1 /*! @page platform Describing the virtual platform
5 As usual, SimGrid is a versatile framework, and you should find the
6 way of describing your platform that best fits your experimental
9 @section pf_first_example First Platform Example
11 Here is a very simple platform file, containing 3 resources (two hosts
12 and one link), and explicitly giving the route between the hosts.
16 <!DOCTYPE platform SYSTEM "https://simgrid.org/simgrid.dtd">
17 <platform version="4.1">
18 <zone id="first zone" routing="Full">
19 <!-- the resources -->
20 <host id="host1" speed="1Mf"/>
21 <host id="host2" speed="2Mf"/>
22 <link id="link1" bandwidth="125MBps" latency="100us"/>
23 <!-- the routing: specify how the hosts are interconnected -->
24 <route src="host1" dst="host2">
25 <link_ctn id="link1"/>
32 Then, every resource (specified with @ref pf_tag_host, @ref
33 pf_tag_link or others) must be located within a given **networking
34 zone**. Each zone is in charge of the routing between its
35 resources. It means that when an host wants to communicate with
36 another host of the same zone, it is the zone's duty to find the list
37 of links that are involved in the communication. Here, since the @ref
38 pf_tag_zone tag has **Full** as a **routing attribute**, all routes
39 must be explicitely given using the @ref pf_tag_route and @ref
40 pf_tag_linkctn tags (this @ref pf_rm "routing model" is both simple
41 and inefficient :) It is OK to not specify the route between two
42 hosts, as long as the processes located on them never try to
45 A zone can contain several zones itself, leading to a hierarchical
46 decomposition of the platform. This can be more efficient (as the
47 inter-zone routing gets factorized with @ref pf_tag_zoneroute), and
48 allows to have more than one routing model in your platform. For
49 example, you could have a coordinate-based routing for the WAN parts
50 of your platforms, a full routing within each datacenter, and a highly
51 optimized routing within each cluster of the datacenter. In this
52 case, determining the route between two given hosts gets @ref
53 routing_basics "somewhat more complex" but SimGrid still computes
54 these routes for you in a time- and space-efficient manner.
55 Here is an illustration of these concepts:
57 
59 Circles represent processing units and squares represent network
60 routers. Bold lines represent communication links. The zone "AS2"
61 models the core of a national network interconnecting a small flat
62 cluster (AS4) and a larger hierarchical cluster (AS5), a subset of a
63 LAN (AS6), and a set of peers scattered around the world (AS7).
65 @section pf_res Resource description
67 @subsection pf_res_computing Computing Resources
70 @subsubsection pf_tag_cluster <cluster>
72 ``<cluster />`` represents a machine-cluster. It is most commonly used
73 when one wants to define many hosts and a network quickly. Technically,
74 ``cluster`` is a meta-tag: <b>from the inner SimGrid point of
75 view, a cluster is a network zone where some optimized routing is defined</b>.
76 The default inner organization of the cluster is as follow:
82 ____________|__________|_____________ backbone
84 l0| l1| l2| l97| l96 | | l99
90 Here, a set of <b>host</b>s is defined. Each of them has a <b>link</b>
91 to a central backbone (backbone is a link itself, as a link can
92 be used to represent a switch, see the switch / link section
93 below for more details about it). A <b>router</b> allows to connect a
94 <b>cluster</b> to the outside world. Internally,
95 SimGrid treats a cluster as a network zone containing all hosts: the router is the default
96 gateway for the cluster.
98 There is an alternative organization, which is as follows:
112 The principle is the same, except that there is no backbone. This representation
113 can be obtained easily: just do not set the bb_* attributes.
116 Attribute name | Mandatory | Values | Description
117 --------------- | --------- | ------ | -----------
118 id | yes | string | The identifier of the cluster. Facilitates referring to this cluster.
119 prefix | yes | string | Each node of the cluster has to have a name. This name will be prefixed with this prefix.
120 suffix | yes | string | Each node of the cluster will be suffixed with this suffix
121 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.
122 speed | yes | int | Same as the ``speed`` attribute of the ``@<host@>`` tag.
123 core | no | int (default: 1) | Same as the ``core`` attribute of the ``@<host@>`` tag.
124 bw | yes | int | Bandwidth for the links between nodes and backbone (if any). See the @ref pf_tag_link "link section" for syntax/details.
125 lat | yes | int | Latency for the links between nodes and backbone (if any). See <b>link</b> section for syntax/details.
126 sharing_policy | no | string | Sharing policy for the links between nodes and backbone (if any). See <b>link</b> section for syntax/details.
127 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>).
128 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>).
129 bb_sharing_policy | no | string | Sharing policy for the backbone (if any). See <b>link</b> section for syntax/details.
130 limiter_link | no | int | Bandwidth for limiter link (if any). This adds a specific link for each node, to set the maximum bandwidth reached when communicating in both directions at the same time. In theory this value should be 2*bw for splitduplex links, but in reality this might be less. This value will depend heavily on the communication model, and on the cluster's hardware, so no default value can be set, this has to be measured. More details can be obtained in <a href="https://hal.inria.fr/hal-00919507/"> "Toward Better Simulation of MPI Applications on Ethernet/TCP Networks"</a>
131 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 @ref pf_sharing_policy_fatpipe "@b FATPIPE".
132 loopback_lat | no | int | Latency for loopback (if any). See <b>link</b> section for syntax/details. See loopback_bw for more info.
133 topology | no | FLAT@|TORUS@|FAT_TREE@|DRAGONFLY (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>, FAT_TREE, and DRAGONFLY attributes for this tag.
134 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. Please refer to the specific documentation for @ref simgrid::kernel::routing::FatTreeZone "FatTree NetZone", @ref simgrid::kernel::routing::DragonflyZone "Dragonfly NetZone".
137 the router name is defined as the resulting String in the following
141 router_name = prefix + clusterId + "_router" + suffix;
145 #### Cluster example ####
147 Consider the following two (and independent) uses of the ``cluster`` tag:
150 <cluster id="my_cluster_1" prefix="" suffix="" radical="0-262144"
151 speed="1e9" bw="125e6" lat="5E-5"/>
153 <cluster id="my_cluster_2" prefix="c-" suffix=".me" radical="0-99"
154 speed="1e9" bw="125e6" lat="5E-5"
155 bb_bw="2.25e9" bb_lat="5E-4"/>
158 The second example creates one router and 100 machines with the following names:
160 c-my_cluster_2_router.me
168 @subsubsection pf_cabinet <cabinet>
171 This tag is only available when the routing mode of the network zone
172 is set to ``Cluster``.
174 The ``<cabinet />`` tag is, like the @ref pf_tag_cluster "<cluster>" tag,
175 a meta-tag. This means that it is simply a shortcut for creating a set of (homogenous) hosts and links quickly;
176 unsurprisingly, this tag was introduced to setup cabinets in data centers quickly. Unlike
177 <cluster>, however, the <cabinet> assumes that you create the backbone
178 and routers yourself; see our examples below.
182 Attribute name | Mandatory | Values | Description
183 --------------- | --------- | ------ | -----------
184 id | yes | string | The identifier of the cabinet. Facilitates referring to this cluster.
185 prefix | yes | string | Each node of the cabinet has to have a name. This name will be prefixed with this prefix.
186 suffix | yes | string | Each node of the cabinet will be suffixed with this suffix
187 radical | yes | string | Regexp used to generate cabinet 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.
188 speed | yes | int | Same as the ``speed`` attribute of the @ref pf_tag_host "<host>" tag.
189 bw | yes | int | Bandwidth for the links between nodes and backbone (if any). See the @ref pf_tag_link "link section" for syntax/details.
190 lat | yes | int | Latency for the links between nodes and backbone (if any). See the @ref pf_tag_link "link section" for syntax/details.
193 Please note that as of now, it is impossible to change attributes such as,
194 amount of cores (always set to 1), the sharing policy of the links (always set to @ref pf_sharing_policy_splitduplex "SPLITDUPLEX").
198 The following example was taken from ``examples/platforms/meta_cluster.xml`` and
199 shows how to use the cabinet tag.
202 <zone id="my_cluster1" routing="Cluster">
203 <cabinet id="cabinet1" prefix="host-" suffix=".cluster1"
204 speed="1Gf" bw="125MBps" lat="100us" radical="1-10"/>
205 <cabinet id="cabinet2" prefix="host-" suffix=".cluster1"
206 speed="1Gf" bw="125MBps" lat="100us" radical="11-20"/>
207 <cabinet id="cabinet3" prefix="host-" suffix=".cluster1"
208 speed="1Gf" bw="125MBps" lat="100us" radical="21-30"/>
210 <backbone id="backbone1" bandwidth="2.25GBps" latency="500us"/>
215 Please note that you must specify the @ref pf_backbone "<backbone>"
216 tag by yourself; this is not done automatically and there are no checks
217 that ensure this backbone was defined.
