# wait period on a per-BGP protocol session
# basis. The lower bound of the range is always
# 0.0. Use 0.0 for no jitter.
startup_jitter_bound %F
# Use NHI addressing in debugging output. This means
# that AS numbers, IP prefixes and IP addresses will
# be converted to their NHI address equivalents.
use_nhi_addressing %S # (true|false)
# show various identifying values for BGP routers
show_id_data %S # (true|false)
# show neighbor info for each BGP speaker
show_nb_info %S # (true|false)
# show the configuration of IBGP clusters
show_ibgp_clusters %S # (true|false)
# show manually (DML) configured BGP attributes
show_timer_config %S # (true|false)
# show jittered timer interval values at startup
show_jitter %S # (true|false)
# show when BGP Start events occur
show_start_event %S # (true|false)
# show when BGP Stop events occur
show_stop_event %S # (true|false)
# show when TransConnOpen events occur
show_transopen %S # (true|false)
# show when TransConnClose events occur
show_transclose %S # (true|false)
# show when TransConnOpenFail events occur
show_transfail %S # (true|false)
# show when TransFatalError events occur
show_transfatal %S # (true|false)
# show when peer connections are established
show_conn_estab %S # (true|false)
# show negotiated Hold Timer Interval values
show_hold_value %S # (true|false)
# show Keep Alive Timer Interval values in use
show_ka_value %S # (true|false)
# show when Open msgs are sent
show_snd_open %S # (true|false)
# show when Open msgs are received
show_rcv_open %S # (true|false)
# show when KeepAlive msgs are sent
show_snd_ka %S # (true|false)
# show when KeepAlive msgs are received
show_rcv_ka %S # (true|false)
# show when Update msgs are sent
show_snd_update %S # (true|false)
# show when Update msgs are received
show_rcv_update %S # (true|false)
# show when Notification msgs are sent
show_snd_notif %S # (true|false)
# show when Notification msgs are received
show_rcv_notif %S # (true|false)
# show when KeepAlive Timers are set
show_set_ka %S # (true|false)
# show when Hold Timers are set
show_set_hold %S # (true|false)
# show when MinAdver Timers are set
show_set_minadver %S # (true|false)
# show when KeepAlive Timers expire
show_ka_exp %S # (true|false)
# show when ConnectRetry Timers expire
show_connretry_exp %S # (true|false)
# show when Hold Timers expire
show_hold_exp %S # (true|false)
# show when MinAdver Timers expire
show_minadver_exp %S # (true|false)
# show steps in handle_update() execution
show_handle_update %S # (true|false)
# show when new routes are added to routing table
show_added_route %S # (true|false)
# show degree of preference whenever calculated
show_dop_calc %S # (true|false)
# show steps of Decision Process execution
show_dec_proc %S # (true|false)
# show when the external update process begins
show_ext_update %S # (true|false)
# show when routes are added to fwding table
show_fwd_table_add %S # (true|false)
# show when routes are removed from fwding table
show_fwd_table_rmv %S # (true|false)
# print fwd table each time BGP changes it
show_fwd_tables %S # (true|false)
# show msgs related to route aggregation
show_aggregation %S # (true|false)
# show when MinAdverTmr expires w/no msgs waiting
show_no_msg_waiting %S # (true|false)
# show when BGP FSM state changes
show_state_changes %S # (true|false)
# show when notable Sockets events occur
show_socket_events %S # (true|false)
# show when route reflection is performed
show_reflection %S # (true|false)
# show inbound policy rule filtering messages
show_in_policy %S # (true|false)
# show outbound policy rule filtering messages
show_out_policy %S # (true|false)
# print the Adj-RIBs-In each time it changes
show_ribs_in %S # (true|false)
# print the Loc-RIB each time it changes
show_loc_rib %S # (true|false)
# print the Adj-RIBs-Out each time it changes
show_ribs_out %S # (true|false)
# dump Adj-RIBs-In at simulation end
dump_ribs_in %S # (true|false)
# dump Loc-RIB at simulation end
dump_loc_rib %S # (true|false)
# dump Adj-RIBs-Out at simulation end
dump_ribs_out %S # (true|false)
# dump the forwarding table at simulation end
dump_fwd_tables %S # (true|false)
# dump the stability state at simulation end
dump_stability %S # (true|false)
]
]
|
Attribute AS_status boundary in an included Net signifies that it is
a Net enclosing an Autonomous System.
