1 /*! \page platform Platform Description
5 In order to run any simulation, SimGrid must be provided with three things:
6 something to run (i.e., your code), a description of the platform on which you
7 want to simulate your application and lastly information about the deployment
8 process: Which process should be deployed to which processor/core?
10 For the last two items, there are essentially two possible ways you can provide
11 this information as an input:
12 \li You can program it, either by using the Lua console (
13 \ref MSG_Lua_funct) or, if you're using MSG, some of MSG's platform and
14 deployment functions (\ref msg_simulation). If you want to use this,
15 check the particular documentation. (You can also check the section
16 \ref pf_flexml_bypassing, however, this documentation is deprecated;
17 there is a new, but undocumented, way to do it properly).
18 \li You can use two XML files: one contains the platform description while
19 the other contains the deployment instructions.
21 For more information on SimGrid's deployment features, please refer to
22 the \ref deployment documentation.
24 The platform description may be intricate. This documentation is all
25 about how to write this file: The basic concepts are introduced. Furthermore,
26 advanced options are explained. Additionally, some hints and tips on how to
27 write a good platform description are given.
29 \section pf_overview Some words about XML and DTD
31 We chose to use XML not only because it's extensible but also because many
32 tools (and plugins for existing tools) are available that facilitate editing and
33 validating XML files. Furthermore, libraries that parse XML are often already
34 available and very well tested.
36 The XML checking is done based on the Document Type Definition (DTD) file,
38 <a href="http://simgrid.gforge.inria.fr/simgrid.dtd">http://simgrid.gforge.inria.fr/simgrid.dtd</a>.
40 If you read the DTD, you should notice the following:
41 \li The platform tags contain a version attribute; the current version is 3.
42 This property might be used in the future to provide backwards
44 \li The DTD contains definitions for the two files used by SimGrid (i.e.,
45 platform description and deployment).
47 \section pf_basics Basic concepts
49 Nowadays, the Internet is composed of a bunch of independently managed
50 networks. Within each of those networks, there are entry and exit
51 points (most of the time, you can both enter and exit through the same
52 point); this allows to leave the current network and reach other
53 networks, possibly even in other locations.
54 At the upper level, such a network is called
55 <b>Autonomous System (AS)</b>, while at the lower level it is named
56 sub-network, or LAN (local area network).
57 They are indeed autonomous: routing is defined
58 (within the limits of his network) by the administrator, and so, those
59 networks can operate without a connection to other
60 networks. So-called gateways allow you to go from one network to
61 another, if such a (physical) connection exists. Every node in one network
62 that can be directly reached (i.e., without traversing other nodes) from
63 another network is called a gateway.
64 Each autonomous system consists of equipment such as cables (network links),
65 routers and switches as well as computers.
67 The structure of the SimGrid platform description relies exactly on the same
68 concept as a real-life platform (see above). Every resource (computers,
69 network equipment etc.) belongs to an AS, which can be defined by using the
70 \<AS\> tag. Within an AS, the routing between its elements can be defined
71 abitrarily. There are several modes for routing, and exactly one mode must be
72 selected by specifying the routing attribute in the AS tag:
75 <AS id="AS0" routing="Full">
79 Other supported values for the routing attribute can be found below, Section
83 There is also the ``<route>`` tag; this tag takes two attributes, ``src`` (source)
84 and ``dst`` (destination). Both source and destination must be valid identifiers
85 for routers (these will be introduced later). Contained by the ``<route>`` are
86 network links; these links must be used in order to communicate from the source
87 to the destination specified in the tag. Hence, a route merely describes
88 how to reach a router from another router.
91 More information and (code-)examples can be found in the Section \ref pf_rm.
94 An AS can also contain one or more AS; this allows you to
95 define the hierarchy of your platform.
97 Within each AS, the following types of resources exist:
98 \li <b>host</b>: a hostmachine; contains processors/cores etc.
99 \li <b>router</b>: a router or a gateway.
100 \li <b>link</b>: a link that defines a connection between two (or
101 more) resources. Every link has a bandwidth and a latency.
102 \li <b>cluster</b>: like a real cluster, contains many hosts
103 interconnected by some dedicated network. Each cluster is itself an AS.
105 Between these elements, a routing has to be defined. The AS is
106 supposed to be Autonomous, hence this has to be done at the AS level. The AS
107 handles two different types of entities (<b>host/router</b> and
108 <b>AS</b>); you are responsible to define routes between those elements,
109 otherwise entities will be unconnected and therefore unreachable from other
110 entities. Although several algorithms for routing are built into SimGrid (see
111 \ref pf_rm), you might encounter a case where you want to define
112 routes manually (for instance, due to specific requirements of your
115 There are three tags to use:
116 \li <b>ASroute</b>: to define routes between two <b>AS</b>
117 \li <b>route</b>: to define routes between two <b>host/router</b>
118 \li <b>bypassRoute</b>: to define routes between two <b>AS</b> that
119 will bypass default routing (as specified by the ``routing`` attribute
120 supplied to ``<AS>``, see above).
122 Here is an illustration of these concepts:
124 ![An illustration of an AS hierarchy. Here, AS1 contains 5 other AS' who in turn may contain other AS' as well.](AS_hierarchy.png)
125 Circles represent processing units and squares represent network routers. Bold
126 lines represent communication links. AS2 models the core of a national
127 network interconnecting a small flat cluster (AS4) and a larger
128 hierarchical cluster (AS5), a subset of a LAN (AS6), and a set of peers
129 scattered around the world (AS7).
131 \section pf_pftags Resource description
133 \subsection pf_As Platform: The \<AS\> tag
135 The concept of an AS was already outlined above (Section \ref pf_basics);
136 recall that the AS is so important because it groups other resources (such
137 as routers/hosts) together (in fact, these resources must be contained by
140 Available attributes :
142 Attribute name | Mandatory | Values | Description
143 --------------- | --------- | ------ | -----------
144 id | yes | String | The identifier of an AS; facilitates referring to this AS. ID must be unique.
145 routing | yes | Full\| Floyd\| Dijkstra\| DijkstraCache\| none\| Vivaldi\| Cluster | See Section \ref pf_rm for details.
150 <AS id="AS0" routing="Full">
151 <host id="host1" power="1000000000"/>
152 <host id="host2" power="1000000000"/>
153 <link id="link1" bandwidth="125000000" latency="0.000100"/>
154 <route src="host1" dst="host2"><link_ctn id="link1"/></route>
158 In this example, AS0 contains two hosts (host1 and host2). The route
159 between the hosts goes through link1.
161 \subsection pf_Cr Computing resources: hosts, clusters and peers.
163 \subsubsection pf_host The tag <host/>
165 A <b>host</b> represents a computer/node card. Every host is able to execute
166 code and it can send and receive data to/from other hosts. Most importantly,
167 a host can contain more than 1 core.
171 Attribute name | Mandatory | Values | Description
172 --------------- | --------- | ------ | -----------
173 id | yes | String | The identifier of the host. facilitates referring to this AS.
174 power | yes | double (must be > 0.0) | Computational power of every core of this host in FLOPS. Must be larger than 0.0.
175 core | no | int (Default: 1) | The number of cores of this host. If more than one core is specified, the "power" parameter refers to every core, i.e., the total computational power is #cores*power.<br /> If 6 cores are specified, up to 6 tasks can be executed without sharing the computational power; if more than 6 tasks are executed, computational power will be shared among these tasks. <br /> <b>Warning:</b> Although functional, this model was never scientifically assessed.
176 availability | no | int | <b>Specify if the percentage of power available.</b> (What? TODO)
177 availability_file| no | string | (Relative or absolute) filename to use as input; must contain availability traces for this host. The syntax of this file is defined below. <br /> <b>Note:</b> The filename must be specified with your system's format.
178 state | no | ON\|OFF<br/> (Default: ON) | Is this host running or not?