219 The hosts generated in the above example are named host-1.cluster, host-2.cluster1
223 @subsection pf_ne Network equipments
225 There are two tags at all times available to represent network entities and
226 several other tags that are available only in certain contexts.
227 1. ``<link>``: Represents a entity that has a limited bandwidth, a
228 latency, and that can be shared according to TCP way to share this
231 The concept of links in SimGrid may not be intuitive, as links are not
232 limited to connecting (exactly) two entities; in fact, you can have more than
233 two equipments connected to it. (In graph theoretical terms: A link in
234 SimGrid is not an edge, but a hyperedge)
236 2. ``<router/>``: Represents an entity that a message can be routed
237 to, but that is unable to execute any code. In SimGrid, routers have also
238 no impact on the performance: Routers do not limit any bandwidth nor
239 do they increase latency. As a matter of fact, routers are (almost) ignored
240 by the simulator when the simulation has begun.
242 3. ``<backbone/>``: This tag is only available when the containing network zone is
243 used as a cluster (i.e., mode="Cluster")
246 If you want to represent an entity like a switch, you must use ``<link>`` (see section). Routers are used
247 to run some routing algorithm and determine routes (see Section @ref pf_routing for details).
249 @subsubsection pf_tag_link <link>
251 Network links can represent one-hop network connections. They are
252 characterized by their id and their bandwidth; links can (but may not) be subject
257 Attribute name | Mandatory | Values | Description
258 --------------- | --------- | ------ | -----------
259 id | yes | string | The identifier of the link to be used when referring to it.
260 bandwidth | yes | string | Maximum bandwidth for this link, along with its unit.
261 latency | no | double (default: 0.0) | Latency for this link.
262 sharing_policy | no | @ref sharing_policy_shared "SHARED"@|@ref pf_sharing_policy_fatpipe "FATPIPE"@|@ref pf_sharing_policy_splitduplex "SPLITDUPLEX" (default: SHARED) | Sharing policy for the link.
263 bandwidth_file | no | string | Allows you to use a file as input for bandwidth.
264 latency_file | no | string | Allows you to use a file as input for latency.
265 state_file | no | string | Allows you to use a file as input for states.
268 #### Possible shortcuts for ``latency`` ####
270 When using the latency attribute, you can specify the latency by using the scientific
271 notation or by using common abbreviations. For instance, the following three tags
275 <link id="LINK1" bandwidth="125000000" latency="5E-6"/>
276 <link id="LINK1" bandwidth="125000000" latency="5us"/>
277 <link id="LINK1" bandwidth="125000000" latency="0.000005"/>
280 Here, the second tag uses "us", meaning "microseconds". Other shortcuts are:
282 Name | Abbreviation | Time (in seconds)
283 ---- | ------------ | -----------------
284 Week | w | 7 * 24 * 60 * 60
285 Day | d | 24 * 60 * 60
289 Millisecond | ms | 0.001 = 10^(-3)
290 Microsecond | us | 0.000001 = 10^(-6)
291 Nanosecond | ns | 0.000000001 = 10^(-9)
292 Picosecond | ps | 0.000000000001 = 10^(-12)
294 #### Sharing policy ####
296 @anchor sharing_policy_shared
297 By default a network link is @b SHARED, i.e., if two or more data flows go
298 through a link, the bandwidth is shared fairly among all data flows. This
299 is similar to the sharing policy TCP uses.
301 @anchor pf_sharing_policy_fatpipe
302 On the other hand, if a link is defined as a @b FATPIPE,
303 each flow going through this link will be provided with the complete bandwidth,
304 i.e., no sharing occurs and the bandwidth is only limiting each flow individually.
305 Please note that this is really on a per-flow basis, not only on a per-host basis!
306 The complete bandwidth provided by this link in this mode
307 is ``number_of_flows*bandwidth``, with at most ``bandwidth`` being available per flow.
309 Using the FATPIPE mode allows to model backbones that won't affect performance
312 @anchor pf_sharing_policy_splitduplex
313 The last mode available is @b SPLITDUPLEX. This means that SimGrid will
314 automatically generate two links (one carrying the suffix _UP and the other the
315 suffix _DOWN) for each ``<link>`` tag. This models situations when the direction
316 of traffic is important.
319 Transfers from one side to the other will interact similarly as
320 TCP when ACK returning packets circulate on the other direction. More
321 discussion about it is available in the description of link_ctn description.
323 In other words: The SHARED policy defines a physical limit for the bandwidth.
324 The FATPIPE mode defines a limit for each application,
325 with no upper total limit.
328 Tip: By using the FATPIPE mode, you can model big backbones that
329 won't affect performance (except latency).
334 <link id="SWITCH" bandwidth="125000000" latency="5E-5" sharing_policy="FATPIPE" />
337 #### Expressing dynamism and failures ####
339 Similar to hosts, it is possible to declare links whose state, bandwidth
340 or latency changes over time (see Section @ref pf_host_dynamism for details).
342 In the case of network links, the ``bandwidth`` and ``latency`` attributes are
343 replaced by the ``bandwidth_file`` and ``latency_file`` attributes.
344 The following XML snippet demonstrates how to use this feature in the platform
345 file. The structure of the files "link1.bw" and "link1.lat" is shown below.
348 <link id="LINK1" state_file="link1.fail" bandwidth="80000000" latency=".0001" bandwidth_file="link1.bw" latency_file="link1.lat" />
352 Even if the syntax is the same, the semantic of bandwidth and latency
353 trace files differs from that of host availability files. For bandwidth and
354 latency, the corresponding files do not
355 express availability as a fraction of the available capacity but directly in
356 bytes per seconds for the bandwidth and in seconds for the latency. This is
357 because most tools allowing to capture traces on real platforms (such as NWS)
358 express their results this way.
360 ##### Example of "link1.bw" file #####
368 In this example, the bandwidth changes repeatedly, with all changes
369 being repeated every 12 seconds.
371 At the beginning of the the simulation, the link's bandwidth is 80,000,000
372 B/s (i.e., 80 Mb/s); this value was defined in the XML snippet above.
373 After four seconds, it drops to 40 Mb/s (line 2), and climbs
374 back to 60 Mb/s after another 4 seconds (line 3). The value does not change any
375 more until the end of the period, that is, after 12 seconds have been simulated).
376 At this point, periodicity kicks in and this behavior is repeated: Seconds
377 12-16 will experience 80 Mb/s, 16-20 40 Mb/s etc.).
379 ##### Example of "link1.lat" file #####
388 In this example, the latency varies with a period of 5 seconds.
389 In the xml snippet above, the latency is initialized to be 0.0001s (100µs). This
390 value will be kept during the first second, since the latency_file contains
391 changes to this value at second one, two and three.
392 At second one, the value will be 0.001, i.e., 1ms. One second later it will
393 be adjusted to 0.01 (or 10ms) and one second later it will be set again to 1ms. The
394 value will not change until second 5, when the periodicity defined in line 1
395 kicks in. It then loops back, starting at 100µs (the initial value) for one second.
397 #### The ``<prop/>`` tag ####
399 Similar to the ``<host>`` tag, a link may also contain the ``<prop/>`` tag; see the host
400 documentation (Section @ref pf_tag_host) for an example.
403 @subsubsection pf_backbone <backbone/>
406 This tag is <b>only available</b> when the containing network zone uses the "Cluster" routing mode!
408 Using this tag, you can designate an already existing link to be a backbone.
410 Attribute name | Mandatory | Values | Description
411 --------------- | --------- | ------ | -----------
412 id | yes | string | Name of the link that is supposed to act as a backbone.
414 @subsection pf_storage Storage
417 This is a prototype version that should evolve quickly, hence this
418 is just some doc valuable only at the time of writing.
419 This section describes the storage management under SimGrid ; nowadays
420 it's only usable with MSG. It relies basically on linux-like concepts.
421 You also may want to have a look to its corresponding section in
422 @ref msg_file ; access functions are organized as a POSIX-like
425 @subsubsection pf_sto_conc Storage - Main Concepts
427 The storage facilities implemented in SimGrid help to model (and account for)
428 storage devices, such as tapes, hard-drives, CD or DVD devices etc.
