Pular para o conteúdo
DispatchAtlas
Buscar

Modelo de benchmarks

dispatchatlas.bench define a evidência de benchmarks antes de os solvers ou as campanhas a consumirem. Uma família de benchmarks declara sua taxonomia, perfil de domínio, classe de perfil, suposições, evidência de citação, envelope de escala, espaço de nomes de semente, e esquema de saída. A materialização valida cada problema gerado com dispatchatlas.core, caracteriza a instância, envolve-a em um envelope de proveniência, e registra hashes estáveis.

O catálogo cobre famílias genéricas de escalonamento de otimização-combinatória e escalonamento de distributed-computing como pares co-iguais, de modo que a plataforma não é uma ferramenta apenas-de-distributed-computing.

Exemplo executável: examples/benchmark_continuum.py gera, caracteriza, e cataloga um atlas de benchmarks contínuo em tempo real.

Famílias de escalonamento

Cada família de escalonamento é materializada como um par de catálogo de primeira-classe com ao menos um perfil gerador. A tabela de catálogo abaixo é gerada a partir do registro de geradores de benchmarks e da matriz de citação, de modo que seus totais de família e citações de fonte são contáveis a partir das próprias linhas. Um gráfico de distribuição de famílias acima da tabela mostra como os perfis de família se distribuem entre as categorias de famílias de escalonamento.

Generated from the benchmark generator registry and the citation matrix: 69 family profiles across 8 scheduling families distributed-computing (47), flow-shop (7), job-shop (6), machine-scheduling (3), open-shop (1), rcpsp (3), rcpsp-max (1), setup-flow-shop (1).

Benchmark family distributiondistributed-computing47flow-shop7job-shop6machine-scheduling3rcpsp3open-shop1rcpsp-max1setup-flow-shop1
How the 69 benchmark families distribute across scheduling-family categories — the IoE / Edge–Fog–Cloud continuum families dominate, while the classical combinatorial-scheduling families (job-shop, flow-shop, machine-scheduling, RCPSP, setup flow-shop) demonstrate the corpus’s reach into general scheduling. Hover or focus a bar to read its count.

Showing 69 of 69 family profiles.