179 state_file | no | string | Same mechanism as availability_file.<br /> <b>Note:</b> The filename must be specified with your system's format.
180 coordinates | no | string | Must be provided when choosing the Vivaldi, coordinate-based routing model for the AS the host belongs to. More details can be found in the Section \ref pf_P2P_tags.
182 ### Possible children: ###
184 Tag name | Description | Documentation
185 ------------ | ----------- | -------------
186 <mount/> | Defines mounting points between some storage resource and the host. | \ref pf_sto_mo
187 <prop/> | The prop tag allows you to define additional information on this host following the attribute/value schema. You may want to use it to give information to the tool you use for rendering your simulation, for example. | N/A
192 <host id="host1" power="1000000000"/>
193 <host id="host2" power="1000000000">
194 <prop id="color" value="blue"/>
195 <prop id="rendershape" value="square"/>
200 \anchor pf_host_dynamism
201 ### Expressing dynamism ###
203 SimGrid provides mechanisms to change a hosts' availability over
204 time, using the ``availability_file`` attribute to the ``<host>`` tag
205 and a separate text file whose syntax is exemplified below.
207 #### Adding a trace file ####
210 <platform version="1">
211 <host id="bob" power="500000000" availability_file="bob.trace" />
215 #### Example of "bob.trace" file ####
224 Let us begin to explain this example by looking at line 2. (Line 1 will become clear soon).
225 The first column describes points in time, in this case, time 0. The second column
226 describes the relative amount of power this host is able to deliver (relative
227 to the maximum performance specified in the ``<host>`` tag). (Clearly, the
228 second column needs to contain values that are not smaller than 0 and not larger than 1).
229 In this example, our host will deliver 500 Mflop/s at time 0, as 500 Mflop/s is the
230 maximum performance of this host. At time 11.0, it will
231 deliver half of its maximum performance, i.e., 250 Mflop/s until time 20.0 when it will
232 will start delivering 80\% of its power. In this example, this amounts to 400 Mflop/s.
234 Since the periodicity in line 1 was set to be 1.0, i.e., 1 timestep, this host will
235 continue to provide 500 Mflop/s from time 21. From time 32 it will provide 250 MFlop/s and so on.
237 ### Changing initial state ###
239 It is also possible to specify whether the host is up or down by setting the
240 ``state`` attribute to either <b>ON</b> (default value) or <b>OFF</b>.
242 #### Example: Expliciting the default value "ON" ####
245 <platform version="1">
246 <host id="bob" power="500000000" state="ON" />
250 If you want this host to be unavailable, simply substitute ON with OFF.
252 ### Expressing churn ###
254 To express the fact that a host can change state over time (as in P2P
255 systems, for instance), it is possible to use a file describing the time
256 at which the host is turned on or off. An example of the content
257 of such a file is presented below.
259 #### Adding a state file ####
262 <platform version="1">
263 <host id="bob" power="500000000" state_file="bob.fail" />
267 #### Example of "bob.fail" file ####
275 A negative value means <b>down</b> (i.e., OFF) while a positive one means <b>up and
276 running</b> (i.e., ON). From time 0.0 to time 1.0, the host is on. At time 1.0, it is
277 turned off and at time 2.0, it is turned on again until time 12 (2.0 plus the
278 periodicity 10.0). It will be turned on again at time 13.0 until time 23.0, and
283 \subsubsection pf_cluster <cluster>
285 ``<cluster />`` represents a machine-cluster. It is most commonly used
286 when one wants to define many hosts and a network quickly. Technically,
287 ``cluster`` is a meta-tag: <b>from the inner SimGrid point of
288 view, a cluster is an AS where some optimized routing is defined</b>.
289 The default inner organization of the cluster is as follow:
295 ____________|__________|_____________ backbone
297 l0| l1| l2| l97| l96 | | l99
303 Here, a set of <b>host</b>s is defined. Each of them has a <b>link</b>
304 to a central backbone (backbone is a link itself, as a link can
305 be used to represent a switch, see the switch / link section
306 below for more details about it). A <b>router</b> allows to connect a
307 <b>cluster</b> to the outside world. Internally,
308 SimGrid treats a cluster as an AS containing all hosts: the router is the default
309 gateway for the cluster.
311 There is an alternative organization, which is as follows:
325 The principle is the same, except that there is no backbone. This representation
326 can be obtained easily: just do not set the bb_* attributes.
329 Attribute name | Mandatory | Values | Description
330 --------------- | --------- | ------ | -----------
331 id | yes | string | The identifier of the cluster. Facilitates referring to this cluster.
332 prefix | yes | string | Each node of the cluster has to have a name. This name will be prefixed with this prefix.
333 suffix | yes | string | Each node of the cluster will be suffixed with this suffix
334 radical | yes | string | Regexp used to generate cluster nodes name. Syntax: "10-20" will give you 11 machines numbered from 10 to 20, "10-20;2" will give you 12 machines, one with the number 2, others numbered as before. The produced number is concatenated between prefix and suffix to form machine names.
335 power | yes | int | Same as the ``power`` attribute of the ``<host>`` tag.
336 core | no | int (default: 1) | Same as the ``core`` attribute of the ``<host>`` tag.
337 bw | yes | int | Bandwidth for the links between nodes and backbone (if any). <b>See <b>link</b> section for syntax/details.</b>
338 lat | yes | int | Latency for the links between nodes and backbone (if any). See <b>link</b> section for syntax/details.
339 sharing_policy | no | string | Sharing policy for the links between nodes and backbone (if any). See <b>link</b> section for syntax/details.
340 bb_bw | no | int | Bandwidth for backbone (if any). See <b>link</b> section for syntax/details. If bb_bw and bb_lat (see below) attributes are omitted, no backbone is created (alternative cluster architecture <b>described before</b>).
341 bb_lat | no | int | Latency for backbone (if any). See <b>link</b> section for syntax/details. If bb_lat and bb_bw (see above) attributes are omitted, no backbone is created (alternative cluster architecture <b>described before</b>).
342 bb_sharing_policy | no | string | Sharing policy for the backbone (if any). See <b>link</b> section for syntax/details.
343 availability_file | no | string | Allows you to use a file as input for availability. Similar to <b>hosts</b> attribute.
344 state_file | no | string | Allows you to use a file as input for states. Similar to <b>hosts</b> attribute.
345 loopback_bw | no | int | Bandwidth for loopback (if any). See <b>link</b> section for syntax/details. If loopback_bw and loopback_lat (see below) attributes are omitted, no loopback link is created and all intra-node communication will use the main network link of the node. Loopback link is a <b>FATPIPE</b>.
346 loopback_lat | no | int | Latency for loopback (if any). See <b>link</b> section for syntax/details. See loopback_bw for more info.
347 topology | no | FLAT\|TORUS\|FAT_TREE (default: FLAT) | Network topology to use. SimGrid currently supports FLAT (with or without backbone, as described before), <a href="http://en.wikipedia.org/wiki/Torus_interconnect">TORUS </a> and FAT_TREE attributes for this tag.
348 topo_parameters | no | string | Specific parameters to pass for the topology defined in the topology tag. For torus networks, comma-separated list of the number of nodes in each dimension of the torus. For fat trees, refer to \ref AsClusterFatTree "AsClusterFatTree documentation".
352 the router name is defined as the resulting String in the following
356 router_name = prefix + clusterId + _router + suffix;
360 #### Cluster example ####
362 Consider the following two (and independent) uses of the ``cluster`` tag:
365 <cluster id="my_cluster_1" prefix="" suffix="" radical="0-262144"
366 power="1e9" bw="125e6" lat="5E-5"/>
368 <cluster id="my_cluster_2" prefix="c-" suffix=".me" radical="0-99"
369 power="1e9" bw="125e6" lat="5E-5"
370 bb_bw="2.25e9" bb_lat="5E-4"/>
373 The second example creates one router and 100 machines with the following names:
375 c-my_cluster_2_router.me
383 \subsubsection pf_peer <peer/>
385 This tag represents a peer, as in Peer-to-Peer (P2P) networks. However, internally,
386 SimGrid transforms a peer into an AS (similar to Cluster). Hence, this tag
387 is virtually only a shortcut that comes with some pre-defined resources
388 and values. These are:
390 \li A tiny AS whose routing type is cluster is created
392 \li Two links: One for download and one for upload. This is
393 convenient to use and simulate stuff under the last mile model (e.g., ADSL peers).