429 A typical situation is depicted in the figure below:
431 @image html ./webcruft/storage_sample_scenario.png
432 @image latex ./webcruft/storage_sample_scenario.png "storage_sample_scenario" width=@textwidth
434 In this figure, two hosts called Bob and Alice are interconnected via a network
435 and each host is physically attached to a disk; it is not only possible for each host to
436 mount the disk they are attached to directly, but they can also mount disks
437 that are in a remote location. In this example, Bob mounts Alice's disk remotely
438 and accesses the storage via the network.
440 SimGrid provides 3 different entities that can be used to model setups
441 that include storage facilities:
443 Entity name | Description
444 --------------- | -----------
445 @ref pf_storage_entity_storage_type "storage_type" | Defines a template for a particular kind of storage (such as a hard-drive) and specifies important features of the storage, such as capacity, performance (read/write), contents, ... Different models of hard-drives use different storage_types (because the difference between an SSD and an HDD does matter), as they differ in some specifications (e.g., different sizes or read/write performance).
446 @ref pf_tag_storage "storage" | Defines an actual instance of a storage type (disk, RAM, ...); uses a ``storage_type`` template (see line above) so that you don't need to re-specify the same details over and over again.
447 @ref pf_tag_mount "mount" | Must be wrapped by a @ref pf_tag_host tag; declares which storage(s) this host has mounted and where (i.e., the mountpoint).
450 @anchor pf_storage_content_file
451 ### %Storage Content File ###
453 In order to assess exactly how much time is spent reading from the storage,
454 SimGrid needs to know what is stored on the storage device (identified by distinct (file-)name, like in a file system)
455 and what size this content has.
458 The content file is never changed by the simulation; it is parsed once
459 per simulation and kept in memory afterwards. When the content of the
460 storage changes, only the internal SimGrid data structures change.
462 @anchor pf_storage_content_file_structure
463 #### Structure of a %Storage Content File ####
465 Here is an excerpt from two storage content file; if you want to see the whole file, check
466 the file ``examples/platforms/content/storage_content.txt`` that comes with the
469 SimGrid essentially supports two different formats: UNIX-style filepaths should
470 follow the well known format:
473 /lib/libsimgrid.so.3.6.2 12710497
477 /bin/simgrid_update_xml 5018
478 /bin/graphicator 66986
479 /bin/simgrid-colorizer 2993
484 Windows filepaths, unsurprisingly, use the windows style:
487 @Windows@avastSS.scr 41664
488 @Windows@bfsvc.exe 75264
489 @Windows@bootstat.dat 67584
490 @Windows@CoreSingleLanguage.xml 31497
492 @Windows@dchcfg64.exe 335464
493 @Windows@dcmdev64.exe 93288
497 The different file formats come at a cost; in version 3.12 (and most likely
498 in later versions, too), copying files from windows-style storages to unix-style
499 storages (and vice versa) is not supported.
501 @anchor pf_storage_content_file_create
502 #### Generate a %Storage Content File ####
504 If you want to generate a storage content file based on your own filesystem (or at least a filesystem you have access to),
505 try running this command (works only on unix systems):
508 find . -type f -exec ls -1s --block=1 {} @; 2>/dev/null | awk '{ print $2 " " $1}' > ./content.txt
511 @subsubsection pf_storage_entities The Storage Entities
513 These are the entities that you can use in your platform files to include
514 storage in your model. See also the list of our @ref pf_storage_example_files "example files";
515 these might also help you to get started.
517 @anchor pf_storage_entity_storage_type
518 #### @<storage_type@> ####
520 Attribute name | Mandatory | Values | Description
521 --------------- | --------- | ------ | -----------
522 id | yes | string | Identifier of this storage_type; used when referring to it
523 model | no | string | In the future, this will allow to change the performance model to use
524 size | yes | string | Specifies the amount of available storage space; you can specify storage like "500GiB" or "500GB" if you want. (TODO add a link to all the available abbreviations)
525 content | yes | string | Path to a @ref pf_storage_content_file "Storage Content File" on your system. This file must exist.
527 This tag must contain some predefined model properties, specified via the <model_prop> tag. Here is a list,
528 see below for an example:
530 Property id | Mandatory | Values | Description
531 --------------- | --------- | ------ | -----------
532 Bwrite | yes | string | Bandwidth for write access; in B/s (but you can also specify e.g. "30MBps")
533 Bread | yes | string | Bandwidth for read access; in B/s (but you can also specify e.g. "30MBps")
536 A storage_type can also contain the <b><prop></b> tag. The <prop> tag allows you
537 to associate additional information to this <storage_type> and follows the
538 attribute/value schema; see the example below. You may want to use it to give information to
539 the tool you use for rendering your simulation, for example.
541 Here is a complete example for the ``storage_type`` tag:
543 <storage_type id="single_HDD" size="4000">
544 <model_prop id="Bwrite" value="30MBps" />
545 <model_prop id="Bread" value="100MBps" />
546 <prop id="Brand" value="Western Digital" />
550 @subsubsection pf_tag_storage <storage>
552 Attributes | Mandatory | Values | Description
553 -------------- | --------- | ------ | -----------
554 id | yes | string | Identifier of this ``storage``; used when referring to it
555 typeId | yes | string | Here you need to refer to an already existing @ref pf_storage_entity_storage_type "@<storage_type@>"; the storage entity defined by this tag will then inherit the properties defined there.
556 attach | yes | string | Name of a host (see Section @ref pf_tag_host) to which this storage is <i>physically</i> attached to (e.g., a hard drive in a computer)
557 content | no | string | When specified, overwrites the content attribute of @ref pf_storage_entity_storage_type "@<storage_type@>"
559 Here are two examples:
562 <storage id="Disk1" typeId="single_HDD" attach="bob" />
564 <storage id="Disk2" typeId="single_SSD"
565 content="content/win_storage_content.txt" />
568 The first example is straightforward: A disk is defined and called "Disk1"; it is
569 of type "single_HDD" (shown as an example of @ref pf_storage_entity_storage_type "@<storage_type@>" above) and attached
570 to a host called "bob" (the definition of this host is omitted here).
572 The second storage is called "Disk2", is still of the same type as Disk1 but
573 now specifies a new content file (so the contents will be different from Disk1)
574 and the filesystem uses the windows style; finally, it is attached to a second host,
575 called alice (which is again not defined here).
577 @subsubsection pf_tag_mount <mount>
579 | Attribute | Mandatory | Values | Description |
580 | ----------- | ----------- | -------- | ------------- |
581 | id | yes | string | Refers to a @ref pf_tag_storage "<storage>" entity that will be mounted on that computer |
582 | name | yes | string | Path/location to/of the logical reference (mount point) of this disk
584 This tag must be enclosed by a @ref pf_tag_host tag. It then specifies where the mountpoint of a given storage device (defined by the ``id`` attribute)
585 is; this location is specified by the ``name`` attribute.
587 Here is a simple example, taken from the file ``examples/platform/storage.xml``:
590 <storage_type id="single_SSD" size="500GiB">
591 <model_prop id="Bwrite" value="60MBps" />
592 <model_prop id="Bread" value="200MBps" />
595 <storage id="Disk2" typeId="single_SSD"
596 content="content/win_storage_content.txt"
598 <storage id="Disk4" typeId="single_SSD"
599 content="content/small_content.txt"
602 <host id="alice" speed="1Gf">
603 <mount storageId="Disk2" name="c:"/>
606 <host id="denise" speed="1Gf">
607 <mount storageId="Disk2" name="c:"/>
608 <mount storageId="Disk4" name="/home"/>
612 This example is quite interesting, as the same device, called "Disk2", is mounted by
613 two hosts at the same time! Note, however, that the host called ``alice`` is actually
614 attached to this storage, as can be seen in the @ref pf_tag_storage "<storage>"
615 tag. This means that ``denise`` must access this storage through the network, but SimGrid automatically takes
616 care of that for you.
618 Furthermore, this example shows that ``denise`` has mounted two storages with different
619 filesystem types (unix and windows). In general, a host can mount as many storage devices as
623 Again, the difference between ``attach`` and ``mount`` is simply that
624 an attached storage is always physically inside (or connected to) that machine;
625 for instance, a USB stick is attached to one and only one machine (where it's plugged-in)
626 but it can only be mounted on others, as mounted storage can also be a remote location.
628 ###### Example files #####
630 @verbinclude example_filelist_xmltag_mount
632 @subsubsection pf_storage_example_files Example files
634 Several examples were already discussed above; if you're interested in full examples,
635 check the the following platforms:
637 1. ``examples/platforms/storage.xml``
638 2. ``examples/platforms/remote_io.xml``
640 If you're looking for some examplary C code, you may find the source code
641 available in the directory ``examples/msg/io/`` useful.