Benchmark families — 69 rows, build-inlined from the public family bundle.
Distinctive againstEvidenceCitation statusSource citations
accelerator-coscheduling
accelerator-coscheduling
a heterogeneous datacenter job needs a compute resource and a scarce accelerator at the same time, so every job co-allocates two simultaneous resource demands held together for its whole run; the accelerator pool is scarce, so jobs sharing an accelerator serialize on it while jobs on disjoint resources run in parallel, and the scheduler reasons over a multi-resource co-allocation problem rather than a single-unit-demand one
distributed-computingioe-completepublished single-resource continuum schedulers (one unit resource demand per task, not the simultaneous co-allocation of a compute resource and a scarce accelerator held together under a Pareto contract)smokecitation-backed
  • ghodsi-drf-2011 Ghodsi, Zaharia, Hindman, Konwinski, Shenker, and Stoica, USENIX NSDI 2011.
  • gandiva-2018 Xiao et al., USENIX OSDI 2018, 595-610.
aerial-edge
aerial-edge
a loitering UAV is a flying fog node that serves the ground region beneath it for a fixed loiter window before moving to the next pass, so sorties group into successive loiter passes pinned to the fog tier the platform embodies while overhead
distributed-computingioe-completeaerial-edge MEC simulators (no flying-fog loiter placement)smokecitation-backed
  • edgecloudsim-2018 Sonmez, Ozgovde, and Ersoy, Transactions on Emerging Telecommunications Technologies, 29(11):e3493, 2018.
  • aerial-mec-survey-2022 Song, Qin, Hao, Hou, Wang, and Sun, arXiv:2208.13965, 2022.
anytime-inference
anytime-inference
an edge accelerator serves deep-learning inference requests that each complete a mandatory minimal-accuracy early-exit branch and may run an optional refinement to full accuracy when their latency deadline allows; every request declares a mandatory duration below its full duration and a tight latency deadline, and arrivals are spaced shorter than a full inference so requests queue and contend, so the scheduler decides which requests refine and which deliver the early-exit result -- an imprecise-computation quality-versus-timeliness trade-off
distributed-computingioe-completepublished edge-cloud split inference and datacenter inference serving (a fixed full-accuracy computation per request), neither of which lets a request drop an optional refinement to meet its deadline so the schedule order trades accuracy for timelinesssmokecitation-backed
blocking-flow-shop
blocking-flow-shop
flow-shopclassicalsmokecitation-backed
bulk-synchronous-graph
bulk-synchronous-graph
an iterative graph computation runs as a sequence of supersteps separated by global barriers, so each superstep's vertex partitions compute and exchange messages and every partition of the next superstep waits on all partitions of the prior one; the slowest partition therefore gates each superstep, and the scheduler reasons over a barrier-synchronized partition-balancing problem rather than an overlap-friendly pipeline or an independent-task one
distributed-computingioe-completepublished pipeline or independent-task schedulers (an overlap-friendly wavefront or unsynchronized tasks, not supersteps separated by global barriers where the slowest partition gates each round under a Pareto contract)smokecitation-backed
  • valiant-bsp-1990 Valiant, Communications of the ACM, 33(8):103-111, 1990.
  • pregel-2010 Malewicz, Austern, Bik, Dehnert, Horn, Leiser, and Czajkowski, ACM SIGMOD 2010, 135-146.
carbon-aware
carbon-aware
flexible jobs defer to low-carbon-intensity windows under a time-varying grid carbon signal while honoring their SLA deadlines
distributed-computingioe-completeElectricity-Maps grid carbon-intensity and CityLearn carbon-aware community signalssmokecitation-backed
  • electricity-maps-grid-ci Electricity Maps, electricitymaps-contrib (open-source parsers and public grid carbon-intensity data).
  • citylearn-v2 Nweye, Kaspar, Buscemi, et al., 2024 (arXiv:2405.03848); Vazquez-Canteli et al., ACM BuildSys 2019.
cloud-independent
cloud-edge-independent
distributed-computingdomain-specificsmokecitation-backed
  • cloudsim-2011 Calheiros, Ranjan, Beloglazov, De Rose, and Buyya, Software: Practice and Experience, 2011.
  • dynamic-cloudsim-2015 Bux and Leser, Future Generation Computer Systems, 2015.
  • edge-vision-2016 Shi, Cao, Zhang, Li, and Xu, IEEE Internet of Things Journal, 2016.
coflow-scheduling
coflow-scheduling
a distributed-computing stage completes only when the last network transfer of its coflow lands, not the first, so a coflow's completion time is the maximum over its member flows; every coflow emits data-heavy flow tasks that place freely across the fabric plus a barrier task that depends on all of them, so the barrier gates the group and the coflow-completion-time is an all-or-nothing footprint
distributed-computingioe-completedatacenter coflow schedulers (no continuum tier-placement barrier)smokecitation-backed
compact-job-shop
compact-job-shop
job-shopclassicalsmokecitation-backed
  • or-library-1990 Beasley, Journal of the Operational Research Society, 41(11):1069-1072, 1990.
  • taillard-1993 Taillard, European Journal of Operational Research, 1993.
confidential-edge
confidential-edge
tasks are classified by data sensitivity -- a confidential task that touches protected data must execute inside the enclave-capable trusted tier so its data never leaves the trusted boundary, while a public task draws a free placement affinity across the fabric
distributed-computingioe-completeedge enclave runtimes (no security-classified Pareto placement)smokecitation-backed
  • occlum-asplos-2020 Shen, Tian, Chen, Chen, Wang, Xu, Yan, and Xia, International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) 2020.
  • confidential-edge-zobaed-2025 Zobaed and Amini Salehi, Software: Practice and Experience, 2025.
cyber-physical
cyber-physical
control cycles arrive periodically under a time-varying tariff
distributed-computingioe-completeperiodic hard-real-time task models and smart-grid demand-side scheduling formulations (no edge-fog-cloud tier placement under a Pareto contract)smokecitation-backed
  • liu-layland-1973 Liu and Layland, Journal of the ACM, 20(1):46-61, 1973.
  • smart-grid-dsm-2012 Logenthiran, Srinivasan, and Shun, IEEE Transactions on Smart Grid, 3(3):1244-1252, 2012.
  • citylearn-v2 Nweye, Kaspar, Buscemi, et al., 2024 (arXiv:2405.03848); Vazquez-Canteli et al., ACM BuildSys 2019.
data-locality
data-locality
a query over a large dataset is cheaper to run where the data already resides than to ship the data across the WAN fabric; each task's input lives on one tier (edge sensor logs, fog warm aggregates, or cloud cold archives) and the task pins to that tier so its heavy input never crosses the fabric, with the data tiers cycled so placement spans the whole edge-fog-cloud continuum
distributed-computingioe-completecluster locality schedulers (no continuum data-residency placement)smokecitation-backed
datacenter-colocation
datacenter-colocation
latency-sensitive service jobs and deferrable batch jobs share multi-tenant cells, so high-priority-band jobs claim capacity ahead of low-band jobs under contention
distributed-computingioe-completeGoogle Borg ClusterData2019 priority-tiered cell tracessmokecitation-backed
  • google-cluster-data Reiss, Wilkes, and Hellerstein, Google cluster-usage traces, 2011; Wilkes, ClusterData2019 v3, 2020.
  • borg-2020 Tirmazi, Barker, Deng, Haque, Qin, Hand, Harchol-Balter, and Wilkes, EuroSys 2020.
digital-twin-sync
digital-twin-sync
each physical asset periodically syncs its state to its fog/cloud twin and must finish within a freshness (Age-of-Information) window before the twin's state goes stale
distributed-computingioe-completedigital-twin edge frameworks (no joint Pareto placement)smokecitation-backed
  • digital-twin-diten-2022 Tang, Chen, Koketsu Rodrigues, Zhao, and Kato, IEEE Open Journal of the Communications Society, 3:1360-1381, 2022.
disaggregated-memory
disaggregated-memory
a CXL-pooled cloud platform backs each socket with a small local DRAM tier and a shared far-memory pool, and every VM draws a long-tailed memory working set, so a few memory-hungry tenants dominate a socket's local budget while far-memory access inflates a VM's runtime in proportion to the working set it spills to the pool; the scheduler reasons over local-versus-pool placement rather than core-count placement
distributed-computingioe-completepublished CXL memory-pooling and tiered-memory systems (socket-local page placement, no edge-fog-cloud tier scheduling under a Pareto contract)smokecitation-backed
  • pond-cxl-2023 Li, Berger, Hsu, Ernst, Zardoshti, Novakovic, Shah, Rajadnya, Lee, Agarwal, Hill, Fontoura, and Bianchini, ASPLOS 2023.
  • tpp-cxl-2023 Maruf, Wang, Dhanotia, Weiner, Agarwal, Bhattacharya, Petersen, Chowdhury, Kanaujia, and Chauhan, ASPLOS 2023.
distributed-assembly-flow-shop
distributed-assembly-flow-shop
flow-shopclassicalsmokecitation-backed
  • hatami-ruiz-2013 Hatami, Ruiz, and Andrés-Romano, International Journal of Production Research, 51(17):5292-5308, 2013.
  • taillard-1993 Taillard, European Journal of Operational Research, 1993.
distributed-flexible-job-shop
distributed-flexible-job-shop
job-shopstructurally-complexsmokecitation-backed
distributed-permutation-flow-shop
distributed-permutation-flow-shop
flow-shopclassicalsmokecitation-backed
  • naderi-ruiz-2010 Naderi and Ruiz, Computers & Operations Research, 37(4):754-768, 2010.
  • taillard-1993 Taillard, European Journal of Operational Research, 1993.
distributed-training-gang
distributed-training-gang
a GPU cluster runs synchronous data-parallel training jobs; each job is a gang of workers that must START TOGETHER on distinct accelerators (every all-reduce step synchronizes the workers), so a job cannot begin until enough accelerators are free simultaneously; the workers reuse the accelerator pool and jobs arrive over time, so jobs queue and the scheduler decides which job acquires a full simultaneously-free worker set first -- an all-or-nothing gang co-start, not an independent placement of each worker
distributed-computingioe-completepublished GPU-cluster and inference-serving families that place each task independently; none requires a whole job's worker set to co-start simultaneously on distinct accelerators, so no other family forbids a partial start -- the gang-scheduling all-or-nothing constraint under a Pareto contractsmokecitation-backed
  • gang-scheduling-1995 Feitelson and Rudolph, Job Scheduling Strategies for Parallel Processing (JSSPP), LNCS 949, Springer, 1995, 1-18.
  • large-minibatch-sgd-2017 Goyal, Dollar, Girshick, Noordhuis, Wesolowski, Kyrola, Tulloch, Jia, and He, arXiv:1706.02677, 2017.
distributed-transaction
distributed-transaction
a partitioned database runs transactions that each acquire exclusive locks on a variable read/write set of data shards, so every transaction co-allocates a randomly drawn subset of shards held together for its whole run; two transactions whose shard sets intersect serialize while disjoint transactions commit in parallel, so the scheduler reasons over a variable-cardinality lock-conflict graph rather than a fixed two-resource hold
distributed-computingioe-completepublished replica-placement schedulers (a single shard pinned per task for locality, not a variable-cardinality exclusive lock set co-allocated per transaction forming a conflict graph under a Pareto contract)smokecitation-backed
  • calvin-2012 Thomson, Diamond, Weng, Ren, Shao, and Abadi, ACM SIGMOD 2012.
  • spanner-2012 Corbett et al., USENIX OSDI 2012.
edge-offloading
edge-offloading-mec
each task chooses between local edge execution and remote offload
distributed-computingioe-completeiFogSim MEC offloading scenariossmokecitation-backed
  • ifogsim-2017 Gupta, Vahid Dastjerdi, Ghosh, and Buyya, Software: Practice and Experience, 47(9):1275-1296, 2017.
  • ifogsim2-2022 Mahmud, Pallewatta, Goudarzi, and Buyya, Journal of Systems and Software, 190:111351, 2022.
  • edge-vision-2016 Shi, Cao, Zhang, Li, and Xu, IEEE Internet of Things Journal, 2016.
  • shanghai-telecom-edge Wang, Guo, Zhang, Yang, Zhou, and Shen, IEEE Transactions on Mobile Computing, 20(3):939-953, 2021.
edge-placement