394 \li It connects the two links to the host
395 \li It creates a router (a gateway) that serves as an entry point for this peer zone.
396 This router has coordinates.
400 Attribute name | Mandatory | Values | Description
401 --------------- | --------- | ------ | -----------
402 id | yes | string | The identifier of the peer. Facilitates referring to this peer.
403 power | yes | int | See the description of the ``host`` tag for this attribute
404 bw_in | yes | int | Bandwidth downstream
405 bw_out | yes | int | Bandwidth upstream
406 lat | yes | double | Latency for both up- and downstream, in seconds.
407 coordinates | no | string | Coordinates of the gateway for this peer. Example value: 12.8 14.4 6.4
408 sharing_policy | no | SHARED\|FULLDUPLEX (default: FULLDUPLEX) | Sharing policy for links. See <b>link</b> description for details.
409 availability_file| no | string | Availability file for the peer. Same as host availability file. See <b>host</b> description for details.
410 state_file | no | string | State file for the peer. Same as host state file. See <b>host</b> description for details.
412 Internally, SimGrid transforms any ``<peer/>`` construct such as
415 coordinates="12.8 14.4 6.4"
421 into an ``<AS>`` (see Sections \ref pf_basics and \ref pf_As). In fact, this example of the ``<peer/>`` tag
422 is completely equivalent to the following declaration:
425 <AS id="as_FOO" routing="Cluster">
426 <host id="peer_FOO" power="1.5Gf"/>
427 <link id="link_FOO_UP" bandwidth="2.25GBps" latency="500us"/>
428 <link id="link_FOO_DOWN" bandwidth="2.25GBps" latency="500us"/>
429 <router id="router_FOO" coordinates="25.5 9.4 1.4"/>
430 <host_link id="peer_FOO" up="link_FOO_UP" down="link_FOO_DOWN"/>
435 \subsection pf_ne Network equipments: links and routers
437 There are two tags available to represent network entities:
438 1. ``<link>``: Represents a entity that has a limited bandwidth, a
439 latency, and that can be shared according to TCP way to share this
442 The concept of links in SimGrid may not be intuitive, as links are not limited
443 to connecting (exactly) two entities; in fact, you can have more than two equipments
444 connected to it. (In graph theoretical terms: A link in SimGrid is not an edge,
448 2. ``<router/>``: Represents an entity that a message can be routed
449 to, but that is unable to execute any code. In SimGrid, routers have also
450 no impact on the performance: Routers do not limit any bandwidth nor
451 do they increase latency. As a matter of fact, routers are (almost) ignored
452 by the simulator when the simulation has begun.
455 If you want to represent an entity like a switch, you must use ``<link>`` (see section). Routers are used
456 to run some routing algorithm and determine routes (see Section \ref pf_routing for details).
459 \subsubsection pf_router <router/>
461 %As said before, <b>router</b> is used only to give some information
462 for routing algorithms. So, it does not have any attributes except :
466 Attribute name | Mandatory | Values | Description
467 --------------- | --------- | ------ | -----------
468 id | yes | string | The identifier of the router to be used when referring to it.
469 coordinates | yes | string | Must be provided when choosing the Vivaldi, coordinate-based routing model for the AS the router belongs to. More details can be found in the Section \ref pf_P2P_tags.
474 <router id="gw_dc1_horizdist"/>
477 \subsubsection pf_link <link>
479 Network links can represent one-hop network connections. They are
480 characterized by their id and their bandwidth; links can (but may not) be subject
485 Attribute name | Mandatory | Values | Description
486 --------------- | --------- | ------ | -----------
487 id | yes | string | The identifier of the link to be used when referring to it.
488 bandwidth | yes | int | Maximum bandwidth for this link, given in bytes/s
489 latency | no | double (default: 0.0) | Latency for this link.
490 sharing_policy | no | SHARED\|FATPIPE\|FULLDUPLEX (default: SHARED) | Sharing policy for the link.
491 state | no | ON\|OFF (default: ON) | Allows you to to turn this link on or off (working / not working)
492 bandwidth_file | no | string | Allows you to use a file as input for bandwidth.
493 latency_file | no | string | Allows you to use a file as input for latency.
494 state_file | no | string | Allows you to use a file as input for states.
497 #### Possible shortcuts for ``latency`` ####
499 When using the latency attribute, you can specify the latency by using the scientific
500 notation or by using common abbreviations. For instance, the following three tags
504 <link id="LINK1" bandwidth="125000000" latency="5E-6"/>
505 <link id="LINK1" bandwidth="125000000" latency="5us"/>
506 <link id="LINK1" bandwidth="125000000" latency="0.000005"/>
509 Here, the second tag uses "us", meaning "microseconds". Other shortcuts are:
511 Name | Abbreviation | Time (in seconds)
512 ---- | ------------ | -----------------
513 Week | w | 7 * 24 * 60 * 60
514 Day | d | 24 * 60 * 60
518 Millisecond | ms | 0.001 = 10^(-3)
519 Microsecond | us | 0.000001 = 10^(-6)
520 Nanosecond | ns | 0.000000001 = 10^(-9)
521 Picosecond | ps | 0.000000000001 = 10^(-12)
523 #### Sharing policy ####
525 By default a network link is SHARED, i.e., if two or more data flows go
526 through a link, the bandwidth is shared fairly among all data flows. This
527 is similar to the sharing policy TCP uses.
529 On the other hand, if a link is defined as a FATPIPE,
530 each flow going through this link will be provided with the complete bandwidth,
531 i.e., no sharing occurs and the bandwidth is only limiting each flow individually.
532 The complete bandwidth provided by this link in this mode
533 is ``#flows*bandwidth``, with at most ``bandwidth`` being available per flow.
535 Using the FATPIPE mode allows to model backbones that won't affect performance
538 The last mode available is FULLDUPLEX. This means that SimGrid will
539 automatically generate two links (one carrying the suffix _UP and the other the
540 suffix _DOWN) for each ``<link>`` tag. This models situations when the direction
541 of traffic is important.
544 Transfers from one side to the other will interact similarly as
545 TCP when ACK returning packets circulate on the other direction. More
546 discussion about it is available in the description of link_ctn description.
549 In other words: The SHARED policy defines a physical limit for the bandwidth.
550 The FATPIPE mode defines a limit for each application,
551 with no upper total limit.
554 Tip: By using the FATPIPE mode, you can model big backbones that
555 won't affect performance (except latency).
561 <link id="SWITCH" bandwidth="125000000" latency="5E-5" sharing_policy="FATPIPE" />
564 #### Expressing dynamism and failures ####
566 Similar to hosts, it is possible to declare links whose state, bandwidth
567 or latency changes over time (see Section \ref pf_hosts_dynamism for details).
569 In the case of network links, the ``bandwidth`` and ``latency`` attributes are
570 replaced by the ``bandwidth_file`` and ``latency_file`` attributes.
571 The following XML snippet demonstrates how to use this feature in the platform
572 file. The structure of the files "link1.bw" and "link1.lat" is shown below.
575 <link id="LINK1" state_file="link1.fail" bandwidth="80000000" latency=".0001" bandwidth_file="link1.bw" latency_file="link1.lat" />
579 Even if the syntax is the same, the semantic of bandwidth and latency
580 trace files differs from that of host availability files. For bandwidth and
581 latency, the corresponding files do not
582 express availability as a fraction of the available capacity but directly in
583 bytes per seconds for the bandwidth and in seconds for the latency. This is
584 because most tools allowing to capture traces on real platforms (such as NWS)
585 express their results this way.