643 @subsubsection pf_storage_examples_modelling Modelling different situations
645 The storage functionality of SimGrid is type-agnostic, that is, the implementation
646 does not presume any type of storage, such as HDDs/SSDs, RAM,
647 CD/DVD devices, USB sticks etc.
649 This allows the user to apply the simulator for a wide variety of scenarios; one
650 common scenario would be the access of remote RAM.
652 #### Modelling the access of remote RAM ####
654 How can this be achieved in SimGrid? Let's assume we have a setup where three hosts
655 (HostA, HostB, HostC) need to access remote RAM:
665 An easy way to model this scenario is to setup and define the RAM via the
666 @ref pf_tag_storage "storage" and @ref pf_storage_entity_storage_type "storage type"
667 entities and attach it to a remote dummy host; then, every host can have their own links
668 to this host (modelling for instance certain scenarios, such as PCIe ...)
673 RAM - Dummy -- Host B
678 Now, if read from this storage, the host that mounts this storage
679 communicates to the dummy host which reads from RAM and
680 sends the information back.
683 @section pf_routing Routing
685 To achieve high performance, the routing tables used within SimGrid are
686 static. This means that routing between two nodes is calculated once
687 and will not change during execution. The SimGrid team chose to use this
688 approach as it is rare to have a real deficiency of a resource;
689 most of the time, a communication fails because the links experience too much
690 congestion and hence, your connection stops before the timeout or
691 because the computer designated to be the destination of that message
694 We also chose to use shortest paths algorithms in order to emulate
695 routing. Doing so is consistent with the reality: [RIP](https://en.wikipedia.org/wiki/Routing_Information_Protocol),
696 [OSPF](https://en.wikipedia.org/wiki/Open_Shortest_Path_First), [BGP](https://en.wikipedia.org/wiki/Border_Gateway_Protocol)
697 are all calculating shortest paths. They do require some time to converge, but
698 eventually, when the routing tables have stabilized, your packets will follow
701 @subsection pf_tag_zone <zone>
703 Before SimGrid v3.16, networking zones used to be called Autonomous
704 Systems, but this was misleading as zones may include other zones in a
705 hierarchical manner. If you find any remaining reference to network
706 zones, please report this as a bug.
708 Attribute | Value | Description
709 ----------- | ------------------------------------------------- | ----------------------------------------------
710 id | String (mandatory) | The identifier of this zone (must be unique)
711 routing | One of the existing routing algorithm (mandatory) | See Section @ref pf_rm for details.
715 <zone id="zone0" routing="Full">
716 <host id="host1" speed="1000000000"/>
717 <host id="host2" speed="1000000000"/>
718 <link id="link1" bandwidth="125000000" latency="0.000100"/>
719 <route src="host1" dst="host2"><link_ctn id="link1"/></route>
723 In this example, zone0 contains two hosts (host1 and host2). The route
724 between the hosts goes through link1.
726 @subsection pf_rm Routing models
728 For each network zone, you must define explicitly which routing model will
729 be used. There are 3 different categories for routing models:
731 1. @ref pf_routing_model_shortest_path "Shortest-path" based models: SimGrid calculates shortest
732 paths and manages them. Behaves more or less like most real life
734 2. @ref pf_routing_model_manual "Manually-entered" route models: you have to define all routes
735 manually in the platform description file; this can become
736 tedious very quickly, as it is very verbose.
737 Consistent with some manually managed real life routing.
738 3. @ref pf_routing_model_simple "Simple/fast models": those models offer fast, low memory routing
739 algorithms. You should consider to use this type of model if
740 you can make some assumptions about your network zone.
741 Routing in this case is more or less ignored.
743 @subsubsection pf_raf The router affair
745 Using routers becomes mandatory when using shortest-path based
746 models or when using the bindings to the ns-3 packet-level
747 simulator instead of the native analytical network model implemented
750 For graph-based shortest path algorithms, routers are mandatory, because these
751 algorithms require a graph as input and so we need to have source and
752 destination for each edge.
754 Routers are naturally an important concept ns-3 since the
755 way routers run the packet routing algorithms is actually simulated.
756 SimGrid's analytical models however simply aggregate the routing time
757 with the transfer time.
759 So why did we incorporate routers in SimGrid? Rebuilding a graph representation
760 only from the route information turns out to be a very difficult task, because
761 of the missing information about how routes intersect. That is why we
762 introduced routers, which are simply used to express these intersection points.
763 It is important to understand that routers are only used to provide topological
766 To express this topological information, a <b>route</b> has to be
767 defined in order to declare which link is connected to a router.
770 @subsubsection pf_routing_model_shortest_path Shortest-path based models
772 The following table shows all the models that compute routes using
773 shortest-paths algorithms are currently available in SimGrid. More detail on how
774 to choose the best routing model is given in the Section called @"@ref pf_routing_howto_choose_wisely@".
776 | Name | Description |
777 | --------------------------------------------------- | -------------------------------------------------------------------------- |
778 | @ref pf_routing_model_floyd "Floyd" | Floyd routing data. Pre-calculates all routes once |
779 | @ref pf_routing_model_dijkstra "Dijkstra" | Dijkstra routing data. Calculates routes only when needed |
780 | @ref pf_routing_model_dijkstracache "DijkstraCache" | Dijkstra routing data. Handles some cache for already calculated routes. |
782 All those shortest-path models are instanciated in the same way and are
783 completely interchangeable. Here are some examples:
785 @anchor pf_routing_model_floyd
790 <zone id="zone0" routing="Floyd">
792 <cluster id="my_cluster_1" prefix="c-" suffix=""
793 radical="0-1" speed="1000000000" bw="125000000" lat="5E-5"
794 router_id="router1"/>
796 <zone id="zone1" routing="None">
797 <host id="host1" speed="1000000000"/>
800 <link id="link1" bandwidth="100000" latency="0.01"/>
802 <zoneroute src="my_cluster_1" dst="zone1"
805 <link_ctn id="link1"/>
811 zoneroute given at the end gives a topological information: link1 is
812 between router1 and host1.
814 #### Example platform files ####
816 This is an automatically generated list of example files that use the Floyd
817 routing model (the path is given relative to SimGrid's source directory)
819 @verbinclude example_filelist_routing_floyd
821 @anchor pf_routing_model_dijkstra
824 #### Example platform files ####
826 This is an automatically generated list of example files that use the Dijkstra
827 routing model (the path is given relative to SimGrid's source directory)
829 @verbinclude example_filelist_routing_dijkstra
833 <zone id="zone_2" routing="Dijkstra">
834 <host id="zone_2_host1" speed="1000000000"/>
835 <host id="zone_2_host2" speed="1000000000"/>
836 <host id="zone_2_host3" speed="1000000000"/>
837 <link id="zone_2_link1" bandwidth="1250000000" latency="5E-4"/>
838 <link id="zone_2_link2" bandwidth="1250000000" latency="5E-4"/>
839 <link id="zone_2_link3" bandwidth="1250000000" latency="5E-4"/>
840 <link id="zone_2_link4" bandwidth="1250000000" latency="5E-4"/>
841 <router id="central_router"/>
842 <router id="zone_2_gateway"/>
843 <!-- routes providing topological information -->
844 <route src="central_router" dst="zone_2_host1"><link_ctn id="zone_2_link1"/></route>
845 <route src="central_router" dst="zone_2_host2"><link_ctn id="zone_2_link2"/></route>
846 <route src="central_router" dst="zone_2_host3"><link_ctn id="zone_2_link3"/></route>
847 <route src="central_router" dst="zone_2_gateway"><link_ctn id="zone_2_link4"/></route>
851 @anchor pf_routing_model_dijkstracache
852 ### DijkstraCache ###
854 DijkstraCache example:
856 <zone id="zone_2" routing="DijkstraCache">
857 <host id="zone_2_host1" speed="1000000000"/>
859 (platform unchanged compared to upper example)
862 #### Example platform files ####
864 This is an automatically generated list of example files that use the DijkstraCache
865 routing model (the path is given relative to SimGrid's source directory):
867 Editor's note: At the time of writing, no platform file used this routing model - so
868 if there are no example files listed here, this is likely to be correct.