Optional attribute .Net.bgpoptions in the top-level Net is used
for diagnostic and informational purposes.
The bgpoptions attribute is a list of the DML attributes available for use
as BGP debugging, validation, and output options. Their values never change
the functional behavior of BGP itself. The bgpoptions attribute, if
utilized, must be an immediate subattribute of the top-level `Net' attribute.
The default value of each of the boolean BGP options is `false'.
|
|
|
Standard configuration schema for class SSF.OS.OSPF.sOSPF
|
|
schema fragment |
explanation
|
Net [
...
ospfoptions [
show_ifaces %S # show ospf interface data structure
show_AS_area %S # show AS and area number
show_rtr_type %S # show router types
show_hello_pkts %S # show hello packets and link connection
show_lsas %S # show lsa packets
show_lsdb %S # show link state databases
show_rtg_tbl %S # show routing tables
]
Net [
ospf_area %I
....
]
]
|
Optional attribute .Net.ospfoptions in the top-level Net is used
solely for debugging and informational purposes. All its subattributes have value "true" or "false",
and control the printing of diagnostics.
Attribute ospf_area in an included Net signifies that it encloses
an OSPF area. Release sOSPF_0.0.18 supports only one area 0 (backbone) and an ospf_area
attribute must appear together in the same Net where an
AS_status attribute appears. Thus currently we use OSPF as an internal routing protocol
within each AS:
Net [
AS_status boundary
ospf_area 0
...
]
This limitation will be removed in a future release.
|
|
|
Standard configuration schema for classes SSF.OS.TCP.test.tcpClient, Server
|
|
schema fragment |
explanation
|
ProtocolSession [ name client use SSF.OS.TCP.test.tcpClient
start_time 1.0 # earliest time to send request to server
start_window 1.0 # send request to server at randomly chosen time
# in interval [start_time, start_time+start_window]
file_size 10000000 # requested file size (payload bytes)
request_size 4 # client request datagram size (bytes)
show_report true # print client-server session summary report
debug false # print verbose client/server diagnostics
# new atributes in ssfnet_1.3 - backward compatible with ssfnet_1.2
------------------------------------------------------------------
server_list "mylist" # identifies server list(s) name in traffic pattern
# if omitted concatenate all server lists in
# matching pattern
# only one of either "file_size" or "random_file_size" must be used.
# It is an error to specify both "file_size" or "random_file_size"
# simultaneously.
# Generated random variate interpreted as number of payload bytes,
# truncated to the integer value > 1.
random_file_size [
distribution [
# distribution-dependent parameters are here.
# see SSF.Util.Random.RandomStream for 21 distinct distributions
name "Exponential" # distribution class name
lambda 0.001 # mean = 1/lambda
]
]
# if both "off_time" and "random_off_time" are omitted,
# client requests a file only once.
# It is an error to specify both "off_time" and "random_off_time"
# simultaneously.
# An off_time (fixed or random) specifies client sleep time (seconds)
# between completion of a file transfer request and the next request.
# off_time 10.0 # commented out because random_off_time
# is specified in this example
random_off_time [
distribution [
name "Exponential" # distribution class name
lambda 0.01 # mean = 1/lambda
# distribution-dependent params here...
]
]
# If connection request times out or a connection is dropped
# due to any error msg (peer reset, TCP abort, no route found, etc)
# then the client may send a file request again as follows:
# i) if "error_recovery" is omitted or "false", client suspends
# operation indefinitely (backward compatible with ssfnet1.2)
# ii) if "error_recovery true", then:
# - if both "off_time" and "random_off_time" are omitted,
# client suspends operation indefinitely
# - if either "off_time" or "random_off_time" are specified,
# client will try connecting to the same server after
# a delay equal to the next value of "off_time".
error_recovery true
]