588 ##### Example of "link1.bw" file #####
596 In this example, the bandwidth changes repeatedly, with all changes
597 being repeated every 12 seconds.
599 At the beginning of the the simulation, the link's bandwidth is 80,000,000
600 B/s (i.e., 80 Mb/s); this value was defined in the XML snippet above.
601 After four seconds, it drops to 40 Mb/s (line 2), and climbs
602 back to 60 Mb/s after another 4 seconds (line 3). The value does not change any
603 more until the end of the period, that is, after 12 seconds have been simulated).
604 At this point, periodicity kicks in and this behavior is repeated: Seconds
605 12-16 will experience 80 Mb/s, 16-20 40 Mb/s etc.).
607 ##### Example of "link1.lat" file #####
616 In this example, the latency varies with a period of 5 seconds.
617 In the xml snippet above, the latency is initialized to be 0.0001s (100µs). This
618 value will be kept during the first second, since the latency_file contains
619 changes to this value at second one, two and three.
620 At second one, the value will be 0.001, i.e., 1ms. One second later it will
621 be adjusted to 0.01 (or 10ms) and one second later it will be set again to 1ms. The
622 value will not change until second 5, when the periodicity defined in line 1
623 kicks in. It then loops back, starting at 100µs (the initial value) for one second.
626 #### The ``<prop/>`` tag ####
628 Similar to ``<host>``, the link may also contain the ``<prop/>`` tag; see the host
629 documentation (Section \ref pf_host) for an example.
634 \subsection pf_storage Storage
637 This is a prototype version that should evolve quickly, this
638 is just some doc valuable only at the time of writing this doc</b>
639 This section describes the storage management under SimGrid ; nowadays
640 it's only usable with MSG. It relies basically on linux-like concepts.
641 You also may want to have a look to its corresponding section in \ref
642 msg_file_management ; functions access are organized as a POSIX-like
646 \subsubsection pf_sto_conc Storage Main concepts
647 Basically there is 3 different entities to know :
648 \li the <b>storage_type</b>: here you define some kind of storage that
649 you will instantiate many type on your platform. Think of it like
650 a definition of throughput of a specific disk.
651 \li the <b>storage</b>: instance of a <b>storage_type</b>. Defines a
652 new storage of <b>storage_type</b>
653 \li the <b>mount</b>: says that the storage is located into this
656 the content of a storage has to be defined in a content file that
657 contains the content. The path to this file has to be passed within
658 the <b>content</b> attribute . Here is a way to generate it:
661 find /path/you/want -type f -exec ls -l {} \; 2>/dev/null > ./content.txt
664 \subsubsection pf_sto_sttp storage_type
667 <b>storage_type</b> attributes :
668 \li <b>id (mandatory)</b>: the identifier of the storage_type to be
669 used when referring to it.
670 \li <b>model (mandatory)</b>: Unused for now by the simulator (but
672 \li <b>content</b>: default value 0. The file containing the disk
673 content. (may be moved soon or later to <b>storage</b> tag.
675 The tag must contains some predefined model prop, as may do some other
677 <b>storage_type</b> mandatory <b>model_prop</b> :
678 \li <b>Bwrite</b>: value in B/s. Write throughput
679 \li <b>Bread</b>: value in B/s. Read throughput
680 \li <b>Bconnexion</b>: value in B/s. Connection throughput (i.e. the
681 throughput of the storage connector).
683 A storage_type can also contain the <b>prop</b> tag. The prop tag allows you
684 to define additional information on this storage_type following the
685 attribute/value schema. You may want to use it to give information to
686 the tool you use for rendering your simulation, for example.
689 <storage_type id="single_HDD" model="linear_no_lat" size="4000" content_type="txt_unix">
690 <model_prop id="Bwrite" value="30MBps" />
691 <model_prop id="Bread" value="100MBps" />
692 <model_prop id="Bconnection" value="150MBps" />
693 <b><prop id="Brand" value="Western Digital" /></b>
697 \subsubsection pf_sto_st storage
699 <b>storage_type</b> attributes :
700 \li <b>id (mandatory)</b>: the identifier of the storage to be used
701 when referring to it.
702 \li <b>typeId (mandatory)</b>: the identifier of the storage_type that
703 this storage belongs to.
704 \li <b>attach (mandatory)</b>: the host (name) to which the storage is
707 \subsubsection pf_sto_mo mount
709 <b>mount</b> attributes :
710 \li <b>id (mandatory)</b>: the id of the <b>storage</b> that must be
711 mounted on that computer.
712 \li <b>name (mandatory)</b>: the name that will be the logical
713 reference to this disk (the mount point).
715 \subsubsection pf_sto_mst mstorage
716 <b>Note : unused for now</b>
717 <b>mstorage</b> attributes :
718 \li <b>typeId (mandatory)</b>: the id of the <b>storage</b> that must
719 be mounted on that computer.
720 \li <b>name (mandatory)</b>: the name that will be the logical
721 reference to this disk (the mount point).
723 \section pf_routing Routing
725 To achieve high performance, the routing tables used within SimGrid are
726 static. This means that routing between two nodes is calculated once
727 and will not change during execution. The SimGrid team chose to use this
728 approach as it is rare to have a real deficiency of a resource;
729 most of the time, a communication fails because the links experience too much
730 congestion and hence, your connection stops before the timeout or
731 because the computer designated to be the destination of that message
734 We also chose to use shortest paths algorithms in order to emulate
735 routing. Doing so is consistent with the reality: RIP, OSPF, BGP are
736 all calculating shortest paths. They have some convergence time, but
737 at the end, so when the platform is stable (and this should be the
738 moment you want to simulate something using SimGrid) your packets will
739 follow the shortest paths.
741 \subsection pf_rm Routing models
743 Within each AS, you have to define a routing model to use. You have
744 basically 3 main kind of routing models :
746 \li Shortest-path based models: you let SimGrid calculates shortest
747 paths and manage it. Behaves more or less as most real life
749 \li Manually-entered route models: you'll have to define all routes
750 manually by yourself into the platform description file.
751 Consistent with some manually managed real life routing.
752 \li Simple/fast models: those models offers fast, low memory routing
753 algorithms. You should consider to use it if you can make some
754 assumptions about your AS. Routing in this case is more or less
757 \subsubsection pf_raf The router affair
759 Expressing routers becomes mandatory when using shortest-path based
760 models or when using ns-3 or the bindings to the GTNetS packet-level
761 simulator instead of the native analytical network model implemented
764 For graph-based shortest path algorithms, routers are mandatory,
765 because both algorithms need a graph, and so we need to have source
766 and destination for each edge.
768 Routers are naturally an important concept in GTNetS or ns-3 since the
769 way they run the packet routing algorithms is actually simulated.
770 Instead, the SimGrid’s analytical models aggregate the routing time
771 with the transfer time. Rebuilding a graph representation only from
772 the route information turns to be a very difficult task, because of
773 the missing information about how routes intersect. That is why we
774 introduced a \<router\> tag, which is simply used to express these
775 intersection points. The only attribute accepted by this tag an id. It
776 is important to understand that the \<router\> tag is only used to
777 provide topological information.
779 To express those topological information, some <b>route</b> have to be
780 defined saying which link is between which routers. Description or the
781 route syntax is given below, as well as example for the different
784 \subsubsection pf_rm_sh Shortest-path based models
786 Here is the complete list of such models, that computes routes using
787 classic shortest-paths algorithms. How to choose the best suited
788 algorithm is discussed later in the section devoted to it.
790 \li <b>Floyd</b>: Floyd routing data. Pre-calculates all routes once.
791 \li <b>Dijkstra</b>: Dijkstra routing data ,calculating routes when
793 \li <b>DijkstraCache</b>: Dijkstra routing data. Handle some cache for
794 already calculated routes.