870 @verbinclude example_filelist_routing_dijkstra_cache
872 @subsubsection pf_routing_model_manual Manually-entered route models
874 | Name | Description |
875 | ---------------------------------- | ------------------------------------------------------------------------------ |
876 | @ref pf_routing_model_full "Full" | You have to enter all necessary routers manually; that is, every single route. This may consume a lot of memory when the XML is parsed and might be tedious to write; i.e., this is only recommended (if at all) for small platforms. |
878 @anchor pf_routing_model_full
883 <zone id="zone0" routing="Full">
884 <host id="host1" speed="1000000000"/>
885 <host id="host2" speed="1000000000"/>
886 <link id="link1" bandwidth="125000000" latency="0.000100"/>
887 <route src="host1" dst="host2"><link_ctn id="link1"/></route>
891 #### Example platform files ####
893 This is an automatically generated list of example files that use the Full
894 routing model (the path is given relative to SimGrid's source directory):
896 @verbinclude example_filelist_routing_full
898 @subsubsection pf_routing_model_simple Simple/fast models
900 | Name | Description |
901 | ---------------------------------------- | ------------------------------------------------------------------------------ |
902 | @ref pf_routing_model_cluster "Cluster" | This is specific to the @ref pf_tag_cluster "<cluster/>" tag and should not be used by the user, as several assumptions are made. |
903 | @ref pf_routing_model_none "None" | No routing at all. Unless you know what you're doing, avoid using this mode in combination with a non-constant network model. |
904 | @ref pf_routing_model_vivaldi "Vivaldi" | Perfect when you want to use coordinates. Also see the corresponding @ref pf_P2P_tags "P2P section" below. |
906 @anchor pf_routing_model_cluster
910 In this mode, the @ref pf_cabinet "<cabinet/>" tag is available.
912 #### Example platform files ####
914 This is an automatically generated list of example files that use the Cluster
915 routing model (the path is given relative to SimGrid's source directory):
917 @verbinclude example_filelist_routing_cluster
919 @anchor pf_routing_model_none
923 This model does exactly what it's name advertises: Nothing. There is no routing
924 available within this model and if you try to communicate within the zone that
925 uses this model, SimGrid will fail unless you have explicitly activated the
926 @ref options_model_select_network_constant "Constant Network Model" (this model charges
927 the same for every single communication). It should
928 be noted, however, that you can still attach an @ref pf_tag_zoneroute "ZoneRoute",
929 as is demonstrated in the example below:
931 @verbinclude platforms/cluster_and_one_host.xml
933 #### Example platform files ####
935 This is an automatically generated list of example files that use the None
936 routing model (the path is given relative to SimGrid's source directory):
938 @verbinclude example_filelist_routing_none
941 @anchor pf_routing_model_vivaldi
944 For more information on how to use the [Vivaldi Coordinates](https://en.wikipedia.org/wiki/Vivaldi_coordinates),
945 see also Section @ref pf_P2P_tags "P2P tags".
947 Note that it is possible to combine the Vivaldi routing model with other routing models;
948 an example can be found in the file @c examples/platforms/cloud.xml. This
949 examples models a NetZone using Vivaldi that contains other NetZones that use different
952 #### Example platform files ####
954 This is an automatically generated list of example files that use the None
955 routing model (the path is given relative to SimGrid's source directory):
957 @verbinclude example_filelist_routing_vivaldi
960 @subsection ps_dec Defining routes
962 There are currently four different ways to define routes:
964 | Name | Description |
965 | ------------------------------------------------- | ----------------------------------------------------------------------------------- |
966 | @ref pf_tag_route "route" | Used to define route between host/router |
967 | @ref pf_tag_zoneroute "zoneRoute" | Used to define route between different zones |
968 | @ref pf_tag_bypassroute "bypassRoute" | Used to supersede normal routes as calculated by the network model between host/router; e.g., can be used to use a route that is not the shortest path for any of the shortest-path routing models. |
969 | @ref pf_tag_bypassasroute "bypassZoneRoute" | Used in the same way as bypassRoute, but for zones |
971 Basically all those tags will contain an (ordered) list of references
972 to link that compose the route you want to define.
974 Consider the example below:
977 <route src="Alice" dst="Bob">
978 <link_ctn id="link1"/>
979 <link_ctn id="link2"/>
980 <link_ctn id="link3"/>
984 The route here from host Alice to Bob will be first link1, then link2,
985 and finally link3. What about the reverse route? @ref pf_tag_route "Route" and
986 @ref pf_tag_zoneroute "zoneroute" have an optional attribute @c symmetrical, that can
987 be either @c YES or @c NO. @c YES means that the reverse route is the same
988 route in the inverse order, and is set to @c YES by default. Note that
989 this is not the case for bypass*Route, as it is more probable that you
990 want to bypass only one default route.
992 For an @ref pf_tag_zoneroute "zoneroute", things are just slightly more complicated, as you have
993 to give the id of the gateway which is inside the zone you want to access ...
994 So it looks like this:
997 <zoneroute src="zone1" dst="zone2"
998 gw_src="router1" gw_dst="router2">
999 <link_ctn id="link1"/>
1003 gw == gateway, so when any message are trying to go from zone1 to zone2,
1004 it means that it must pass through router1 to get out of the zone, then
1005 pass through link1, and get into zone2 by being received by router2.
1006 router1 must belong to zone1 and router2 must belong to zone2.
1008 @subsubsection pf_tag_linkctn <link_ctn>
1010 This entity has only one purpose: Refer to an already existing
1011 @ref pf_tag_link "<link/>" when defining a route, i.e., it
1012 can only occur as a child of @ref pf_tag_route "<route/>"
1014 | Attribute name | Mandatory | Values | Description |
1015 | --------------- | --------- | ------ | ----------- |
1016 | id | yes | String | The identifier of the link that should be added to the route. |
1017 | direction | maybe | UP@|DOWN | If the link referenced by @c id has been declared as @ref pf_sharing_policy_splitduplex "SPLITDUPLEX", this indicates which direction the route traverses through this link: UP or DOWN. If you don't use SPLITDUPLEX, do not use this attribute or SimGrid will not find the right link.
1019 #### Example Files ####
1021 This is an automatically generated list of example files that use the @c <link_ctn/>
1022 entity (the path is given relative to SimGrid's source directory):
1024 @verbinclude example_filelist_xmltag_linkctn
1026 @subsubsection pf_tag_zoneroute <zoneRoute>
1028 The purpose of this entity is to define a route between two
1029 NetZones. Recall that all zones form a tree, so to connect two
1030 sibiling zones, you must give such a zoneRoute specifying the source
1031 and destination zones, along with the gateway in each zone (ie, the
1032 point to reach within that zone to reach the netzone), and the list of
1033 links in the ancestor zone to go from one zone to another.
1035 So, to go from an host @c src_host that is within zone @c src, to an
1036 host @c dst_host that is within @c dst, you need to:
1038 - move within zone @c src, from @c src_host to the specified @c gw_src;
1039 - traverse all links specified by the zoneRoute (they are supposed to be within the common ancestor);
1040 - move within zone @c dst, from @c gw_dst to @c dst_host.
1042 #### Attributes ####
1044 | Attribute name | Mandatory | Values | Description |
1045 | --------------- | --------- | ------ | ----------- |
1046 | src | yes | String | The identifier of the source zone |
1047 | dst | yes | String | See the @c src attribute |
1048 | gw_src | yes | String | The gateway that will be used within the src zone; this can be any @ref pf_tag_host "Host" or @ref pf_router "Router" defined within the src zone. |
1049 | gw_dst | yes | String | Same as @c gw_src, but with the dst zone instead. |
1050 | symmetrical | no | YES@|NO (Default: YES) | If this route is symmetric, the opposite route (from dst to src) will also be declared implicitly. |
1055 <zone id="zone0" routing="Full">
1056 <cluster id="my_cluster_1" prefix="c-" suffix=".me"
1057 radical="0-149" speed="1000000000" bw="125000000" lat="5E-5"
1058 bb_bw="2250000000" bb_lat="5E-4"/>
1060 <cluster id="my_cluster_2" prefix="c-" suffix=".me"
1061 radical="150-299" speed="1000000000" bw="125000000" lat="5E-5"
1062 bb_bw="2250000000" bb_lat="5E-4"/>
1064 <link id="backbone" bandwidth="1250000000" latency="5E-4"/>
1066 <zoneroute src="my_cluster_1" dst="my_cluster_2"
1067 gw_src="c-my_cluster_1_router.me"
1068 gw_dst="c-my_cluster_2_router.me">
1069 <link_ctn id="backbone"/>
1071 <zoneroute src="my_cluster_2" dst="my_cluster_1"
1072 gw_src="c-my_cluster_2_router.me"
1073 gw_dst="c-my_cluster_1_router.me">
1074 <link_ctn id="backbone"/>
1079 @subsubsection pf_tag_route <route>
1081 The principle is the same as for
1082 @ref pf_tag_zoneroute "ZoneRoute": The route contains a list of links that
1083 provide a path from @c src to @c dst. Here, @c src and @c dst can both be either a
1084 @ref pf_tag_host "host" or @ref pf_router "router". This is mostly useful for the
1085 @ref pf_routing_model_full "Full routing model" as well as for the
1086 @ref pf_routing_model_shortest_path "shortest-paths" based models (as they require
1087 topological information).