ProtocolSession [
name server use SSF.OS.TCP.test.tcpServer
port 10 # server's well known port
client_limit 13 # max number of simultaneous established client
# connections, if omitted, default is unlimited.
# Connection requests arriving when limit is
# reached are dropped to time out.
queue_limit 7 # maximum size of pending connection request
# queue, Qlimit, if omitted default is 5.
# Determines the total number of pending
# connection requests (see SSF.OS.TCP.tcpSocket).
# If exceeded, the listening socket sends RESET
# to client and drops the request.
request_size 4 # client request datagram size (bytes)
show_report true # print client-server session summary report
debug false # print verbose client/server diagnostics
]
|
See method config() in classes SSF.OS.TCP.test.tcpClient, tcpServer.
|
|
|
Standard configuration schema for classes SSF.OS.UDP.test.udpStreamClient, Server
|
|
schema fragment |
explanation
|
ProtocolSession [ name client use SSF.OS.UDP.test.udpStreamClient
start_time 1.0 # earliest time to send request to server
start_window 1.0 # send request to server at randomly chosen time
# in interval [start_time, start_time+start_window]
file_size 10000000 # requested file size (payload bytes)
request_size 4 # client request datagram size (bytes)
datagram_size 1000 # server datagram payload size (bytes)
show_report true # print client-server session summary report
debug false # print verbose client/server diagnostics
]
ProtocolSession [ name server use SSF.OS.UDP.test.udpStreamServer
port 10 # server's well known port
client_limit 10 # max number of contemporaneously allowed clients
# if omitted, default is no limit
request_size 4 # client request datagram size (bytes)
datagram_size 1000 # server datagram payload size (bytes)
send_interval 0.1 # time interval (in sec) between consecutive datagrams,
# if undefined, datagrams are sent back to back
show_report true # print client-server session summary report
debug false # print verbose client/server diagnostics
]
|
Trivial example of a streaming server.
See method config() in classes SSF.OS.UDP.test.udpStreamClient, udpStreamServer.
|
|
|
Standard configuration schema for classes SSF.OS.WWW.httpClient, Server
|
|
schema fragment |
explanation
|
ProtocolSession [name client use SSF.OS.WWW.httpClient
# earliest time to send the first request to server (seconds).
start_time 1.0
# HTTP session arrival process in this client. The first session
# begins at start_time + inter_session_time (in seconds).
# Each session is with a server chosen uniformly at random from
# the servers specified in the .Net.traffic for this client.
inter_session_time [
distribution [
name "Exponential" # distribution class name
lambda 0.01 # mean = 1/lambda
]
]
# distribution of number of pages per session with selected server.
# if random variable = 0, round up to 1
pages_in_session [
distribution [
name "Exponential" # distribution class name
lambda 0.2 # mean = 1/lambda
]
]
# distribution of user-think-times (off-times) between consecutive
# in-session page requests, counted from the time when the previous
# page delivery is completed (in seconds).
inter_page_time [
distribution [
name "Pareto" # distribution class name
k 25.0 # scale (cutoff) parameter
alpha 2.0 # shape (exponent) parameter
]
]
# option to pipeline requests (i.e. send before waiting for
# completion of previous request
# pipeline_requests false NOT IMPLEMENTED
# distribution of delays between consecutive requests in a session.
# If pipeline_requests = true, counted from the time of previous request;
# If pipeline_requests = false, counted from the time when the previous
# request was completed. (In seconds.)
inter_request_time [
distribution [
name "Pareto" # distribution class name
k 0.1667 # scale (cutoff) parameter
alpha 1.5 # shape (exponent) parameter
]
]
http_hdr_size 1000 # nominal HTTP header size (virtual bytes)
# read from a socket before reading data
persistent true # true: one persistent TCP connection
# per session (HTTP1.1),
# false: one TCP connection for each
# request/response (HTTP1.0).
show_report true # print client-server session summary report
debug false # print verbose client/server diagnostics
# new atributes in ssfnet_1.3 - backward compatible with ssfnet_1.2
------------------------------------------------------------------
server_list "mylist" # identifies server list(s) name in traffic pattern
# if omitted concatenate all server lists in
# matching pattern
]
|
See method config() in class SSF.OS.UDP.WWW.httpClient.