796 All those shortest-path models are instanciated the same way. Here are
801 <AS id="AS0" routing="Floyd">
803 <cluster id="my_cluster_1" prefix="c-" suffix=""
804 radical="0-1" power="1000000000" bw="125000000" lat="5E-5"
805 router_id="router1"/>
807 <AS id="AS1" routing="none">
808 <host id="host1" power="1000000000"/>
811 <link id="link1" bandwidth="100000" latency="0.01"/>
813 <ASroute src="my_cluster_1" dst="AS1"
816 <link_ctn id="link1"/>
822 ASroute given at the end gives a topological information: link1 is
823 between router1 and host1.
827 <AS id="AS_2" routing="Dijsktra">
828 <host id="AS_2_host1" power="1000000000"/>
829 <host id="AS_2_host2" power="1000000000"/>
830 <host id="AS_2_host3" power="1000000000"/>
831 <link id="AS_2_link1" bandwidth="1250000000" latency="5E-4"/>
832 <link id="AS_2_link2" bandwidth="1250000000" latency="5E-4"/>
833 <link id="AS_2_link3" bandwidth="1250000000" latency="5E-4"/>
834 <link id="AS_2_link4" bandwidth="1250000000" latency="5E-4"/>
835 <router id="central_router"/>
836 <router id="AS_2_gateway"/>
837 <!-- routes providing topological information -->
838 <route src="central_router" dst="AS_2_host1"><link_ctn id="AS_2_link1"/></route>
839 <route src="central_router" dst="AS_2_host2"><link_ctn id="AS_2_link2"/></route>
840 <route src="central_router" dst="AS_2_host3"><link_ctn id="AS_2_link3"/></route>
841 <route src="central_router" dst="AS_2_gateway"><link_ctn id="AS_2_link4"/></route>
845 DijsktraCache example :
847 <AS id="AS_2" routing="DijsktraCache">
848 <host id="AS_2_host1" power="1000000000"/>
850 (platform unchanged compared to upper example)
853 \subsubsection pf_rm_me Manually-entered route models
855 \li <b>Full</b>: You have to enter all necessary routes manually
859 <AS id="AS0" routing="Full">
860 <host id="host1" power="1000000000"/>
861 <host id="host2" power="1000000000"/>
862 <link id="link1" bandwidth="125000000" latency="0.000100"/>
863 <route src="host1" dst="host2"><link_ctn id="link1"/></route>
867 \subsubsection pf_rm_sf Simple/fast models
869 \li <b>none</b>: No routing (Unless you know what you are doing, avoid
870 using this mode in combination with a non Constant network model).
873 <AS id="exitAS" routing="none">
874 <router id="exit_gateway"/>
877 \li <b>Vivaldi</b>: Vivaldi routing, so when you want to use
878 coordinates. See the corresponding section P2P below for details.
879 \li <b>Cluster</b>: Cluster routing, specific to cluster tag, should
880 not be used, except internally.
882 \subsection ps_dec Defining routes
884 The principle of route definition is the same for the 4 available tags
885 for doing it. Those for tags are:
887 \li <b>route</b>: to define route between host/router
888 \li <b>ASroute</b>: to define route between AS
889 \li <b>bypassRoute</b>: to bypass normal routes as calculated by the
890 network model between host/router
891 \li <b>bypassASroute</b>: same as bypassRoute, but for AS
893 Basically all those tags will contain an (ordered) list of references
894 to link that compose the route you want to define.
896 Consider the example below:
899 <route src="Alice" dst="Bob">
900 <link_ctn id="link1"/>
901 <link_ctn id="link2"/>
902 <link_ctn id="link3"/>
906 The route here from host Alice to Bob will be first link1, then link2,
907 and finally link3. What about the reverse route ? <b>route</b> and
908 <b>ASroute</b> have an optional attribute <b>symmetrical</b>, that can
909 be either YES or NO. YES means that the reverse route is the same
910 route in the inverse order, and is set to YES by default. Note that
911 this is not the case for bypass*Route, as it is more probable that you
912 want to bypass only one default route.
914 For an ASroute, things are just slightly more complicated, as you have
915 to give the id of the gateway which is inside the AS you're talking
916 about you want to access ... So it looks like this :
920 <ASroute src="AS1" dst="AS2"
921 gw_src="router1" gw_dst="router2">
922 <link_ctn id="link1"/>
926 gw == gateway, so when any message are trying to go from AS1 to AS2,
927 it means that it must pass through router1 to get out of the AS, then
928 pass through link1, and get into AS2 by being received by router2.
929 router1 must belong to AS1 and router2 must belong to AS2.
931 \subsubsection pf_linkctn link_ctn
933 a <b>link_ctn</b> is the tag that is used in order to reference a
934 <b>link</b> in a route. Its id is the link id it refers to.
936 <b>link_ctn</b> attributes :
937 \li <b>id (mandatory)</b>: Id of the link this tag refers to
938 \li <b>direction</b>: if the link referenced by <b>id</b> has been
939 declared as FULLDUPLEX, this is used to indicate in which
940 direction the route you're defining is going through this link.
941 Possible values "UP" or "DOWN".
943 \subsubsection pf_asro ASroute
945 ASroute tag purpose is to let people write manually their routes
946 between AS. It's useful when you're in Full model.
948 <b>ASroute</b> attributes :
949 \li <b>src (mandatory)</b>: the source AS id.
950 \li <b>dst (mandatory)</b>: the destination AS id.
951 \li <b>gw_src (mandatory)</b>: the gateway to be used within the AS.
952 Can be any <b>host</b> or \b router defined into the \b src AS or
953 into one of the AS it includes.
954 \li <b>gw_dst (mandatory)</b>: the gateway to be used within the AS.
955 Can be any <b>host</b> or \b router defined into the \b dst AS or
956 into one of the AS it includes.
957 \li <b>symmetrical</b>: if the route is symmetric, the reverse route
958 will be the opposite of the one defined. Can be either YES or NO,
961 <b>Example of ASroute with Full</b>
963 <AS id="AS0" routing="Full">
964 <cluster id="my_cluster_1" prefix="c-" suffix=".me"
965 radical="0-149" power="1000000000" bw="125000000" lat="5E-5"
966 bb_bw="2250000000" bb_lat="5E-4"/>
968 <cluster id="my_cluster_2" prefix="c-" suffix=".me"
969 radical="150-299" power="1000000000" bw="125000000" lat="5E-5"
970 bb_bw="2250000000" bb_lat="5E-4"/>
972 <link id="backbone" bandwidth="1250000000" latency="5E-4"/>
974 <ASroute src="my_cluster_1" dst="my_cluster_2"
975 gw_src="c-my_cluster_1_router.me"
976 gw_dst="c-my_cluster_2_router.me">
977 <link_ctn id="backbone"/>
979 <ASroute src="my_cluster_2" dst="my_cluster_1"
980 gw_src="c-my_cluster_2_router.me"
981 gw_dst="c-my_cluster_1_router.me">
982 <link_ctn id="backbone"/>
987 \subsubsection pf_ro route
988 The principle is the same as ASroute : <b>route</b> contains list of
989 links that are in the path between src and dst, except that it is for
990 routes between a src that can be either <b>host</b> or \b router and a
991 dst that can be either <b>host</b> or \b router. Useful for Full
992 as well as for the shortest-paths based models, where you
993 have to give topological information.
996 <b>route</b> attributes :
997 \li <b>src (mandatory)</b>: the source id.
998 \li <b>dst (mandatory)</b>: the destination id.