1090 | Attribute name | Mandatory | Values | Description |
1091 | --------------- | --------- | ---------------------- | ----------- |
1092 | src | yes | String | The value given to the source's "id" attribute |
1093 | dst | yes | String | The value given to the destination's "id" attribute. |
1094 | symmetrical | no | YES@| NO (Default: YES) | If this route is symmetric, the opposite route (from dst to src) will also be declared implicitly. |
1099 A route in the @ref pf_routing_model_full "Full routing model" could look like this:
1101 <route src="Tremblay" dst="Bourassa">
1102 <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"/>
1106 A route in the @ref pf_routing_model_shortest_path "Shortest-Path routing model" could look like this:
1108 <route src="Tremblay" dst="Bourassa">
1113 You must only have one link in your routes when you're using them to provide
1114 topological information, as the routes here are simply the edges of the
1115 (network-)graph and the employed algorithms need to know which edge connects
1116 which pair of entities.
1118 @subsubsection pf_tag_bypassasroute bypasszoneroute
1120 As said before, once you choose
1121 a model, it (most likely; the constant network model, for example, doesn't) calculates routes for you. But maybe you want to
1122 define some of your routes, which will be specific. You may also want
1123 to bypass some routes defined in lower level zone at an upper stage:
1124 <b>bypasszoneroute</b> is the tag you're looking for. It allows to
1125 bypass routes defined between already defined between zone (if you want
1126 to bypass route for a specific host, you should just use byPassRoute).
1127 The principle is the same as zoneroute: <b>bypasszoneroute</b> contains
1128 list of links that are in the path between src and dst.
1130 #### Attributes ####
1132 | Attribute name | Mandatory | Values | Description |
1133 | --------------- | --------- | ---------------------- | ----------- |
1134 | src | yes | String | The value given to the source zone's "id" attribute |
1135 | dst | yes | String | The value given to the destination zone's "id" attribute. |
1136 | gw_src | yes | String | The value given to the source gateway's "id" attribute; this can be any host or router within the src zone |
1137 | gw_dst | yes | String | The value given to the destination gateway's "id" attribute; this can be any host or router within the dst zone|
1138 | symmetrical | no | YES@| NO (Default: YES) | If this route is symmetric, the opposite route (from dst to src) will also be declared implicitly. |
1143 <bypasszoneRoute src="my_cluster_1" dst="my_cluster_2"
1144 gw_src="my_cluster_1_router"
1145 gw_dst="my_cluster_2_router">
1146 <link_ctn id="link_tmp"/>
1150 This example shows that link @c link_tmp (definition not displayed here) directly
1151 connects the router @c my_cluster_1_router in the source cluster to the router
1152 @c my_cluster_2_router in the destination router. Additionally, as the @c symmetrical
1153 attribute was not given, this route is presumed to be symmetrical.
1155 @subsubsection pf_tag_bypassroute bypassRoute
1157 As said before, once you choose
1158 a model, it (most likely; the constant network model, for example, doesn't) calculates routes for you. But maybe you want to
1159 define some of your routes, which will be specific. You may also want
1160 to bypass some routes defined in lower level zone at an upper stage:
1161 <b>bypassRoute</b> is the tag you're looking for. It allows to bypass
1162 routes defined between <b>host/router</b>. The principle is the same
1163 as route: <b>bypassRoute</b> contains list of links references of
1164 links that are in the path between src and dst.
1166 #### Attributes ####
1168 | Attribute name | Mandatory | Values | Description |
1169 | --------------- | --------- | ---------------------- | ----------- |
1170 | src | yes | String | The value given to the source zone's "id" attribute |
1171 | dst | yes | String | The value given to the destination zone's "id" attribute. |
1172 | symmetrical | no | YES @| NO (Default: YES) | If this route is symmetric, the opposite route (from dst to src) will also be declared implicitly. |
1177 <bypassRoute src="host_1" dst="host_2">
1178 <link_ctn id="link_tmp"/>
1182 This example shows that link @c link_tmp (definition not displayed here) directly
1183 connects host @c host_1 to host @c host_2. Additionally, as the @c symmetrical
1184 attribute was not given, this route is presumed to be symmetrical.
1186 @subsection pb_baroex Basic Routing Example
1188 Let's say you have an zone named zone_Big that contains two other zone, zone_1
1189 and zone_2. If you want to make a host (h1) from zone_1 with another one
1190 (h2) from zone_2 then you'll have to proceed as follows:
1191 @li First, you have to ensure that a route is defined from h1 to the
1192 zone_1's exit gateway and from h2 to zone_2's exit gateway.
1193 @li Then, you'll have to define a route between zone_1 to zone_2. As those
1194 zone are both resources belonging to zone_Big, then it has to be done
1195 at zone_big level. To define such a route, you have to give the
1196 source zone (zone_1), the destination zone (zone_2), and their respective
1197 gateway (as the route is effectively defined between those two
1198 entry/exit points). Elements of this route can only be elements
1199 belonging to zone_Big, so links and routers in this route should be
1200 defined inside zone_Big. If you choose some shortest-path model,
1201 this route will be computed automatically.
1203 As said before, there are mainly 2 tags for routing:
1204 @li <b>zoneroute</b>: to define routes between two <b>zone</b>
1205 @li <b>route</b>: to define routes between two <b>host/router</b>
1207 As we are dealing with routes between zone, it means that those we'll
1208 have some definition at zone_Big level. Let consider zone_1 contains 1
1209 host, 1 link and one router and zone_2 3 hosts, 4 links and one router.
1210 There will be a central router, and a cross-like topology. At the end
1211 of the crosses arms, you'll find the 3 hosts and the router that will
1212 act as a gateway. We have to define routes inside those two zone. Let
1213 say that zone_1 contains full routes, and zone_2 contains some Floyd
1214 routing (as we don't want to bother with defining all routes). As
1215 we're using some shortest path algorithms to route into zone_2, we'll
1216 then have to define some <b>route</b> to gives some topological
1217 information to SimGrid. Here is a file doing it all:
1220 <zone id="zone_Big" routing="Dijkstra">
1221 <zone id="zone_1" routing="Full">
1222 <host id="zone_1_host1" speed="1000000000"/>
1223 <link id="zone_1_link" bandwidth="1250000000" latency="5E-4"/>
1224 <router id="zone_1_gateway"/>
1225 <route src="zone_1_host1" dst="zone_1_gateway">
1226 <link_ctn id="zone_1_link"/>
1229 <zone id="zone_2" routing="Floyd">
1230 <host id="zone_2_host1" speed="1000000000"/>
1231 <host id="zone_2_host2" speed="1000000000"/>
1232 <host id="zone_2_host3" speed="1000000000"/>
1233 <link id="zone_2_link1" bandwidth="1250000000" latency="5E-4"/>
1234 <link id="zone_2_link2" bandwidth="1250000000" latency="5E-4"/>
1235 <link id="zone_2_link3" bandwidth="1250000000" latency="5E-4"/>
1236 <link id="zone_2_link4" bandwidth="1250000000" latency="5E-4"/>
1237 <router id="central_router"/>
1238 <router id="zone_2_gateway"/>
1239 <!-- routes providing topological information -->
1240 <route src="central_router" dst="zone_2_host1"><link_ctn id="zone_2_link1"/></route>
1241 <route src="central_router" dst="zone_2_host2"><link_ctn id="zone_2_link2"/></route>
1242 <route src="central_router" dst="zone_2_host3"><link_ctn id="zone_2_link3"/></route>
1243 <route src="central_router" dst="zone_2_gateway"><link_ctn id="zone_2_link4"/></route>
1245 <link id="backbone" bandwidth="1250000000" latency="5E-4"/>
1247 <zoneroute src="zone_1" dst="zone_2"
1248 gw_src="zone_1_gateway"
1249 gw_dst="zone_2_gateway">
1250 <link_ctn id="backbone"/>
1255 @section pf_other Other tags
1257 The following tags can be used inside a @<platform@> tag even if they are not
1258 directly describing the platform:
1260 - @ref pf_tag_config passes configuration options, e.g. to change the network model;
1261 - @ref pf_tag_prop gives user-defined properties to various elements
1263 @subsection pf_tag_config <config>
1265 Adding configuration flags into the platform file is particularly
1266 useful when the described platform is best used with specific
1267 flags. For example, you could finely tune SMPI in your platform file directly.