|
ProtocolSession [name server use SSF.OS.WWW.httpServer
port 80 # server's HTTP port
client_limit 10 # max number of contemporaneously allowed clients
# if omitted, default is unlimited number
# distribution of the number of objects in a page request:
# if random variable = 0, round up to 1
# NOTE: the response to page request contains the 1st object
objects_in_page [
distribution [
name "Pareto" # distribution class name
k 0.6667 # scale (cutoff) parameter
alpha 1.2 # shape (exponent) parameter
]
]
# distribution of sizes of objects (in bytes)
# if random variable = 0, round up to 1
object_size [
distribution [
name "Pareto" # distribution class name
k 2000.0 # scale (cutoff) parameter
alpha 1.2 # shape (exponent) parameter
]
]
# distribution of server delays between request and response (in seconds)
response_delay [
distribution [
name "Exponential" # distribution class name
lambda 0.01 # mean = 1/lambda
]
]
http_hdr_size 1000 # nominal HTTP header size (virtual bytes)
show_report true # print client-server session summary report
debug false # print verbose client/server diagnostics
]
|
See method config() in class SSF.OS.UDP.WWW.httpServer.
|
|
|
SSFNet component addressing |
|---|
|
SSFNET supports multiple addressing and numbering schemes for
networks, hosts/routers, and interfaces.
The sections below describe their implementation in SSFNet 1.1.
|
|
Topological NHI addressing of Nets, Hosts, Interfaces n:n:n:h(i) |
|---|
|
The basis of all addressing schemes is network topology, and in SSFNET,
NHI (Net, Host, Interface) strings are the basic notation for
identifying nodes and edges in the network graph.
Each Net, Host, Router, and NIC must either specify an integer ID (the
id
attribute) or an integer ID range (the idrange attribute).
At a given level (e.g., within a given Net) the IDs and ID ranges of
sub-Nets may not overlap; they need not be compact, however.
The same is true of Hosts and Routers, and of NICs.
The global NHI address of a network is created by concatenating the ID
of each network from the outermost to the innermost, separated by colons
(":").
Similarly, the global NHI address of a host or router is created by
concatenating the global NHI address of its containing network with the
Host ID, separated by a colon. Finally, the NHI address of a NIC is
created by adding the NIC ID, in parenthesis, to the containing Host NIC
address.
For example, in this nested Net:
Net [ id 3
Net [id 7
Net [id 8
Host [id 9 ...
interface [id 2]
]
]
]
]
|
The outmost Net has NHI address 3, the next Net has address 3:7,
and the innermost Net has address 3:7:8. The Host has NHI address
3:7:8:9, and the interface has NHI address 3:7:8:9(2).
NHI numbers are manually specified by the modeler's use of ID
attributes. The ids have integer values that need not be
sequentially assigned, although it is strongly recommended. The component
link does not have the id attribute.
The IDs are never automatically generated; omitting an id or idrange
attribute at any level other than the topmost level is an error. As per
the Net schema, the root Net should not have an id, idrange, or cidr
address specified; only imported Nets acquire these.
The id attributes for Nets and for hosts and routers belong to the same
address space, thus should be all distinct.
Within each host or router instance the interface.id integer values uniquely
identify network interfaces.
The above definitions of id attributes permit to compose bigger networks
from instances of Net without the need to re-number their internal components'
ids.
Note that when a Net is included in another Net, each of its NHI identifiers
simply acquires one more N: at the front of the N:N:...:N:H(I) string,
when it is addressed from the importing Net. More examples:
Example 1: in the simplest case of a Net containing only hosts, routers and
links, use H(I) only, e.g. 33(2) for host 33, interface 2.