999 \li <b>symmetrical</b>: if the route is symmetric, the reverse route
1000 will be the opposite of the one defined. Can be either YES or NO,
1003 <b>route example in Full</b>
1005 <route src="Tremblay" dst="Bourassa">
1006 <link_ctn id="4"/><link_ctn id="3"/><link_ctn id="2"/><link_ctn id="0"/><link_ctn id="1"/><link_ctn id="6"/><link_ctn id="7"/>
1010 <b>route example in a shortest-path model</b>
1012 <route src="Tremblay" dst="Bourassa">
1016 Note that when using route to give topological information, you have
1017 to give routes with one link only in it, as SimGrid needs to know
1018 which host are at the end of the link.
1020 \subsubsection pf_byASro bypassASroute
1022 <b>Note : bypassASroute and bypassRoute are under rewriting to perform
1023 better ; so you may not use it yet</b> As said before, once you choose
1024 a model, it (if so) calculates routes for you. But maybe you want to
1025 define some of your routes, which will be specific. You may also want
1026 to bypass some routes defined in lower level AS at an upper stage :
1027 <b>bypassASroute</b> is the tag you're looking for. It allows to
1028 bypass routes defined between already defined between AS (if you want
1029 to bypass route for a specific host, you should just use byPassRoute).
1030 The principle is the same as ASroute : <b>bypassASroute</b> contains
1031 list of links that are in the path between src and dst.
1033 <b>bypassASroute</b> attributes :
1034 \li <b>src (mandatory)</b>: the source AS id.
1035 \li <b>dst (mandatory)</b>: the destination AS id.
1036 \li <b>gw_src (mandatory)</b>: the gateway to be used within the AS.
1037 Can be any <b>host</b> or \b router defined into the \b src AS or
1038 into one of the AS it includes.
1039 \li <b>gw_dst (mandatory)</b>: the gateway to be used within the AS.
1040 Can be any <b>host</b> or \b router defined into the \b dst AS or
1041 into one of the AS it includes.
1042 \li <b>symmetrical</b>: if the route is symmetric, the reverse route
1043 will be the opposite of the one defined. Can be either YES or NO,
1046 <b>bypassASroute Example</b>
1048 <bypassASRoute src="my_cluster_1" dst="my_cluster_2"
1049 gw_src="my_cluster_1_router"
1050 gw_dst="my_cluster_2_router">
1051 <link_ctn id="link_tmp"/>
1055 \subsubsection pf_byro bypassRoute
1056 <b>Note : bypassASRoute and bypassRoute are under rewriting to perform
1057 better ; so you may not use it yet</b> As said before, once you choose
1058 a model, it (if so) calculates routes for you. But maybe you want to
1059 define some of your routes, which will be specific. You may also want
1060 to bypass some routes defined in lower level AS at an upper stage :
1061 <b>bypassRoute</b> is the tag you're looking for. It allows to bypass
1062 routes defined between <b>host/router</b>. The principle is the same
1063 as route : <b>bypassRoute</b> contains list of links references of
1064 links that are in the path between src and dst.
1066 <b>bypassRoute</b> attributes :
1067 \li <b>src (mandatory)</b>: the source AS id.
1068 \li <b>dst (mandatory)</b>: the destination AS id.
1069 \li <b>symmetrical</b>: if the route is symmetric, the reverse route
1070 will be the opposite of the one defined. Can be either YES or NO,
1073 <b>bypassRoute Example</b>
1075 <bypassRoute src="host_1" dst="host_2">
1076 <link_ctn id="link_tmp"/>
1081 \subsection pb_baroex Basic Routing Example
1083 Let's say you have an AS named AS_Big that contains two other AS, AS_1
1084 and AS_2. If you want to make a host (h1) from AS_1 with another one
1085 (h2) from AS_2 then you'll have to proceed as follows:
1086 \li First, you have to ensure that a route is defined from h1 to the
1087 AS_1's exit gateway and from h2 to AS_2's exit gateway.
1088 \li Then, you'll have to define a route between AS_1 to AS_2. As those
1089 AS are both resources belonging to AS_Big, then it has to be done
1090 at AS_big level. To define such a route, you have to give the
1091 source AS (AS_1), the destination AS (AS_2), and their respective
1092 gateway (as the route is effectively defined between those two
1093 entry/exit points). Elements of this route can only be elements
1094 belonging to AS_Big, so links and routers in this route should be
1095 defined inside AS_Big. If you choose some shortest-path model,
1096 this route will be computed automatically.
1098 As said before, there are mainly 2 tags for routing :
1099 \li <b>ASroute</b>: to define routes between two <b>AS</b>
1100 \li <b>route</b>: to define routes between two <b>host/router</b>
1102 As we are dealing with routes between AS, it means that those we'll
1103 have some definition at AS_Big level. Let consider AS_1 contains 1
1104 host, 1 link and one router and AS_2 3 hosts, 4 links and one router.
1105 There will be a central router, and a cross-like topology. At the end
1106 of the crosses arms, you'll find the 3 hosts and the router that will
1107 act as a gateway. We have to define routes inside those two AS. Let
1108 say that AS_1 contains full routes, and AS_2 contains some Floyd
1109 routing (as we don't want to bother with defining all routes). As
1110 we're using some shortest path algorithms to route into AS_2, we'll
1111 then have to define some <b>route</b> to gives some topological
1112 information to SimGrid. Here is a file doing it all :
1115 <AS id="AS_Big" routing="Dijsktra">
1116 <AS id="AS_1" routing="Full">
1117 <host id="AS_1_host1" power="1000000000"/>
1118 <link id="AS_1_link" bandwidth="1250000000" latency="5E-4"/>
1119 <router id="AS_1_gateway"/>
1120 <route src="AS_1_host1" dst="AS_1_gateway">
1121 <link_ctn id="AS_1_link"/>
1124 <AS id="AS_2" routing="Floyd">
1125 <host id="AS_2_host1" power="1000000000"/>
1126 <host id="AS_2_host2" power="1000000000"/>
1127 <host id="AS_2_host3" power="1000000000"/>
1128 <link id="AS_2_link1" bandwidth="1250000000" latency="5E-4"/>
1129 <link id="AS_2_link2" bandwidth="1250000000" latency="5E-4"/>
1130 <link id="AS_2_link3" bandwidth="1250000000" latency="5E-4"/>
1131 <link id="AS_2_link4" bandwidth="1250000000" latency="5E-4"/>
1132 <router id="central_router"/>
1133 <router id="AS_2_gateway"/>
1134 <!-- routes providing topological information -->
1135 <route src="central_router" dst="AS_2_host1"><link_ctn id="AS_2_link1"/></route>
1136 <route src="central_router" dst="AS_2_host2"><link_ctn id="AS_2_link2"/></route>
1137 <route src="central_router" dst="AS_2_host3"><link_ctn id="AS_2_link3"/></route>
1138 <route src="central_router" dst="AS_2_gateway"><link_ctn id="AS_2_link4"/></route>
1140 <link id="backbone" bandwidth="1250000000" latency="5E-4"/>
1142 <ASroute src="AS_1" dst="AS_2"
1143 gw_src="AS_1_gateway"
1144 gw_dst="AS_2_gateway">
1145 <link_ctn id="backbone"/>
1150 \section pf_other_tags Tags not (directly) describing the platform
1152 There are 3 tags, that you can use inside a \<platform\> tag that are
1153 not describing the platform:
1154 \li random: it allows you to define random generators you want to use
1155 for your simulation.
1156 \li config: it allows you to pass some configuration stuff like, for
1157 example, the network model and so on. It follows the
1158 \li include: simply allows you to include another file into the
1161 \subsection pf_conf config
1162 <b>config</b> attributes :
1163 \li <b>id (mandatory)</b>: the identifier of the config to be used
1164 when referring to it.