edge-placement
services place on edge servers near their user population and migrate as demand shifts across base-station coverage cells
distributed-computingioe-completeEUA edge-user-allocation and Shanghai-Telecom base-station placement tracessmokecitation-backed
  • eua-dataset Lai, He, Abdelrazek, Chen, Bao, Grundy, Hosking, and Yang, ICSOC 2018.
  • shanghai-telecom-edge Wang, Guo, Zhang, Yang, Zhou, and Shen, IEEE Transactions on Mobile Computing, 20(3):939-953, 2021.
elastic-serverless-autoscale
elastic-serverless-autoscale
a serverless platform serves function invocations on a shared worker pool, and every invocation is moldable: it may run on one, two, or four concurrent workers, where a wider allocation runs shorter by a sublinear speedup but spends more total worker-seconds; each request declares its execution modes and arrivals are spaced shorter than a base invocation so the pool is contended, so the scheduler picks each invocation's worker width -- a moldable latency-versus-cost choice rather than a fixed resource hold
distributed-computingioe-completepublished serverless cold-start and fixed-width co-allocation families (accelerator co-scheduling, fpga partitioning), each of which holds one fixed resource set per task; none lets a request choose among several worker-count modes so the schedule order trades latency for resource cost under a Pareto contractsmokecitation-backed
facility-assignment
facility-assignment
machine-schedulingclassicalsmokecitation-backed
  • hooker-2007 Hooker, Operations Research, 55(3):588-602, 2007.
failure-recovery
failure-recovery
a failed task re-places its checkpoint state to a surviving tier
distributed-computingioe-completeBorg cluster failure-event tracessmokecitation-backed
  • google-cluster-data Reiss, Wilkes, and Hellerstein, Google cluster-usage traces, 2011; Wilkes, ClusterData2019 v3, 2020.
  • borg-2020 Tirmazi, Barker, Deng, Haque, Qin, Hand, Harchol-Balter, and Wilkes, EuroSys 2020.
federated-learning
federated-learning
each training round selects a subset of heterogeneous, straggler-prone edge clients that train on non-IID local data, then a fog or cloud aggregator combines their updates
distributed-computingioe-completeFedScale and Oort federated-learning device-participation benchmarkssmokecitation-backed
  • fedscale-2022 Lai, Dai, Singapuram, Liu, Zhu, Madhyastha, and Chowdhury, ICML 2022 (PMLR 162); arXiv:2105.11367.
  • oort-2021 Lai, Zhu, Madhyastha, and Chowdhury, USENIX OSDI 2021.
flexible-job-shop
flexible-job-shop
job-shopclassicalsmokecitation-backed
flow-shop
permutation-flow-shop
flow-shopclassicalsmokecitation-backed
  • taillard-1993 Taillard, European Journal of Operational Research, 1993.
  • vallada-ruiz-framinan-2015 Vallada, Ruiz, and Framinan, European Journal of Operational Research, 240(3):666-677, 2015.
fpga-partitioning
fpga-partitioning
a multi-tenant reconfigurable FPGA hosts tenant kernels that each occupy a contiguous region of fabric tiles, so every kernel co-allocates a contiguous run of tiles held together for its whole residency; two kernels whose tile intervals overlap cannot be co-resident and serialize while kernels on disjoint tile spans run in parallel, so the scheduler reasons over an interval-overlap conflict graph rather than a fixed two-resource hold or a random-subset lock set
distributed-computingioe-completepublished accelerator co-scheduling (a fixed compute-plus-accelerator pair) and shard-lock transactions (a random subset of resources), neither of which constrains the co-allocated set to a spatially contiguous tile interval whose overlaps form an interval conflict graph under a Pareto contractsmokecitation-backed
frontierco-fjsp
frontierco-fjsp
job-shopclassicalsmokecitation-backed
  • frontierco Feng, Sun, Li, Talwalkar, and Yang, arXiv:2505.16952, 2026 (ICLR 2026).
  • behnke-geiger-2012 Behnke and Geiger, Helmut-Schmidt-University Hamburg working paper, 2012.
  • naderi-roshanaei-2022 Naderi and Roshanaei, INFORMS Journal on Optimization, 4(1):1-28, 2022.
generative-inference-serving
generative-inference-serving
a transformer inference replica batches autoregressive requests that each hold key-value-cache memory proportional to their token count for the whole decode, so a long-tailed sequence mix fragments a fixed cache budget and the scheduler reasons over memory-bound admission rather than GPU-count placement; every request's decode duration and KV-cache demand scale with its drawn token count
distributed-computingioe-completepublished single-replica generative-model serving systems (no edge-fog-cloud tier placement under a Pareto contract)smokecitation-backed
gpu-ml
gpu-ml
training and inference jobs claim accelerators and gang-schedule replicas
distributed-computingioe-completeAlibaba PAI, Philly, and Helios GPU-cluster tracessmokecitation-backed
hybrid-flow-shop
hybrid-flow-shop
flow-shopclassicalsmokecitation-backed
  • ruiz-vazquez-2010 Ruiz and Vázquez-Rodríguez, European Journal of Operational Research, 205(1):1-18, 2010.
  • taillard-1993 Taillard, European Journal of Operational Research, 1993.
immersive-xr
immersive-xr
each extended-reality frame runs a latency-critical perception, render, and display pipeline placed across the device, edge, and cloud within a hard motion-to-photon deadline
distributed-computingioe-completeILLIXR extended-reality systems testbedsmokecitation-backed
  • illixr-2021 Huzaifa, Desai, Grayson, et al., IEEE IISWC 2021.
intermittent-edge
intermittent-edge
a batteryless sensor harvests ambient energy into a small buffer, runs until the buffer depletes, then sleeps to recharge; a job too large for one duty-cycle window is checkpointed at power loss and resumed in the next, so it is a precedence chain of edge-pinned per-window segments each bounded by the constant energy window
distributed-computingioe-completeintermittent-computing runtimes (no continuum energy-window placement)smokecitation-backed
iot-edge
iot-edge
many small sensor readings arrive periodically and aggregate at the edge
distributed-computingioe-completepublished wireless-sensor-network telemetry datasets and in-network aggregation deployments (raw sensor readings, not edge-fog-cloud tier scheduling under a Pareto contract)smokecitation-backed
  • intel-lab-data Bodik, Hong, Guestrin, Madden, Paskin, and Thibaux, Intel Berkeley Research Lab / MIT CSAIL, 2004.
  • milano-cdr-2015 Barlacchi, De Nadai, Larcher, et al., Scientific Data, 2:150055, 2015.
job-shop
job-shop
job-shopclassicalsmokecitation-backed
  • or-library-1990 Beasley, Journal of the Operational Research Society, 41(11):1069-1072, 1990.
  • taillard-1993 Taillard, European Journal of Operational Research, 1993.
  • applegate-cook-1991 Applegate and Cook, ORSA Journal on Computing, 3(2):149-156, 1991.
  • demirkol-mehta-uzsoy-1998 Demirkol, Mehta, and Uzsoy, European Journal of Operational Research, 109(1):137-141, 1998.
  • storer-wu-vaccari-1992 Storer, Wu, and Vaccari, Management Science, 38(10):1495-1509, 1992.
  • rl4co-2023 Berto, Hua, Park, et al., arXiv:2306.17100, 2023 (KDD 2025).
kv-cache-placement
kv-cache-placement
a distributed key-value cache tier serves a catalog of cache objects whose request rate follows a heavy Zipfian popularity skew, so a few hot objects absorb most of the traffic; each object is placement-flexible, carrying one single-node mode per cache-tier node, so the scheduler chooses which tier node hosts it, and an object's working set -- the transfer volume staged onto its host tier -- scales with its popularity rank, so the hottest object carries the largest working set and a cold-tail object the smallest; a naive uniform placement strands a hot, large-working-set object on a far tier and pays its transfer across the fabric, while a locality-aware placement pins the hottest objects to near tiers to shrink makespan and cost
distributed-computingioe-completepublished cache and content-placement families (consistent-hashing replica placement, CDN content distribution) that place each object uniformly or by a hash; none scales each object's working set with a Zipfian popularity rank so the hot objects carry a strictly larger transfer volume, making popularity-skewed near-tier pinning the lever a locality-aware placement pulls under the Pareto contractsmokecitation-backed
machine-scheduling
machine-scheduling-unrelated
machine-schedulingclassicalsmokecitation-backed
  • or-library-1990 Beasley, Journal of the Operational Research Society, 41(11):1069-1072, 1990.
microservice-dag
microservice-dag
services form an acyclic call graph pinned by role to a tier
distributed-computingioe-completeAlibaba v2021 microservice-trace call graphssmokecitation-backed
mixed-criticality
mixed-criticality
a safety-critical real-time mix runs tasks of differing criticality, and a high-criticality task is budgeted with a conservative high-assurance worst-case execution time and a tight deadline while a low-criticality task carries a smaller best-effort budget and a loose deadline, so the criticality tiering lives in the duration and deadline structure; the scheduler reasons over which assured-criticality tasks to guarantee under contention rather than a uniform-assurance deadline-scheduling one
distributed-computingioe-completepublished uniform-assurance real-time deadline schedulers (one worst-case execution time and deadline class per task, not criticality-tiered WCET budgets with tighter high-assurance deadlines under a Pareto contract)smokecitation-backed
moe-expert-parallel
moe-expert-parallel
a sparsely-activated mixture-of-experts model routes each token batch to one expert and the experts are spread across devices, and expert popularity is long-tailed, so a few hot experts receive most token batches while many stay cold and the all-to-all routing exchange dominates fabric traffic; the scheduler reasons over an expert-placement and load-balancing problem rather than a dense uniform-replica serving one
distributed-computingioe-completepublished dense generative-model serving systems (uniform per-replica KV-cache admission, not sparse token-to-expert routing under load imbalance and a Pareto contract)smokecitation-backed
multi-objective-pfsp
multi-objective-pfsp
flow-shopclassicalsmokecitation-backed
  • minella-ruiz-2008 Minella, Ruiz, and Ciavotta, INFORMS Journal on Computing, 20(3):451-471, 2008.
  • taillard-1993 Taillard, European Journal of Operational Research, 1993.
multi-project-rcpsp
multi-project-rcpsp
rcpspstructurally-complexsmokecitation-backed
  • van-eynde-vanhoucke-2020 Van Eynde and Vanhoucke, Journal of Scheduling, 23(3):301-325, 2020.
  • psplib-1997 Kolisch and Sprecher, European Journal of Operational Research, 96(1):205-216, 1997.
multi-tenant-fair-share
multi-tenant-fair-share
a shared cluster serves several tenants whose workloads compete for one node pool; each task is placement-flexible, carrying one single-node mode per pool node, so the scheduler chooses which node it occupies; tenants are sized asymmetrically, so even a load-balanced placement leaves the heavy tenants holding a larger fraction of their busiest node -- a higher dominant resource share -- than the light ones, and a fairness-aware scheduler rebalances placement to shrink the dominant-share spread
distributed-computingioe-completepublished multi-tenant colocation families (Borg-style priority colocation, vm allocation) that fix each task's resource and score makespan or cost; none lets the scheduler choose each tenant task's node and scores the dominant-resource-share spread between tenants as a fairness objectivesmokecitation-backed
  • ghodsi-drf-2011 Ghodsi, Zaharia, Hindman, Konwinski, Shenker, and Stoica, USENIX NSDI 2011.
  • borg-2020 Tirmazi, Barker, Deng, Haque, Qin, Hand, Harchol-Balter, and Wilkes, EuroSys 2020.
network-slicing
network-slicing
isolated slice classes (latency-critical, broadband, massive-IoT) each carry their own service-level deadline and placement, and same-class slices spread across tiers for resilience
distributed-computingioe-complete5G slicing orchestration (no joint Pareto placement)smokecitation-backed
  • network-slicing-afolabi-2018 Afolabi, Taleb, Samdanis, Ksentini, and Flinck, IEEE Communications Surveys & Tutorials, 20(3):2429-2453, 2018.
no-wait-flow-shop
no-wait-flow-shop
flow-shopclassicalsmokecitation-backed