Example 2: if a Net contains two Nets numbered 1, 2, then interface 1
on host 7 in Net 1 is 1:7(1), and interface 1 on host 7 in Net 2 is
2:7(1). The corresponding hosts are 1:7 and 2:7, respectively.
Example 3: If Net 1 in Example 2 is further composed of Nets numbered 1,2,3,4,
then an interface belonging to lowest level Net 4 could be identified
in the master Net as 1:4:98(12), or in words: interface 12 of router 98
in Net 4 in Net 1".
|
|
CIDR hierarchy addressing i/j/k/l |
|---|
|
Each Link and Net (other than the root (toplevel) Net) in a DML model has associated with it a
CIDR block address, consisting of 1 or more integers separated by slashes ("/").
If cidr attributes are entirely omitted, they will be automatically computed
by SSF.Net.Net as follows: links and Nets are assigned CIDR block addresses relative to that
of the Net in which they are defined, starting from CIDR block zero.
Simply speaking, cidr attributes specify a logical tree of subblocks of the toplevel IP address
block allocated to a model. The values of cidr attributes are slash-separated integers
similarly to a UNIX directory tree, and thus
manually specified CIDR addresses may be multilevel: e.g.,
link [ cidr 1/2/4 ...] means "this network link shall get IP addresses from block 4 in block 2 in block 1".
No two Nets or links may manually specify the same CIDR address; this is
flagged as an error. In general, modelers should defer to the framework
for automatic CIDR assignment unless they know what they're doing: It is
easy to make errors,
and that feature should be used with caution. The manually assigned CIDR attributes
are not checked for consistency in SSFNet 1.x.
In the future such checking may be a job for an offline design tool.
Read about Variable Length Subnet Mask
IP address assignment,
and see an example in a
mini-tutorial on cidr
attributes in DML networks.
|
|
IP addressing a.b.c.d/p |
|---|
|
The SSFNet version 1.x supports Variable Length Subnet Mask (VLSM) IP address
assignement method. The IP addresses may be allocated in either of
three mutually exclusive ways:
Manually, using attributes Net.ip, Net.link.ip
and Net.host.interface.ip, Net.router.interface.ip.
Use with caution, as manually assigned IP addresses are not verified for
consistency in this release.
In the future this may be a job for an offline validation tool, not the general-purpose
simulator.
Using the optional attributes Net.cidr and
Net.link.cidr
to guide the IP address assignement algorithm implemented in package SSF.Net.
Fully automatically if neither cidr nor ip
attributes are provided.
Read about Variable Length Subnet Mask
IP address assignment,
and see an example in a
mini-tutorial on cidr
attributes in DML networks.
SSFNET 1.1 allows to assign only one IP address to a network interface.
Future releases may allow multiple IP attributes, as in this example:
interface [id 1 ip "10.0.0.0/32" ip "128.1.2.3/32" ....]
but the current code will not handle this correctly. It will complain
up front because it expects to find a single value for "ip."
|
|
Autonomous Systems numbers |
|---|
|
Net.AS_status
While not a full addressing scheme, SSFNet supports the automatic
assignment of identifier numbers to autonomous systems. An included Net that specifies
the AS_status boundary attribute specifies that it is the enclosing Net of an autonomous
system, and is assigned the next free autonomous system number, starting from
one. It's an error to specify that a Net is the root of an autonomous system
if it is contained within another Net that is the root of an autonomous
system.
|
|
SSFNet 1.1 implementation of addressing |
|---|
|
The SSF.Net.Util.cidrBlock class implements NHI, CIDR, IP, and AS
addressing, and provides convenience methods to translate among these addressing
styles.
Currently the logic of SSF.Net is as follows: if you omit cidr attributes,
it makes default cidr blocks automatically;
if you omit ip attributes, it uses cidr (provided or computed)
to make ip;
if you supply ip, it ignores cidr if also provided.