1167 <b>config</b> tag only purpose is to include <b>prop</b> tags. Valid
1168 id are basically the same as the list of possible parameters you can
1169 use by command line, except that "/" are used for namespace
1170 definition. See the \ref options config and options page for more
1174 <b>config example</b>
1176 <?xml version='1.0'?>
1177 <!DOCTYPE platform SYSTEM "http://simgrid.gforge.inria.fr/simgrid.dtd">
1178 <platform version="3">
1179 <config id="General">
1180 <prop id="maxmin/precision" value="0.000010"></prop>
1181 <prop id="cpu/optim" value="TI"></prop>
1182 <prop id="workstation/model" value="compound"></prop>
1183 <prop id="network/model" value="SMPI"></prop>
1184 <prop id="path" value="~/"></prop>
1185 <prop id="smpi/bw_factor" value="65472:0.940694;15424:0.697866;9376:0.58729"></prop>
1188 <AS id="AS0" routing="Full">
1193 \subsection pf_rand random
1194 Not yet in use, and possibly subject to huge modifications.
1196 \subsection pf_incl include
1197 <b>include</b> tag allows to import into a file platform parts located
1198 in another file. This is done with the intention to help people
1199 combine their different AS and provide new platforms. Those files
1200 should contains XML part that contains either
1201 <b>include,cluster,peer,AS,trace,trace_connect</b> tags.
1203 <b>include</b> attributes :
1204 \li <b>file (mandatory)</b>: filename of the file to include. Possible
1205 values: absolute or relative path, syntax similar to the one in
1208 <b>Note</b>: due to some obscure technical reasons, you have to open
1209 and close tag in order to let it work.
1210 <b>include Example</b>
1212 <?xml version='1.0'?>
1213 <!DOCTYPE platform SYSTEM "http://simgrid.gforge.inria.fr/simgrid.dtd">
1214 <platform version="3">
1215 <AS id="main" routing="Full">
1216 <include file="clusterA.xml"></include>
1217 <include file="clusterB.xml"></include>
1222 \subsection pf_tra trace and trace_connect
1223 Both tags are an alternate way to passe availability, state, and so on
1224 files to entity. Instead of referring to the file directly in the host,
1225 link, or cluster tag, you proceed by defining a trace with an id
1226 corresponding to a file, later a host/link/cluster, and finally using
1227 trace_connect you say that the file trace must be used by the entity.
1228 Get it ? Let's have a look at an example :
1231 <AS id="AS0" routing="Full">
1232 <host id="bob" power="1000000000"/>
1234 <trace id="myTrace" file="bob.trace" periodicity="1.0"/>
1235 <trace_connect trace="myTrace" element="bob" kind="POWER"/>
1238 All constraints you have is that <b>trace_connect</b> is after
1239 <b>trace</b> and <b>host</b> definitions.
1242 <b>trace</b> attributes :
1243 \li <b>id (mandatory)</b>: the identifier of the trace to be used when
1245 \li <b>file</b>: filename of the file to include. Possible values :
1246 absolute or relative path, syntax similar to the one in use on
1247 your system. If omitted, the system expects that you provide the
1248 trace values inside the trace tags (see below).
1249 \li <b>trace periodicity (mandatory)</b>: trace periodicity, same
1250 definition as in hosts (see upper for details).
1252 Here is an example of trace when no file name is provided:
1255 <trace id="myTrace" periodicity="1.0">
1262 <b>trace_connect</b> attributes :
1263 \li <b>kind</b>: the type of trace, possible values
1264 <b>HOST_AVAIL|POWER|LINK_AVAIL|BANDWIDTH|LATENCY,</b> default:
1266 \li <b>trace (mandatory)</b>: the identifier of the trace referenced.
1267 \li <b>element (mandatory)</b>: the identifier of the entity referenced.
1271 \section pf_hints Hints and tips, or how to write a platform efficiently
1273 Now you should know at least the syntax and be able to create a
1274 platform by your own. However, after having ourselves wrote some platforms, there
1275 are some best practices you should pay attention to in order to
1276 produce good platform and some choices you can make in order to have
1277 faster simulations. Here's some hints and tips, then.
1279 \subsection pf_as_h AS Hierarchy
1280 The AS design allows SimGrid to go fast, because computing route is
1281 done only for the set of resources defined in this AS. If you're using
1282 only a big AS containing all resource with no AS into it and you're
1283 using Full model, then ... you'll loose all interest into it. On the
1284 other hand, designing a binary tree of AS with, at the lower level,
1285 only one host, then you'll also loose all the good AS hierarchy can
1286 give you. Remind you should always be "reasonable" in your platform
1287 definition when choosing the hierarchy. A good choice if you try to
1288 describe a real life platform is to follow the AS described in
1289 reality, since this kind of trade-off works well for real life
1292 \subsection pf_exit_as Exit AS: why and how
1293 Users that have looked at some of our platforms may have notice a
1294 non-intuitive schema ... Something like that :
1298 <AS id="AS_4" routing="Full">
1299 <AS id="exitAS_4" routing="Full">
1300 <router id="router_4"/>
1302 <cluster id="cl_4_1" prefix="c_4_1-" suffix="" radical="1-20" power="1000000000" bw="125000000" lat="5E-5" bb_bw="2250000000" bb_lat="5E-4"/>
1303 <cluster id="cl_4_2" prefix="c_4_2-" suffix="" radical="1-20" power="1000000000" bw="125000000" lat="5E-5" bb_bw="2250000000" bb_lat="5E-4"/>
1304 <link id="4_1" bandwidth="2250000000" latency="5E-5"/>
1305 <link id="4_2" bandwidth="2250000000" latency="5E-5"/>
1306 <link id="bb_4" bandwidth="2250000000" latency="5E-4"/>
1307 <ASroute src="cl_4_1"
1309 gw_src="c_4_1-cl_4_1_router"
1310 gw_dst="c_4_2-cl_4_2_router"
1312 <link_ctn id="4_1"/>
1313 <link_ctn id="bb_4"/>
1314 <link_ctn id="4_2"/>
1316 <ASroute src="cl_4_1"
1318 gw_src="c_4_1-cl_4_1_router"
1321 <link_ctn id="4_1"/>
1322 <link_ctn id="bb_4"/>
1324 <ASroute src="cl_4_2"
1326 gw_src="c_4_2-cl_4_2_router"
1329 <link_ctn id="4_2"/>
1330 <link_ctn id="bb_4"/>
1335 In the AS_4, you have an exitAS_4 defined, containing only one router,
1336 and routes defined to that AS from all other AS (as cluster is only a
1337 shortcut for an AS, see cluster description for details). If there was
1338 an upper AS, it would define routes to and from AS_4 with the gateway
1339 router_4. It's just because, as we did not allowed (for performances
1340 issues) to have routes from an AS to a single host/router, you have to
1341 enclose your gateway, when you have AS included in your AS, within an
1342 AS to define routes to it.
1344 \subsection pf_P2P_tags P2P or how to use coordinates
1345 SimGrid allows you to use some coordinated-based system, like vivaldi,
1346 to describe a platform. The main concept is that you have some peers
1347 that are located somewhere: this is the function of the
1348 <b>coordinates</b> of the \<peer\> or \<host\> tag. There's nothing
1349 complicated in using it, here is an example of it:
1352 <?xml version='1.0'?>
1353 <!DOCTYPE platform SYSTEM "http://simgrid.gforge.inria.fr/simgrid.dtd">
1354 <platform version="3">
1356 <config id="General">
1357 <prop id="network/coordinates" value="yes"></prop>
1359 <AS id="AS0" routing="Vivaldi">
1360 <host id="100030591" coordinates="25.5 9.4 1.4" power="1500000000.0" />
1361 <host id="100036570" coordinates="-12.7 -9.9 2.1" power="730000000.0" />
1363 <host id="100429957" coordinates="17.5 6.7 18.8" power="830000000.0" />
1368 Coordinates are then used to calculate latency between two hosts by
1369 calculating the euclidean distance between the two hosts coordinates.
1370 The results express the latency in ms.
1372 Note that the previous example defines a routing directly between hosts but it could be also used to define a routing between AS.