open-shop
open-shop
open-shopclassicalsmokecitation-backed
orbital-edge
orbital-edge
tasks schedule across ground terminals, moving low-earth-orbit satellites, and cloud backhaul under time-varying connectivity as satellites enter and leave coverage and hand work over
distributed-computingioe-completeLENS real-measurement LEO satellite-network tracessmokecitation-backed
pipeline-parallel-training
pipeline-parallel-training
a deep network is split into successive pipeline stages pinned across edge-to-cloud tiers and the training mini-batch is divided into micro-batches, so each micro-batch flows forward stage by stage while each stage runs its micro-batches in issue order; the two precedence families form a diagonal wavefront whose warm-up and cool-down idle slots are the pipeline bubbles, and deeper stages carry rising compute, so the scheduler reasons over a stage-partition and bubble-minimizing problem rather than a synchronous data-parallel all-reduce one
distributed-computingioe-completepublished data-parallel / gang-scheduled training systems (synchronous all-reduce over co-located replicas, not a stage-by-micro-batch pipeline wavefront with warm-up and cool-down bubbles under a Pareto contract)smokecitation-backed
  • gpipe-2019 Huang, Cheng, Bapna, Firat, Chen, Chen, Lee, Ngiam, Le, Wu, and Chen, NeurIPS 2019.
  • pipedream-2019 Narayanan, Harlap, Phanishayee, Seshadri, Devanur, Ganger, Gibbons, and Zaharia, SOSP 2019.
rcpsp
rcpsp-renewable
rcpspstructurally-complexsmokecitation-backed
  • psplib-1997 Kolisch and Sprecher, European Journal of Operational Research, 96(1):205-216, 1997.
rcpsp-max
rcpsp-max
rcpsp-maxstructurally-complexsmokecitation-backed
  • bartusch-moehring-1988 Bartusch, Möhring, and Radermacher, Annals of Operations Research, 16(1):199-240, 1988.
  • psplib-1997 Kolisch and Sprecher, European Journal of Operational Research, 96(1):205-216, 1997.
rcpsp-multi-mode
rcpsp-multi-mode
rcpspstructurally-complexsmokecitation-backed
  • van-peteghem-vanhoucke-2014 Van Peteghem and Vanhoucke, European Journal of Operational Research, 235(1):62-72, 2014.
  • psplib-1997 Kolisch and Sprecher, European Journal of Operational Research, 96(1):205-216, 1997.
reentrant-fab
reentrant-fab
job-shopstructurally-complexsmokecitation-backed
  • smt2020-2020 Kopp, Hassoun, Kalir, and Mönch, IEEE Transactions on Semiconductor Manufacturing, 33(4):522-531, 2020.
replica-placement
replica-placement
a replica runs where its data shard already lives
distributed-computingioe-completeCRUSH replicated-data placementsmokecitation-backed
  • crush-2006 Weil, Brandt, Miller, and Maltzahn, Proceedings of the 2006 ACM/IEEE Conference on Supercomputing (SC'06), 2006.
  • google-cluster-data Reiss, Wilkes, and Hellerstein, Google cluster-usage traces, 2011; Wilkes, ClusterData2019 v3, 2020.
serverless-cold-start
serverless-cold-start
a cold invocation pays a container provisioning penalty
distributed-computingioe-completeAzure Functions serverless-in-the-wild tracessmokecitation-backed
  • azure-public-dataset Cortez et al., SOSP'17 Resource Central; Shahrad et al., USENIX ATC'20 Serverless in the Wild; Hadary et al., USENIX OSDI'20 Protean (AzureTracesForPacking2020).
  • sebs-2021 Copik, Kwasniewski, Besta, Podstawski, and Hoefler, ACM/IFIP Middleware 2021.
service-function-chain
service-function-chain
an NFV packet flow traverses a linear ordered chain of typed virtual network functions -- firewall, intrusion detection, deep packet inspection, address translation -- each pinned to a tier that hosts its function type, so the chain is a strict total order and the flow crosses the edge-fog-cloud fabric in a fixed sequence; the scheduler reasons over a chain-placement problem under an end-to-end latency budget rather than the branching role-pinned call graph of a microservice
distributed-computingioe-completepublished microservice call-graph schedulers (a branching role-pinned acyclic call graph, not a strict linear chain of function-typed network functions under an end-to-end latency budget and a Pareto contract)smokecitation-backed
  • nfv-survey-2016 Mijumbi, Serrat, Gorricho, Bouten, De Turck, and Boutaba, IEEE Communications Surveys & Tutorials, 18(1):236-262, 2016.
  • sfc-placement-2014 Mehraghdam, Keller, and Karl, IEEE CloudNet 2014, 7-13.
setup-flow-shop
setup-flow-shop
setup-flow-shopclassicalsmokecitation-backed
  • allahverdi-survey-2008 Allahverdi, Ng, Cheng, and Kovalyov, European Journal of Operational Research, 187(3):985-1032, 2008.
  • allahverdi-survey-2015 Allahverdi, European Journal of Operational Research, 246(2):345-378, 2015.
smartnic-offload
smartnic-offload
a SmartNIC-accelerated server pairs a fast host CPU with a low-power on-NIC processor, and every microservice draws a long-tailed compute intensity, so most are light enough to offload onto the energy-frugal NIC cores while a few compute-heavy services must stay host-bound; a service's runtime scales with its intensity, so the scheduler reasons over an energy-versus-latency offload-placement problem rather than a uniform host placement one
distributed-computingioe-completepublished mobile-edge computation-offloading models (device-to-edge latency offload, not in-server host-to-NIC energy offload under a Pareto contract)smokecitation-backed
split-inference-serving
split-inference-serving
each inference request partitions a deep model at a layer cut -- a light head runs the early layers on the edge near the sensor and a heavy tail runs the later layers in the cloud, consuming the head's intermediate feature map under a per-request end-to-end latency SLO
distributed-computingioe-completedatacenter inference serving (no edge-cloud partition placement)smokecitation-backed
  • alpaserve-osdi-2023 Li, Zheng, Zhong, Liu, Sheng, Jin, Huang, Chen, Zhang, Gonzalez, and Stoica, USENIX Symposium on Operating Systems Design and Implementation (OSDI) 2023, 663-679.
  • neurosurgeon-asplos-2017 Kang, Hauswald, Gao, Rovinski, Mudge, Mars, and Tang, International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) 2017.
spot-preemptible
spot-preemptible
a cloud provider rents idle capacity at a discount as revocable spot instances reclaimed after a short lease; eviction-tolerant batch work pins to the cloud spot tier under a hard lease deadline (the eviction horizon), while latency-critical interactive work pins to the stable edge on-demand tier with no eviction deadline
distributed-computingioe-completecloud spot schedulers (no continuum eviction-deadline placement)smokecitation-backed
  • harvest-vms-ambati-osdi-2020 Ambati, Goiri, Frujeri, Gun, Wang, Dolan, Corell, Pasupuleti, Moscibroda, Elnikety, Fontoura, and Bianchini, USENIX Symposium on Operating Systems Design and Implementation (OSDI) 2020, 735-751.
  • spot-eviction-yang-www-2022 Yang, Pang, Zhang, Qiao, Wang, Couturier, Bansal, Ram, Qin, Ma, Goiri, Cortez, Baladhandayutham, Ruhle, Rajmohan, Lin, and Zhang, The Web Conference (WWW) 2022 Companion, 152-156.
storage-io-tiering
storage-io-tiering
a tiered storage pool serves I/O-bound jobs whose runtime is dominated by moving a job's I/O volume through the storage node it lands on; each job is placement-flexible, carrying one single-node mode per tier node, and the mode duration is tier-dependent -- a seek floor plus the I/O volume divided by that tier's I/O bandwidth, which differs by tier (a fast cloud array sustains far more bytes/second than a slow edge disk); a job's I/O volume follows a heavy-tailed falloff over its I/O-demand rank, so a few I/O-heavy jobs carry most of the bytes and have a large cross-tier duration spread, while the light tail barely varies; a naive placement strands an I/O-heavy job on a low-bandwidth tier and pays its volume slowly, while a bandwidth-aware placement pins the heavy jobs to fast tiers to shrink makespan and cost
distributed-computingioe-completepublished storage-tiering and hierarchical-storage-management families that migrate blocks between fast and slow tiers by access frequency; none models the heterogeneous-tier I/O bandwidth as a placement-flexible per-tier mode whose duration is the I/O volume divided by that tier's bandwidth, so an I/O-heavy job's cross-tier duration spread is the bottleneck-relief lever a bandwidth-aware placement pulls under the Pareto contractsmokecitation-backed
  • msr-cambridge-block-2008 Narayanan, Donnelly, and Rowstron, ACM Transactions on Storage 4(3):10, 2008 (USENIX FAST 2008); MSR Cambridge enterprise volume block-I/O trace via the SNIA IOTTA repository.
streaming-window
streaming-window
events arrive online in bounded windows and must close within one
distributed-computingioe-completeParallel Workloads Archive online arrivalssmokecitation-backed
  • parallel-workloads-archive Feitelson, Tsafrir, and Krakov, Journal of Parallel and Distributed Computing, 74(10):2967-2982, 2014.
  • dspbench-2020 Bordin, Griebler, Mencagli, Geyer, and Fernandes, IEEE Access, 8:222900-222917, 2020.
time-sensitive-networking
time-sensitive-networking
each time-triggered flow releases on a fixed period and must finish within one cycle under a hard, jitter-free deadline
distributed-computingioe-completeEdgeCloudSim best-effort scenarios (no gating)smokecitation-backed
  • tsn-craciunas-2016 Craciunas, Serna Oliver, Chmelik, and Steiner, RTNS 2016, ACM, 183-192.
  • edgecloudsim-2018 Sonmez, Ozgovde, and Ersoy, Transactions on Emerging Telecommunications Technologies, 29(11):e3493, 2018.
unrelated-parallel-setup
unrelated-parallel-setup
machine-schedulingstructurally-complexsmokecitation-backed
  • vallada-ruiz-2011 Vallada and Ruiz, European Journal of Operational Research, 211(3):612-622, 2011.
  • allahverdi-survey-2015 Allahverdi, European Journal of Operational Research, 246(2):345-378, 2015.
vehicular-offloading
vehicular-offloading
a vehicle's tasks share an arrival time and a roadside-unit dwell deadline, and hand over from the roadside unit to the fog tier as the vehicle drives on
distributed-computingioe-completeEdgeCloudSim / SUMO vehicular-edge mobility scenariossmokecitation-backed
  • lust-scenario Codecà, Frank, Faye, and Engel, IEEE Intelligent Transportation Systems Magazine, 9(2):52-63, 2017.
  • edgecloudsim-2018 Sonmez, Ozgovde, and Ersoy, Transactions on Emerging Telecommunications Technologies, 29(11):e3493, 2018.
video-analytics
video-analytics
each camera streams frames that must be analyzed within a tight real-time latency bound, placed hierarchically with edge inference near the camera and cloud aggregation
distributed-computingioe-completeedge video-analytics clusters (no joint Pareto placement)smokecitation-backed
  • videoedge-2018 Hung, Ananthanarayanan, Bodik, Golubchik, Yu, Bahl, and Philipose, ACM/IEEE Symposium on Edge Computing (SEC) 2018, 115-131.
vm-allocation
vm-allocation
size-heterogeneous virtual-machine deployments pack onto hosts while each deployment's members spread across distinct failure domains for availability
distributed-computingioe-completeAzure Public Dataset Resource Central VM-allocation tracessmokecitation-backed
  • azure-public-dataset Cortez et al., SOSP'17 Resource Central; Shahrad et al., USENIX ATC'20 Serverless in the Wild; Hadary et al., USENIX OSDI'20 Protean (AzureTracesForPacking2020).
  • bitbrains-gwa-t12-2015 Shen, van Beek, and Iosup, IEEE/ACM CCGrid 2015, pp. 465-474; Bitbrains (now Solvinity) datacenter VM trace via the Grid Workloads Archive (GWA-T-12).
  • materna-gwa-t13-2014 Kohne, Spohr, Nagel, and Spinczyk, ACM CCB@Middleware 2014; Materna GmbH datacenter VM trace via the Grid Workloads Archive (GWA-T-13).
  • planetlab-workload-2012 Beloglazov and Buyya, Concurrency and Computation: Practice and Experience 24(13):1397-1420, 2012; PlanetLab/CoMon CPU-utilization workload bundled with CloudSim.
workflow-dag
workflow-dag
distributed-computingstructurally-complexsmokecitation-backed
  • stg-task-graphs Tobita and Kasahara, Journal of Scheduling, 5(5):379-394, 2002.
  • wfcommons-pegasus-instances Coleman, Casanova, Pottier, Kaushik, Deelman, and Ferreira da Silva, Future Generation Computer Systems, 128:16-27, 2022.
  • topcuoglu-heft-2002 Topcuoglu, Hariri, and Wu, IEEE Transactions on Parallel and Distributed Systems, 13(3):260-274, 2002.
  • canon-dag-bias-2019 Canon, El Sayah, and Héam, Euro-Par 2019, LNCS 11725.