1269 | Attribute | Values | Description |
1270 | ---------- | ------------------- | ---------------------------------------------- |
1271 | id | String (optional) | This optional identifier is ignored by SimGrid |
1273 * **Included tags:** @ref pf_tag_prop to specify a given configuration item (see @ref options).
1275 Any such configuration must be given at the very top of the platform file.
1280 <?xml version='1.0'?>
1281 <!DOCTYPE platform SYSTEM "https://simgrid.org/simgrid.dtd">
1282 <platform version="4">
1284 <prop id="maxmin/precision" value="0.000010" />
1285 <prop id="cpu/optim" value="TI" />
1286 <prop id="network/model" value="SMPI" />
1287 <prop id="smpi/bw-factor" value="65472:0.940694;15424:0.697866;9376:0.58729" />
1290 <zone id="zone0" routing="Full">
1294 @subsection pf_tag_prop <prop>
1296 Defines a user-defined property, identified with a name and having a
1297 value. You can specify such properties to most kind of resources:
1298 @ref pf_tag_zone, @ref pf_tag_host, @ref pf_tag_storage,
1299 @ref pf_tag_cluster and @ref pf_tag_link. These values can be retrieved
1300 at runtime with MSG_zone_property() or simgrid::s4u::NetZone::property(),
1301 or similar functions.
1303 | Attribute | Values | Description |
1304 | --------- | ---------------------- | ----------------------------------------------------------------------------------------- |
1305 | id | String (mandatory) | Identifier of this property. Must be unique for a given property holder, eg host or link. |
1306 | value | String (mandatory) | Value of this property; The semantic is completely up to you. |
1308 * **Included tags:** none.
1313 <prop id="Operating System" value="Linux" />
1317 @subsection pf_trace trace and trace_connect
1319 Both tags are an alternate way to pass files containing information on
1320 availability, state etc. to an entity. (See also @ref howto_churn).
1321 Instead of referring to the file directly in the host, link, or
1322 cluster tag, you proceed by defining a trace with an id corresponding
1323 to a file, later a host/link/cluster, and finally using trace_connect
1324 you say that the file trace must be used by the entity.
1330 <zone id="zone0" routing="Full">
1331 <host id="bob" speed="1000000000"/>
1333 <trace id="myTrace" file="bob.trace" periodicity="1.0"/>
1334 <trace_connect trace="myTrace" element="bob" kind="POWER"/>
1338 The order here is important. @c trace_connect must come
1339 after the elements @c trace and @c host, as both the host
1340 and the trace definition must be known when @c trace_connect
1341 is parsed; the order of @c trace and @c host is arbitrary.
1344 #### @c trace attributes ####
1347 | Attribute name | Mandatory | Values | Description |
1348 | --------------- | --------- | ---------------------- | ----------- |
1349 | id | yes | String | Identifier of this trace; this is the name you pass on to @c trace_connect. |
1350 | file | no | String | Filename of the file that contains the information - the path must follow the style of your OS. You can omit this, but then you must specifiy the values inside of <trace> and </trace> - see the example below. |
1351 | trace_periodicity | yes | String | This is the same as for @ref pf_tag_host "hosts" (see there for details) |
1353 Here is an example of trace when no file name is provided:
1356 <trace id="myTrace" periodicity="1.0">
1363 #### @c trace_connect attributes ####
1365 | Attribute name | Mandatory | Values | Description |
1366 | --------------- | --------- | ---------------------- | ----------- |
1367 | kind | no | HOST_AVAIL@|POWER@|<br/>LINK_AVAIL@|BANDWIDTH@|LATENCY (Default: HOST_AVAIL) | Describes the kind of trace. |
1368 | trace | yes | String | Identifier of the referenced trace (specified of the trace's @c id attribute) |
1369 | element | yes | String | The identifier of the referenced entity as given by its @c id attribute |
1371 @section pf_hints Hints, tips and frequently requested features
1373 Now you should know at least the syntax and be able to create a
1374 platform by your own. However, after having ourselves wrote some platforms, there
1375 are some best practices you should pay attention to in order to
1376 produce good platform and some choices you can make in order to have
1377 faster simulations. Here's some hints and tips, then.
1379 @subsection pf_hints_search Finding the platform example that you need
1381 Most platform files that we ship are in the @c examples/platforms
1382 folder. The good old @c grep tool can find the examples you need when
1383 wondering on a specific XML tag. Here is an example session searching
1384 for @ref pf_trace "trace_connect":
1387 % cd examples/platforms
1388 % grep -R -i -n --include="*.xml" "trace_connect" .
1389 ./two_hosts_platform_with_availability_included.xml:26:<trace_connect kind="SPEED" trace="A" element="Cpu A"/>
1390 ./two_hosts_platform_with_availability_included.xml:27:<trace_connect kind="HOST_AVAIL" trace="A_failure" element="Cpu A"/>
1391 ./two_hosts_platform_with_availability_included.xml:28:<trace_connect kind="SPEED" trace="B" element="Cpu B"/>
1392 ./two_hosts.xml:17: <trace_connect trace="Tremblay_power" element="Tremblay" kind="SPEED"/>
1395 @subsection pf_hint_generating How to generate different platform files?
1397 This is actually a good idea to search for a better platform file,
1398 that better fit the need of your study. To be honest, the provided
1399 examples are not representative of anything. They exemplify our XML
1400 syntax, but that's all. small_platform.xml for example was generated
1401 without much thought beyond that.
1403 The best thing to do when possible is to write your own platform file,
1404 that model the platform on which you run your code. For that, you
1405 could use <a href="https://gitlab.inria.fr/simgrid/platform-calibration">our
1406 calibration scripts</a>. This leads to very good fits between the
1407 platform, the model and the needs. The g5k.xml example resulted of
1408 such an effort, which also lead to <a href="https://github.com/lpouillo/topo5k/">an
1409 ongoing attempt</a> to automatically extract the SimGrid platform from
1410 the <a href="http://grid5000.fr/">Grid'5000</a> experimental platform.
1411 But it's hard to come up with generic models. Don't take these files
1412 too seriously. Actually, you should always challenge our models and
1413 your instanciation if the accuracy really matters to you (see <a
1414 href="https://hal.inria.fr/hal-00907887">this discussion</a>).
1416 But such advices only hold if you have a real platform and a real
1417 application at hand. It's moot for more abstract studies working on
1418 ideas and algorithms instead of technical artefacts. Well, in this
1419 case, there unfortunately is nothing better than this old and rusty
1420 <a href="http://pda.gforge.inria.fr/tools/download.html">simulacrum</a>.
1421 This project is dormant since over 10 years (and you will have to
1422 update the generated platforms with <tt>bin/simgrid_update_xml</tt> to
1423 use them), but that's the best we have for this right now....
1425 @subsection pf_zone_h Zone Hierarchy
1426 The network zone design allows SimGrid to go fast, because computing route is
1427 done only for the set of resources defined in the current zone. If you're using
1428 only a big zone containing all resource with no zone into it and you're
1429 using Full model, then ... you'll loose all interest into it. On the
1430 other hand, designing a binary tree of zone with, at the lower level,
1431 only one host, then you'll also loose all the good zone hierarchy can
1432 give you. Remind you should always be "reasonable" in your platform
1433 definition when choosing the hierarchy. A good choice if you try to
1434 describe a real life platform is to follow the zone described in
1435 reality, since this kind of trade-off works well for real life
1438 @subsection pf_exit_zone Exit Zone: why and how
1439 Users that have looked at some of our platforms may have notice a
1440 non-intuitive schema ... Something like that:
1444 <zone id="zone_4" routing="Full">
1445 <zone id="exitzone_4" routing="Full">
1446 <router id="router_4"/>
1448 <cluster id="cl_4_1" prefix="c_4_1-" suffix="" radical="1-20" speed="1000000000" bw="125000000" lat="5E-5" bb_bw="2250000000" bb_lat="5E-4"/>
1449 <cluster id="cl_4_2" prefix="c_4_2-" suffix="" radical="1-20" speed="1000000000" bw="125000000" lat="5E-5" bb_bw="2250000000" bb_lat="5E-4"/>
1450 <link id="4_1" bandwidth="2250000000" latency="5E-5"/>
1451 <link id="4_2" bandwidth="2250000000" latency="5E-5"/>
1452 <link id="bb_4" bandwidth="2250000000" latency="5E-4"/>
1453 <zoneroute src="cl_4_1"
1455 gw_src="c_4_1-cl_4_1_router"
1456 gw_dst="c_4_2-cl_4_2_router">
1457 <link_ctn id="4_1"/>
1458 <link_ctn id="bb_4"/>
1459 <link_ctn id="4_2"/>
1461 <zoneroute src="cl_4_1"
1463 gw_src="c_4_1-cl_4_1_router"
1465 <link_ctn id="4_1"/>
1466 <link_ctn id="bb_4"/>
1468 <zoneroute src="cl_4_2"
1470 gw_src="c_4_2-cl_4_2_router"
1472 <link_ctn id="4_2"/>
1473 <link_ctn id="bb_4"/>
1478 In the zone_4, you have an exitzone_4 defined, containing only one router,
1479 and routes defined to that zone from all other zone (as cluster is only a
1480 shortcut for an zone, see cluster description for details). If there was
1481 an upper zone, it would define routes to and from zone_4 with the gateway
1482 router_4. It's just because, as we did not allowed (for performances
1483 issues) to have routes from an zone to a single host/router, you have to
1484 enclose your gateway, when you have zone included in your zone, within an
1485 zone to define routes to it.