1373 That is for example what is commonly done when using peers (see Section \ref pf_peer).
1375 <?xml version='1.0'?>
1376 <!DOCTYPE platform SYSTEM "http://simgrid.gforge.inria.fr/simgrid.dtd">
1377 <platform version="3">
1379 <config id="General">
1380 <prop id="network/coordinates" value="yes"></prop>
1382 <AS id="AS0" routing="Vivaldi">
1383 <peer id="peer-0" coordinates="173.0 96.8 0.1" power="730Mf" bw_in="13.38MBps" bw_out="1.024MBps" lat="500us"/>
1384 <peer id="peer-1" coordinates="247.0 57.3 0.6" power="730Mf" bw_in="13.38MBps" bw_out="1.024MBps" lat="500us" />
1385 <peer id="peer-2" coordinates="243.4 58.8 1.4" power="730Mf" bw_in="13.38MBps" bw_out="1.024MBps" lat="500us" />
1389 In such a case though, we connect the AS created by the <b>peer</b> tag with the Vivaldi routing mechanism.
1390 This means that to route between AS1 and AS2, it will use the coordinates of router_AS1 and router_AS2.
1391 This is currently a convention and we may offer to change this convention in the DTD later if needed.
1392 You may have noted that conveniently, a peer named FOO defines an AS named FOO and a router named router_FOO, which is why it works seamlessly with the <b>peer</b> tag.
1395 \subsection pf_wisely Choosing wisely the routing model to use
1398 Choosing wisely the routing model to use can significantly fasten your
1399 simulation/save your time when writing the platform/save tremendous
1400 disk space. Here is the list of available model and their
1401 characteristics (lookup : time to resolve a route):
1403 \li <b>Full</b>: Full routing data (fast, large memory requirements,
1405 \li <b>Floyd</b>: Floyd routing data (slow initialization, fast
1406 lookup, lesser memory requirements, shortest path routing only).
1407 Calculates all routes at once at the beginning.
1408 \li <b>Dijkstra</b>: Dijkstra routing data (fast initialization, slow
1409 lookup, small memory requirements, shortest path routing only).
1410 Calculates a route when necessary.
1411 \li <b>DijkstraCache</b>: Dijkstra routing data (fast initialization,
1412 fast lookup, small memory requirements, shortest path routing
1413 only). Same as Dijkstra, except it handles a cache for latest used
1415 \li <b>none</b>: No routing (usable with Constant network only).
1416 Defines that there is no routes, so if you try to determine a
1417 route without constant network within this AS, SimGrid will raise
1419 \li <b>Vivaldi</b>: Vivaldi routing, so when you want to use coordinates
1420 \li <b>Cluster</b>: Cluster routing, specific to cluster tag, should
1423 \subsection pf_switch Hey, I want to describe a switch but there is no switch tag !
1425 Actually we did not include switch tag, ok. But when you're trying to
1426 simulate a switch, the only major impact it has when you're using
1427 fluid model (and SimGrid uses fluid model unless you activate GTNetS,
1428 ns-3, or constant network mode) is the impact of the upper limit of
1429 the switch motherboard speed that will eventually be reached if you're
1430 using intensively your switch. So, the switch impact is similar to a
1431 link one. That's why we are used to describe a switch using a link tag
1432 (as a link is not an edge by a hyperedge, you can connect more than 2
1435 \subsection pf_platform_multipath How to express multipath routing in platform files?
1437 It is unfortunately impossible to express the fact that there is more
1438 than one routing path between two given hosts. Let's consider the
1439 following platform file:
1442 <route src="A" dst="B">
1445 <route src="B" dst="C">
1448 <route src="A" dst="C">
1453 Although it is perfectly valid, it does not mean that data traveling
1454 from A to C can either go directly (using link 3) or through B (using
1455 links 1 and 2). It simply means that the routing on the graph is not
1456 trivial, and that data do not following the shortest path in number of
1457 hops on this graph. Another way to say it is that there is no implicit
1458 in these routing descriptions. The system will only use the routes you
1459 declare (such as <route src="A" dst="C"><link_ctn
1460 id="3"/></route>), without trying to build new routes by aggregating
1463 You are also free to declare platform where the routing is not
1464 symmetric. For example, add the following to the previous file:
1467 <route src="C" dst="A">
1473 This makes sure that data from C to A go through B where data from A
1474 to C go directly. Don't worry about realism of such settings since
1475 we've seen ways more weird situation in real settings (in fact, that's
1476 the realism of very regular platforms which is questionable, but
1477 that's another story).
1479 \section pf_flexml_bypassing Bypassing the XML parser with your own C functions
1480 <b>NOTE THAT THIS DOCUMENTATION, WHILE STILL WORKING, IS STRONGLY DEPRECATED</b>
1482 So you want to bypass the XML files parser, uh? Maybe doing some parameter
1483 sweep experiments on your simulations or so? This is possible, and
1484 it's not even really difficult (well. Such a brutal idea could be
1485 harder to implement). Here is how it goes.
1487 For this, you have to first remember that the XML parsing in SimGrid is done
1488 using a tool called FleXML. Given a DTD, this gives a flex-based parser. If
1489 you want to bypass the parser, you need to provide some code mimicking what
1490 it does and replacing it in its interactions with the SURF code. So, let's
1491 have a look at these interactions.
1493 FleXML parser are close to classical SAX parsers. It means that a
1494 well-formed SimGrid platform XML file might result in the following
1497 - start "platform_description" with attribute version="2"
1498 - start "host" with attributes id="host1" power="1.0"
1500 - start "host" with attributes id="host2" power="2.0"
1502 - start "link" with ...
1504 - start "route" with ...
1505 - start "link_ctn" with ...
1508 - end "platform_description"
1510 The communication from the parser to the SURF code uses two means:
1511 Attributes get copied into some global variables, and a surf-provided
1512 function gets called by the parser for each event. For example, the event
1513 - start "host" with attributes id="host1" power="1.0"
1515 let the parser do something roughly equivalent to:
1517 strcpy(A_host_id,"host1");
1522 In SURF, we attach callbacks to the different events by initializing the
1523 pointer functions to some the right surf functions. Since there can be
1524 more than one callback attached to the same event (if more than one
1525 model is in use, for example), they are stored in a dynar. Example in
1526 workstation_ptask_L07.c:
1528 /* Adding callback functions */
1529 surf_parse_reset_parser();
1530 surfxml_add_callback(STag_surfxml_host_cb_list, &parse_cpu_init);
1531 surfxml_add_callback(STag_surfxml_prop_cb_list, &parse_properties);
1532 surfxml_add_callback(STag_surfxml_link_cb_list, &parse_link_init);
1533 surfxml_add_callback(STag_surfxml_route_cb_list, &parse_route_set_endpoints);
1534 surfxml_add_callback(ETag_surfxml_link_c_ctn_cb_list, &parse_route_elem);
1535 surfxml_add_callback(ETag_surfxml_route_cb_list, &parse_route_set_route);
1537 /* Parse the file */
1538 surf_parse_open(file);
1539 xbt_assert(!surf_parse(), "Parse error in %s", file);
1543 So, to bypass the FleXML parser, you need to write your own version of the
1544 surf_parse function, which should do the following:
1545 - Fill the A_<tag>_<attribute> variables with the wanted values
1546 - Call the corresponding STag_<tag>_fun function to simulate tag start
1547 - Call the corresponding ETag_<tag>_fun function to simulate tag end
1548 - (do the same for the next set of values, and loop)
1550 Then, tell SimGrid that you want to use your own "parser" instead of the stock one:
1552 surf_parse = surf_parse_bypass_environment;
1553 MSG_create_environment(NULL);
1554 surf_parse = surf_parse_bypass_application;
1555 MSG_launch_application(NULL);
1558 A set of macros are provided at the end of
1559 include/surf/surfxml_parse.h to ease the writing of the bypass
1560 functions. An example of this trick is distributed in the file
1561 examples/msg/masterslave/masterslave_bypass.c