Per-instance characterization and download eligibility live in the benchmark catalog; this table is the family-and-citation inventory.

O componente acima carrega o inventário gerado completo — as famílias núcleo clássicas e de distributed-computing abaixo mais o contínuo de famílias Edge–Fog–Cloud. Essas famílias núcleo fundacionais em forma de documento:

FamíliaPerfilClasse de perfilCorpus primário
Escalonamento de máquinas (R||Cmax, máquina não-relacionada)machine-scheduling-unrelatedclassicalOR-Library
Job-shopjob-shop-classicalclassicalOR-Library, Taillard
Job-shop flexível (FJSP)flexible-job-shopclassicalBrandimarte; Hurink-Jurisch-Thole
Flow-shop de permutaçãopermutation-flow-shopclassicalTaillard
Flow-shop de setup dependente-de-sequência (SDST)setup-flow-shopclassicalAllahverdi et al. (2008); Allahverdi (2015)
Escalonamento de projetos com-recursos-restritos (RCPSP)rcpsp-renewablestructurally-complexPSPLIB
Distributed-computing (cloud/edge)cloud-edge-capacitydomain-specificCloudSim; DynamicCloudSim; Edge vision
Distributed-computing (workflow DAG)workflow-dagstructurally-complexStandard Task Graph Set

As famílias de máquina-não-relacionada e job-shop flexível anexam uma matriz de tempo-de-execução CostModel a cada instância, exportada como uma matriz materializada junto ao JSON do problema. A família de flow-shop de setup dependente-de-sequência em vez disso anexa uma matriz de setup CostModel: uma troca entre trabalhos de famílias diferentes em uma máquina custa tempo de setup, de modo que o objetivo de setup recompensa agrupar trabalhos similares. Os corpora padrão nomeados são citados e vinculados apenas e nunca são redistribuídos dentro do repositório.

Várias famílias do contínuo exercitam co-alocação multi-recurso: cada tarefa demanda mais de um recurso ao mesmo tempo e o construtor os mantém juntos por toda a sua duração (ver Contratos de domínio). A família accelerator-coscheduling co-aloca um nó de computação e um acelerador escasso por trabalho; a família distributed-transaction co-aloca um conjunto de bloqueios de cardinalidade-variável de fragmentos de dados por transação; e a família fpga-partitioning co-aloca uma corrida espacialmente contígua de tiles de tecido reconfigurável por kernel de inquilino. As tarefas cujos conjuntos de recursos se intersectam serializam enquanto tarefas disjuntas rodam concorrentemente — o executável examples/inspect_coallocation.py torna a alavanca explícita.

Além da co-alocação, três famílias do contínuo exercitam suas próprias alavancas estruturais. A família elastic-serverless-autoscale exercita execução moldável: cada invocação de função declara mais de um modo de execução — um modo estreito apenas-em-casa e um modo largo que empresta um worker de um pequeno pool de rajada compartilhado para terminar mais cedo — de modo que o escalonamento escolhe um modo por tarefa e a ordem decide quais invocações reclamam o escasso modo largo-e-rápido. A família distributed-training-gang exercita co-escalonamento de gangue: os workers de um trabalho de treinamento data-paralelo síncrono compartilham uma gangue e devem co-iniciar em aceleradores distintos em um lançamento tudo-ou-nada — os workers reutilizam o pool de aceleradores e os trabalhos chegam ao longo do tempo, de modo que um trabalho não pode começar até que aceleradores suficientes se liberem simultaneamente, e a ordem decide qual trabalho adquire seu conjunto completo de workers primeiro. A família multi-tenant-fair-share exercita equidade de recurso-dominante: vários inquilinos de tamanho-assimétrico colocam tarefas flexíveis-em-colocação em um pool de nós compartilhado, e o objetivo de cota-de-recurso-dominante pontua a dispersão entre a cota dominante do inquilino mais- e menos-servido — de modo que a colocação, quais recursos cada inquilino ocupa, é a alavanca que a equilibra ou a enviesa. Os executáveis examples/serverless_autoscale_study.py, examples/distributed_training_gang_study.py, e examples/multi_tenant_fairshare_study.py tornam essas três alavancas explícitas.