1487 @subsection pf_P2P_tags P2P or how to use coordinates
1488 SimGrid allows you to use some coordinated-based system, like vivaldi,
1489 to describe a platform. The main concept is that you have some peers
1490 that are located somewhere: this is the function of the
1491 <b>coordinates</b> of the @<peer@> or @<host@> tag. There's nothing
1492 complicated in using it, here is an example:
1495 <?xml version='1.0'?>
1496 <!DOCTYPE platform SYSTEM "https://simgrid.org/simgrid.dtd">
1497 <platform version="4">
1499 <zone id="zone0" routing="Vivaldi">
1500 <host id="100030591" coordinates="25.5 9.4 1.4" speed="1.5Gf" />
1501 <host id="100036570" coordinates="-12.7 -9.9 2.1" speed="7.3Gf" />
1503 <host id="100429957" coordinates="17.5 6.7 18.8" speed="8.3Gf" />
1508 Coordinates are then used to calculate latency (in microseconds)
1509 between two hosts by calculating the distance between the two hosts
1510 coordinates with the following formula: distance( (x1, y1, z1), (x2,
1511 y2, z2) ) = euclidian( (x1,y1), (x2,y2) ) + abs(z1) + abs(z2)
1513 In other words, we take the euclidian distance on the two first
1514 dimensions, and then add the absolute values found on the third
1515 dimension. This may seem strange, but it was found to allow better
1516 approximations of the latency matrices (see the paper describing
1519 Note that the previous example defines a routing directly between hosts but it could be also used to define a routing between zone.
1520 That is for example what is commonly done when using peers (see Section @ref pf_peer).
1522 <?xml version='1.0'?>
1523 <!DOCTYPE platform SYSTEM "https://simgrid.org/simgrid.dtd">
1524 <platform version="4">
1526 <zone id="zone0" routing="Vivaldi">
1527 <peer id="peer-0" coordinates="173.0 96.8 0.1" speed="730Mf" bw_in="13.38MBps" bw_out="1.024MBps" lat="500us"/>
1528 <peer id="peer-1" coordinates="247.0 57.3 0.6" speed="730Mf" bw_in="13.38MBps" bw_out="1.024MBps" lat="500us" />
1529 <peer id="peer-2" coordinates="243.4 58.8 1.4" speed="730Mf" bw_in="13.38MBps" bw_out="1.024MBps" lat="500us" />
1533 In such a case though, we connect the zone created by the <b>peer</b> tag with the Vivaldi routing mechanism.
1534 This means that to route between zone1 and zone2, it will use the coordinates of router_zone1 and router_zone2.
1535 This is currently a convention and we may offer to change this convention in the DTD later if needed.
1536 You may have noted that conveniently, a peer named FOO defines an zone named FOO and a router named router_FOO, which is why it works seamlessly with the <b>peer</b> tag.
1539 @subsection pf_routing_howto_choose_wisely Choosing wisely the routing model to use
1542 Choosing wisely the routing model to use can significantly fasten your
1543 simulation/save your time when writing the platform/save tremendous
1544 disk space. Here is the list of available model and their
1545 characteristics (lookup: time to resolve a route):
1547 @li <b>Full</b>: Full routing data (fast, large memory requirements,
1549 @li <b>Floyd</b>: Floyd routing data (slow initialization, fast
1550 lookup, lesser memory requirements, shortest path routing only).
1551 Calculates all routes at once at the beginning.
1552 @li <b>Dijkstra</b>: Dijkstra routing data (fast initialization, slow
1553 lookup, small memory requirements, shortest path routing only).
1554 Calculates a route when necessary.
1555 @li <b>DijkstraCache</b>: Dijkstra routing data (fast initialization,
1556 fast lookup, small memory requirements, shortest path routing
1557 only). Same as Dijkstra, except it handles a cache for latest used
1559 @li <b>None</b>: No routing (usable with Constant network only).
1560 Defines that there is no routes, so if you try to determine a
1561 route without constant network within this zone, SimGrid will raise
1563 @li <b>Vivaldi</b>: Vivaldi routing, so when you want to use coordinates
1564 @li <b>Cluster</b>: Cluster routing, specific to cluster tag, should
1567 @subsection pf_loopback I want to specify the characteristics of the loopback link!
1569 Each routing model automatically adds a loopback link for each declared host, i.e.,
1570 a network route from the host to itself, if no such route is declared in the XML
1571 file. This default link has a bandwidth of 498 Mb/s, a latency of 15 microseconds,
1572 and is <b>not</b> shared among network flows.
1574 If you want to specify the characteristics of the loopback link for a given host, you
1575 just have to specify a route from this host to itself with the desired characteristics
1576 in the XML file. This will prevent the routing model to add and use the default
1579 @subsection pf_switch I want to describe a switch but there is no switch tag!
1581 Actually we did not include switch tag. But when you're trying to
1582 simulate a switch, assuming
1583 fluid bandwidth models are used (which SimGrid uses by default unless
1584 ns-3 or constant network models are activated), the limiting factor is
1585 switch backplane bandwidth. So, essentially, at least from
1586 the simulation perspective, a switch is similar to a
1587 link: some device that is traversed by flows and with some latency and
1588 so,e maximum bandwidth. Thus, you can simply simulate a switch as a
1590 can be connected to this "switch", which is then included in routes just
1594 @subsection pf_multicabinets I want to describe multi-cabinets clusters!
1596 You have several possibilities, as usual when modeling things. If your
1597 cabinets are homogeneous and the intercabinet network negligible for
1598 your study, you should just create a larger cluster with all hosts at
1601 In the rare case where your hosts are not homogeneous between the
1602 cabinets, you can create your cluster completely manually. For that,
1603 create an As using the Cluster routing, and then use one
1604 <cabinet> for each cabinet. This cabinet tag can only be used an
1605 As using the Cluster routing schema, and creating
1607 Be warned that creating a cluster manually from the XML with
1608 <cabinet>, <backbone> and friends is rather tedious. The
1609 easiest way to retrieve some control of your model without diving into
1610 the <cluster> internals is certainly to create one separate
1611 <cluster> per cabinet and interconnect them together. This is
1612 what we did in the G5K example platform for the Graphen cluster.
1614 @subsection pf_platform_multipath I want to express multipath routing in platform files!
1616 It is unfortunately impossible to express the fact that there is more
1617 than one routing path between two given hosts. Let's consider the
1618 following platform file:
1621 <route src="A" dst="B">
1624 <route src="B" dst="C">
1627 <route src="A" dst="C">
1632 Although it is perfectly valid, it does not mean that data traveling
1633 from A to C can either go directly (using link 3) or through B (using
1634 links 1 and 2). It simply means that the routing on the graph is not
1635 trivial, and that data do not following the shortest path in number of
1636 hops on this graph. Another way to say it is that there is no implicit
1637 in these routing descriptions. The system will only use the routes you
1638 declare (such as <route src="A" dst="C"><link_ctn
1639 id="3"/></route>), without trying to build new routes by aggregating
1642 You are also free to declare platform where the routing is not
1643 symmetrical. For example, add the following to the previous file:
1646 <route src="C" dst="A">
1652 This makes sure that data from C to A go through B where data from A
1653 to C go directly. Don't worry about realism of such settings since
1654 we've seen ways more weird situation in real settings (in fact, that's
1655 the realism of very regular platforms which is questionable, but
1656 that's another story).