SSF.Net.Util.cidrBlock now supports "strict" rules for CIDR and IP
exclusion (all or nothing, mutual exclusivity). The SSFNet default behavior
is "strict" when
enforcing the rules about CIDR and IP manual assignment;
the "-relax" flag to turn off strictness is a command line option.
(Java SSF).
SSFNET enforces the rule that top-level Net may not define id,
idrange, or cidr attributes.
Suppose a user specifies neither cidr nor ip
attributes in the master Net file.
Then each link within a Net is assigned an independent
CIDR block (there is no default aggregation), each imported Net is assigned
an independent CIDR block (there is no default aggregation); and the program
will recursively count attached interfaces and find the minimum size of
assigned cidr block that is sufficient to accomodate all interfaces.
Sensible Net/subNet design is all that's needed to do reasonable network partitioning and
address assignment, and first-time users may wish to stick to that basis.
Then if a user wishes to be clever, she will purposefully design recursive
Net imports to achieve desired address aggregation.
There is no need to ever specify CIDR attributes if Net import is used to
build the desired CIDR nesting structure implicitly and automatically.
Tree Linkage
Each node in the cidrBlock tree represents one CIDR block, and contains
three hashtables:
-- subblocks, mapping integers to corresponding CIDR sub-blocks
-- subnets, mapping integers to the CIDR blocks corresponding to
the given NHI sub-net
-- interfaces, mapping NHI host/interface addresses (H(I)) to
attachment specifications.
Note that the subnet tree is embedded within the subblock tree, derived
by collapsing edges corresponding to 'placeholder' CIDR blocks. A
placeholder CIDR block is created when the user manually specifies a
multilevel CIDR address for a subnet.
For example, the following net:
Net [
Net [id 0 cidr 4]
Net [id 1 cidr 5/1/4
Net [id 8 cidr 3]
]
]
|
yields the following subblock tree:
(-- (4) (5 (1 (4 (3)))))
and the following subnet tree:
(-- (0) ( 1 (8)))
Note that the subnets of the unnumbered root CIDR block correspond to
the two sub-Nets with NHI addresses "0" and "1", omitting the CIDR
placeholder blocks corresponding to the CIDR addresses "5" and "5/1".
Each cidrBlock B holds back-references to its CIDR parent and its NHI
parent (CP.subblocks includes B, and NP.subnets includes B).
Construction
During construction of a cidrBlock, only the NHI address is considered,
and the new block is entered into the subnets table of its parent.
After construction, the insertBlock method builds the CIDR address,
entering the new block into the subblocks table of its parent (possibly
introducing intermediate placeholder blocks if multilevel CIDR addresses
have been manually specified).
Finally, one pass from the root determines the autonomous system numbers,
another pass determines the IP address requirements, and a third pass
performs IP address assignment to each net and interface. These later
passes are skipped if the modeler supplies any manual IP attributes.
Interface summary at a glance
- see generated Javadocs in ssfnet/doc for details.
|
|
Important things about SSF.Net DML attributes |
|---|
|
1. Important rule 1: How to verify the correctness of _extends
and _find attribute values
that use various user-defined DML tuples to specify a Net component:
The DML schema checker expands all _find and
_extends attributes within
an attribute value, and if the expansion agrees with the master
schema for that attribute (optionally provided by the _schema attribute
value) then all is OK.
2. Important rule 2: when a Net includes a Net, NOTHING gets rewritten in
the imported DML Net tuple.
4. Note that Net.link does not have the id nor idrange attribute.
5. To attach a link to an imported Net's host or router, it must already have
an unattached interface.
6. Cannot attach an interface to included Net's link (meant as a LAN) - it
violates rule 2.
7. VLSM/CIDR partitions are imported with Nets. Only p2p links can cross Net
boundaries, and may be associated with either Net's subdivision CIDR block.
8. In general, if you're not a power user, you should NOT be specifying IP
routes. Always use NHI routes, because they are topologically defined and don't care about
actual IP addresses. Remember that IP routes are not closed
under Net inclusion. IP routes are almost always an overspecification that
will prevent reuse of DML Net or host or router components.
|