Classes de perfil

Classe de perfilSignificado
classicalDeriva de um corpus padrão de otimização-combinatória.
structurally-complexCarrega estrutura de precedência, DAG, ou rede-de-recursos.
ioe-completeCenário distribuído Internet-of-Everything-completo.
trace-backedFundamentado em um traço de carga do mundo-real nomeado.
domain-specificAdaptado a um único domínio operacional.

Rótulos de evidência

RótuloUso
smokeInstâncias determinísticas pequenas para testes, exemplos, docs, e prévias.
exploratoryMaterial plausível que ainda não é respaldado-por-citações nem totalmente caracterizado.
grau de evidência candidatoMaterial respaldado-por-citações à espera de gates de piloto, estatísticos, e de campanha.
grau de evidência de campanha-completaEvidência que passou pelos gates de citação, caracterização, estatísticos, de divulgação, e de qualidade.

Os catálogos de fumaça nunca são evidência de avaliação final. Existem para provar que os geradores, a validação, a caracterização, as verificações de citação, e a persistência funcionam rapidamente. O catálogo do portal e os downloads previsualizam cada família do contínuo nesta pequena escala de fumaça (um pool de três-recursos); uma família de co-alocação não pode exibir paralelismo de recurso-disjunto em um pool tão pequeno, de modo que a estrutura distintiva é uma propriedade de escala-de-pesquisa. build_continuum_full_catalog() materializa cada família em sua escala de pesquisa declarada -- o pool de recursos e a contagem de tarefas maiores onde a estrutura de co-alocação, contenção, e colocação genuinamente se manifesta -- para pacotes de benchmarks de grau-pesquisa.

Taxonomia

A taxonomia cobre estruturas de escalonamento, ambientes, realismo de infraestrutura, características de objetivo, características de restrição, incerteza, e dinamismo. Os exemplos incluem workflows DAG, lotes de tarefas independentes, funções serverless, consolidação de contêineres e VM, ambientes edge e cloud, traços públicos, otimização multi-objetivo, deadlines, localidade de dados, churn, e chegadas dinâmicas.

Matriz de citação

As afirmações de benchmarks são verificadas contra CitationMatrix. As afirmações de grau de evidência candidato e de campanha-completa falham na validação a menos que referenciem fontes respaldadas-por-citações. Material não suportado deve permanecer exploratory até que evidência seja adicionada.

O conjunto de fontes é declarado em default_citation_matrix() com identificadores de fonte estáveis, referências resolúveis, e uma postura de licença registrada. Abrange três níveis: os corpora padrão de otimização-combinatória (citados e vinculados apenas, nunca empacotados), traços de cluster de produção, e um amplo conjunto de datasets contemporâneos do mundo-real do contínuo Edge–Fog–Cloud — traços de cluster de GPU e machine-learning, suites de benchmark de microsserviços e serverless, traços de workflow-científico, traços de trabalhos de supercomputador, datasets de edge-placement e mobilidade, datasets de IoT e demanda-celular, sinais de carbono e energia de rede, cargas de processamento-de-fluxo, benchmarks de participação-de-dispositivos de federated-learning, testbeds de sistemas de realidade-estendida, e traços de rede-satelital de órbita-terrestre-baixa:

Source idReferenceLicense posture
cloudsim-2011CloudSimacadêmico
dynamic-cloudsim-2015DynamicCloudSimacadêmico
edge-vision-2016Edge Computing: Vision and Challengesacadêmico
or-library-1990OR-Libraryapenas-citar-e-vincular
psplib-1997PSPLIBapenas-citar-e-vincular
van-eynde-vanhoucke-2020Resource-constrained multi-project scheduling: benchmark datasets and decoupled schedulingapenas-citar-e-vincular
taillard-1993Benchmarks for basic scheduling problemsapenas-citar-e-vincular
gonzalez-sahni-1976Open shop scheduling to minimize finish timeacadêmico
ruiz-vazquez-2010The hybrid flow shop scheduling problemacadêmico
smt2020-2020SMT2020—A Semiconductor Manufacturing Testbedcite-and-link-only
hooker-2007Planning and Scheduling by Logic-Based Benders Decompositionacadêmico
naderi-ruiz-2010The distributed permutation flowshop scheduling problemacadêmico
hall-sriskandarajah-1996A survey of machine scheduling problems with blocking and no-wait in processacadêmico
minella-ruiz-2008A review and evaluation of multiobjective algorithms for the flowshop scheduling problemacadêmico
bartusch-moehring-1988Scheduling project networks with resource constraints and time windowsacadêmico
hatami-ruiz-2013The distributed assembly permutation flowshop scheduling problemacadêmico
brandimarte-1993Flexible job shop by tabu searchapenas-citar-e-vincular
de-giovanni-pezzella-2010An improved genetic algorithm for the distributed and flexible job-shop scheduling problemacadêmico
frontiercoFrontierCO: real-world large-scale ML-for-CO evaluationusable-and-citable
behnke-geiger-2012Test instances for the flexible job shop scheduling problem with work centerscite-and-link-only
naderi-roshanaei-2022Critical-path-search logic-based Benders decomposition for flexible job shop schedulingcite-and-link-only
hurink-jurisch-thole-1994Multi-purpose-machine job shopapenas-citar-e-vincular
allahverdi-survey-2008A survey of scheduling problems with setup times or costsacadêmico
allahverdi-survey-2015The third comprehensive survey on scheduling problems with setup times/costsacadêmico
vallada-ruiz-2011A genetic algorithm for the unrelated parallel machine scheduling problem with sequence dependent setup timesacadêmico
stg-task-graphsA standard task graph set for fair evaluation of multiprocessor scheduling algorithmspadrão
google-cluster-dataBorg cluster workload tracesCC-BY-4.0, usável e citável
twitter-cache-traces-2020Twitter in-memory cache cluster tracesCC-BY-4.0, usável e citável
m100-exadata-2023M100 ExaData supercomputer telemetry and job traceCC-BY-4.0, usável e citável
borg-2020Borg: the Next Generationacadêmico
azure-public-datasetAzure Public DatasetCC-BY-4.0, usável e citável
alibaba-cluster-dataAlibaba cluster trace programapenas-citar
parallel-workloads-archiveParallel Workloads Archiveapenas-citar
bitbrains-gwa-t12-2015GWA-T-12 Bitbrains business-critical VM traceapenas-citar
materna-gwa-t13-2014GWA-T-13 Materna enterprise datacenter VM traceapenas-citar
msr-cambridge-block-2008MSR Cambridge enterprise block-I/O traces (Write Off-Loading)apenas-citar
planetlab-workload-2012PlanetLab CPU-utilization consolidation tracesapenas-citar
dspbench-2020DSPBench stream-processing benchmark suiteacadêmico
ifogsim-2017iFogSim toolkitacadêmico
edgecloudsim-2018EdgeCloudSimacadêmico
aerial-mec-survey-2022Aerial mobile edge computing surveyacadêmico
crush-2006CRUSH decentralized replica placementacadêmico
smart-grid-dsm-2012Smart-grid demand-side managementacadêmico
liu-layland-1973Scheduling Algorithms for Multiprogramming in a Hard-Real-Time Environmentacadêmico
philly-traces-2019Philly DNN-training cluster traceaberto, usável e citável
helios-traces-2021Helios GPU-datacenter traceaberto, usável e citável
acmetrace-2024AcmeTrace language-model-development cluster traceCC-BY-4.0
alibaba-pai-gpu-2020Alibaba PAI MLaaS GPU traceapenas-citar
azure-inference-trace-2023Azure model-inference serving traceCC-BY-4.0, usável e citável
deathstarbench-2019DeathStarBench microservice suiteApache-2.0, usável e citável
alibaba-microservice-characterization-2021Alibaba microservice characterizationacadêmico
train-ticket-benchmarkTrain-Ticket microservice benchmarkApache-2.0, usável e citável
sebs-2021SeBS serverless benchmark suiteBSD-3-Clause, usável e citável
functionbench-2019FunctionBench serverless workloadsaberto, usável e citável
wfcommons-pegasus-instancesWfCommons scientific-workflow instancesaberto, usável e citável
bharathi-synthetic-workflows-2008Synthetic scientific-workflow libraryaberto, usável e citável
nrel-eagle-jobs-2023NREL Eagle supercomputer job traceCC-BY-4.0, usável e citável
eua-datasetEdge-User-Allocation datasetaberto, usável e citável
shanghai-telecom-edgeShanghai Telecom access traceapenas-citar, pesquisa
lust-scenarioLuxembourg SUMO Traffic scenarioacadêmico, usável e citável
intel-lab-dataIntel Lab Data (54-node in-network aggregation)apenas-citar-e-vincular
milano-cdr-2015Milan and Trentino urban datasetCC-BY-4.0, usável e citável
citylearn-v2CityLearn grid-interactive benchmarkMIT, usável e citável
electricity-maps-grid-ciElectricity Maps grid carbon-intensityparsers abertos (MIT), o nível de dados varia
nexmark-benchmarkNEXMark streaming-query benchmarkaberto, usável e citável
yahoo-streaming-benchmark-2016Yahoo Streaming BenchmarkApache-2.0, usável e citável
fedscale-2022FedScale federated-learning benchmarkApache-2.0, usável e citável
oort-2021Oort guided participant selectionApache-2.0, usável e citável
illixr-2021ILLIXR extended-reality testbedNCSA, usável e citável
lens-2024LENS LEO satellite-network tracesCC-BY-SA-4.0, usável e citável
tsn-craciunas-2016Scheduling Real-Time Communication in IEEE 802.1Qbv TSNacadêmico
network-slicing-afolabi-2018Network Slicing and Softwarization: A Surveyacadêmico
digital-twin-diten-2022Survey on Digital Twin Edge Networks (DITEN) Toward 6Gacadêmico
videoedge-2018VideoEdge: Processing Camera Streams using Hierarchical Clustersacadêmico
alpaserve-osdi-2023AlpaServe: Statistical Multiplexing with Model Parallelism for Deep Learning Servingacadêmico
neurosurgeon-asplos-2017Neurosurgeon: Collaborative Intelligence Between the Cloud and Mobile Edgeacadêmico
occlum-asplos-2020Occlum: Secure and Efficient Multitasking Inside a Single Enclave of Intel SGXacadêmico
confidential-edge-zobaed-2025Confidential Computing across Edge-to-Cloud for Machine Learning: A Survey Studyacadêmico
batteryless-lucia-pldi-2015A Simpler, Safer Programming and Execution Model for Intermittent Systemsacadêmico
batteryless-hester-sensys-2017The Future of Sensing is Batteryless, Intermittent, and Awesomeacadêmico
harvest-vms-ambati-osdi-2020Providing SLOs for Resource-Harvesting VMs in Cloud Platformsacadêmico
spot-eviction-yang-www-2022Spot Virtual Machine Eviction Prediction in Microsoft Cloudacadêmico
delay-scheduling-zaharia-eurosys-2010Delay Scheduling: A Simple Technique for Achieving Locality and Fairness in Cluster Schedulingacadêmico
geo-analytics-iridium-pu-sigcomm-2015Low Latency Geo-distributed Data Analyticsacadêmico
coflow-chowdhury-hotnets-2012Coflow: A Networking Abstraction for Cluster Applicationsacadêmico
varys-chowdhury-sigcomm-2014Efficient Coflow Scheduling with Varysacadêmico
vllm-pagedattention-2023Efficient Memory Management for Large Language Model Serving with PagedAttentionacadêmico
orca-serving-2022Orca: A Distributed Serving System for Transformer-Based Generative Modelsacadêmico
pond-cxl-2023Pond: CXL-Based Memory Pooling Systems for Cloud Platformsacadêmico
tpp-cxl-2023TPP: Transparent Page Placement for CXL-Enabled Tiered-Memoryacadêmico
ipipe-smartnic-2019Offloading Distributed Applications onto SmartNICs using iPipeacadêmico
e3-smartnic-2019E3: Energy-Efficient Microservices on SmartNIC-Accelerated Serversacadêmico
switch-transformers-2022Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsityacadêmico
lina-moe-2023Accelerating Distributed MoE Training and Inference with Linaacadêmico
gpipe-2019GPipe: Efficient Training of Giant Neural Networks using Pipeline Parallelismacadêmico
pipedream-2019PipeDream: Generalized Pipeline Parallelism for DNN Trainingacadêmico
vestal-2007Preemptive Scheduling of Multi-criticality Systems with Varying Degrees of Execution Time Assuranceacadêmico
mixed-criticality-survey-2017A Survey of Research into Mixed Criticality Systemsacadêmico
valiant-bsp-1990A Bridging Model for Parallel Computationacadêmico
pregel-2010Pregel: A System for Large-Scale Graph Processingacadêmico
nfv-survey-2016Network Function Virtualization: State-of-the-Art and Research Challengesacadêmico
sfc-placement-2014Specifying and Placing Chains of Virtual Network Functionsacadêmico
ghodsi-drf-2011Dominant Resource Fairness: Fair Allocation of Multiple Resource Typesacadêmico
gandiva-2018Gandiva: Introspective Cluster Scheduling for Deep Learningacadêmico
calvin-2012Calvin: Fast Distributed Transactions for Partitioned Database Systemsacadêmico
spanner-2012Spanner: Google's Globally-Distributed Databaseacadêmico
amorphos-2018Sharing, Protection, and Compatibility for Reconfigurable Fabric with AmorphOSacadêmico
fpga-online-placement-2003Online Scheduling and Placement of Real-Time Tasks to Partially Reconfigurable Devicesacadêmico
imprecise-computation-1991Algorithms for Scheduling Imprecise Computationsacadêmico
branchynet-2016BranchyNet: Fast Inference via Early Exiting from Deep Neural Networksacadêmico
moldable-parallelizable-tasks-1992Approximate Algorithms for Scheduling Parallelizable Tasksacadêmico
serverless-rise-2019The Rise of Serverless Computingacadêmico
gang-scheduling-1995Parallel Job Scheduling: Issues and Approachesacadêmico
large-minibatch-sgd-2017Accurate, Large Minibatch SGD: Training ImageNet in 1 Houracadêmico
fisher-thompson-1963Probabilistic learning combinations of local job-shop scheduling rulesapenas-citar-e-vincular
lawrence-1984Resource constrained project scheduling: an experimental investigation of heuristic scheduling techniques (Supplement)apenas-citar-e-vincular
adams-balas-zawack-1988The shifting bottleneck procedure for job shop schedulingpadrão
applegate-cook-1991A computational study of the job-shop scheduling problempadrão
storer-wu-vaccari-1992New search spaces for sequencing problems with application to job shop schedulingpadrão
yamada-nakano-1992A genetic algorithm applicable to large-scale job-shop problemsapenas-citar-e-vincular
demirkol-mehta-uzsoy-1998Benchmarks for shop scheduling problemspadrão
dauzere-peres-paulli-1997General multiprocessor job-shop scheduling using tabu searchpadrão
vallada-ruiz-framinan-2015New hard benchmark for flowshop scheduling problems minimising makespanpadrão
ruiz-maroto-alcaraz-2005Flowshop scheduling with sequence dependent setup times using advanced metaheuristicspadrão
van-peteghem-vanhoucke-2014Metaheuristics for the multi-mode resource-constrained project scheduling problem on new dataset instancespadrão
debels-vanhoucke-2007A decomposition-based genetic algorithm for the resource-constrained project-scheduling problempadrão
cicirello-wtsds-benchmarkWeighted tardiness scheduling with sequence-dependent setups: a benchmark librarypadrão
crauwels-potts-vanwassenhove-1998Local search heuristics for the single machine total weighted tardiness scheduling problemacadêmico
van-hoorn-2018The current state of bounds on benchmark instances of the job-shop scheduling problemacadêmico
scheduleopt-benchmarksScheduleOpt benchmark collection: JSPLib and FJSPLib instances with verified boundspadrão
zenodo-pfsp-bks-2021Permutation flow-shop: best-known makespans and schedules for Taillard and VRF benchmarkspadrão
solutionsupdate-ugent-rcpspBest known results for the resource-constrained project scheduling problemindustrial
cloud-datasets-survey-2025Public Datasets for Cloud Computing: A Comprehensive Surveyacadêmico
fog-placement-survey-2023Fog/edge node placement surveyacadêmico
topcuoglu-heft-2002Performance-Effective and Low-Complexity Task Scheduling for Heterogeneous Computingacadêmico
canon-dag-bias-2019A Comparison of Random Task Graph Generation Methods for Scheduling Problemsacadêmico
ifogsim2-2022iFogSim2 extended fog/edge simulatoracadêmico
rl4co-2023RL4CO reinforcement-learning-for-CO benchmarkacadêmico

Suites de referência canônicas

O registro em default_reference_suites() registra as suites de instâncias publicadas canônicas às quais cada família de escalonamento genérica se ancora: identidade, família, contagem de instâncias, ponteiro de recuperação, e os rastreadores de melhores-soluções-conhecidas que publicam limites para a suite. As suites clássicas são apenas citadas e vinculadas — o DispatchAtlas nunca empacota nem redistribui arquivos de instâncias de terceiros.

SuiteFamíliaInstânciasDownloadRastreador BKS
fisher-thompsonjob-shop3OR-Libraryvan-hoorn-2018, scheduleopt-benchmarks
lawrencejob-shop40Espelho JSPLIBvan-hoorn-2018, scheduleopt-benchmarks
adams-balas-zawackjob-shop5Espelho JSPLIBvan-hoorn-2018, scheduleopt-benchmarks
applegate-cook-orbjob-shop10Espelho JSPLIBvan-hoorn-2018, scheduleopt-benchmarks
storer-wu-vaccarijob-shop20Espelho JSPLIBvan-hoorn-2018, scheduleopt-benchmarks
yamada-nakanojob-shop4Espelho JSPLIBvan-hoorn-2018, scheduleopt-benchmarks
taillard-jspjob-shop80Espelho JSPLIBvan-hoorn-2018, scheduleopt-benchmarks
demirkol-dmujob-shop80Espelho JSPLIBscheduleopt-benchmarks
brandimarte-mkjob-shop (flexível)15Espelho SchedulingLabscheduleopt-benchmarks
hurink-fjspjob-shop (flexível)198Espelho SchedulingLabscheduleopt-benchmarks
dauzere-peres-paullijob-shop (flexível)18Espelho SchedulingLabscheduleopt-benchmarks
taillard-pfspflow-shop120OR-Libraryzenodo-pfsp-bks-2021
vrf-pfspflow-shop480Site do grupo SOAzenodo-pfsp-bks-2021
sdst-taillard-ruizflow-shop de setup480Site do grupo SOAas melhores soluções acompanham as instâncias
cicirello-wt-sdsescalonamento de máquinas120Harvard Dataversecicirello-wtsds-benchmark
or-library-smtwtescalonamento de máquinas375OR-Librarycrauwels-potts-vanwassenhove-1998
vallada-ruiz-upmspescalonamento de máquinas1640 (reportado)Site do grupo SOA
psplibrcpsp2040Site do PSPLIBpsplib-1997
mmlibrcpsp4320 (reportado)Página do OR&Ssolutionsupdate-ugent-rcpsp
rg300rcpsp480Página do OR&Ssolutionsupdate-ugent-rcpsp

As suites com um parser empacotado (texto job-shop padrão, matrizes de flow-shop de Taillard, job-shop flexível .fjs, JSON WfFormat do WfCommons) são ingeridas com load_reference_suite(suite_id, instances_root=...) a partir de arquivos que o operador baixa e coloca sob uma árvore resources/ local. A ingestão executa inteiramente offline, reutiliza o mesmo envelope de validação, caracterização, hashing, e proveniência dos geradores sintéticos, e carimba cada problema com seu suite_id e upstream_instance_id. As suites apenas-de-registro são registradas com suas citações e ponteiros de recuperação sem um parser empacotado.

Registros de melhores-soluções-conhecidas

Os valores de melhor-solução-conhecida por-instância nunca são distribuídos com o DispatchAtlas. O operador os ingere como arquivos JSON sob um diretório privado resources/benchmarks/bks/, um arquivo por suite, cada um carregando schema_version, o suite_id, o source_id do rastreador, a data de recuperação, e as entradas de valor (id da instância, objetivo, valor, tipo ótimo-ou-limite-superior, limite inferior opcional). load_best_known_registry valida cada arquivo contra as suites de referência e a matriz de citação e falha fechando em suites desconhecidas, rastreadores desconhecidos, entradas duplicadas, ou limites inconsistentes. Sem um registro ingerido, as métricas de desvio-relativo estão simplesmente indisponíveis — elas nunca são computadas parcialmente, e nenhum valor de melhor-solução-conhecida aparece em qualquer superfície pública.

Divergências de calibração

As famílias genéricas sintéticas são ancoradas às suites canônicas sem alegar reproduzir seus esquemas de geração. As divergências conhecidas são documentadas em vez de ocultadas:

FamíliaConvenção publicadaConvenção sintética
flow-shop de setupsetups SDST-Taillard a 10/50/100/125% do tempo de processamentotrês famílias de setup, custo = família + 1
escalonamento de máquinas (R||Cmax)classes de duração U[1,100] e variantes de máquina-correlacionadafatores de velocidade por-par 0.5–2.0

As comparações contra as convenções publicadas passam pelas instâncias canônicas ingeridas, não pelas famílias sintéticas.

Caracterização

Cada problema materializado recebe descritores normalizados para:

  • densidade de oportunidade
  • esparsidade de compatibilidade
  • contenção e sobrecarga
  • profundidade de dependência
  • pressão de comunicação
  • intensidade de setup
  • viés de carga e heterogeneidade
  • conflito de objetivo
  • incerteza e dinamismo
  • sensibilidade do solver

Ponte de distribution-distance

Cada perfil de grau de evidência de campanha-completa declara uma ponte de distribution-distance: seu estado (synthetic, calibrated-synthetic, trace-backed, ou externally-sourced), evidência de calibração, cenário de domínio, cobertura de transferência e disrupção, e risco de distribution-distance residual. A promoção ao grau de evidência de campanha-completa falha fechando a menos que as coberturas de transferência e disrupção sejam declaradas, e um perfil calibrated-synthetic deve nomear a referência trace-backed contra a qual calibra. As famílias calibrated-synthetic nomeiam sua referência de traço explicitamente; as suites canônicas ingeridas carregam uma ponte externally-sourced, e o adaptador WfCommons é a primeira fonte de instâncias trace-backed analisada-externamente, dando ao distribution_distance_score uma perna de referência trace-backed real.

A métrica de calibração nomeada reporta a distância 1-Wasserstein (movedor-de-terra) por-característica entre as distribuições de características de caracterização de um perfil calibrated-synthetic e as de suas instâncias de referência trace-backed. As distâncias por-característica são agregadas a uma única pontuação de distribution-distance; uma pontuação acima do limiar de divergência-máxima (padrão 0.25) significa que o perfil derivou longe demais de sua referência e falha na calibração. A métrica é reproduzível a partir das instâncias materializadas e do traço de referência nomeado.

from dispatchatlas.bench import (
    build_smoke_catalog,
    metrics_from_instance_set,
    distribution_distance_score,
)
 
catalogs = {c.config.profile_id: c for c in build_smoke_catalog()}
synthetic = metrics_from_instance_set(catalogs["cloud-edge-capacity"])
reference = metrics_from_instance_set(catalogs["workflow-dag"])
score = distribution_distance_score(
    synthetic, reference, reference_trace_id="google-cluster-data"
)

Estratificação e seleção de subconjunto

difficulty_score agrega os descritores de contenção, sobrecarga, profundidade-de-dependência, e sensibilidade-do-solver em uma pontuação de dificuldade normalizada, e stratify_instances agrupa as instâncias materializadas em estratos de dificuldade baixa, média, e alta. select_benchmark_subset escolhe um subconjunto determinístico filtrado por família, classe de perfil, e estrato de dificuldade, ordenado por identificador de problema de modo que a seleção é reproduzível.

Catálogo de fumaça

from dispatchatlas.bench import build_smoke_catalog, smoke_benchmark_provider
 
catalogs = build_smoke_catalog(root_seed=20260527)
provider = smoke_benchmark_provider(root_seed=20260527)
 
first_problem = provider.get_problem(provider.list_problem_ids()[0])

O catálogo de fumaça empacotado inclui as duas famílias de distributed-computing (cloud/edge de tarefa-independente e workflow DAG) junto às quatorze famílias genéricas de escalonamento (escalonamento de máquinas, job-shop, job-shop flexível, flow-shop de permutação, flow-shop de setup dependente-de-sequência, RCPSP, open-shop, flow-shop híbrido, flow-shop de permutação distribuído, flow-shop sem-espera, flow-shop com-bloqueio, flow-shop de montagem distribuído, flow-shop de permutação multi-objetivo, e RCPSP/max) como pares co-iguais. Cada instância é determinística a partir da semente raiz e usa metadados de gerador respaldados-por-citações enquanto permanece rotulada como material de desenvolvimento de fumaça.

Catálogos completos candidatos

Os catálogos de campanha-completa candidatos usam o mesmo caminho de materialização com contagens de problema configuradas maiores e rótulos de evidência mais estritos:

from dispatchatlas.bench import build_full_catalog, full_benchmark_provider
 
catalogs = build_full_catalog(root_seed=2026052713, problem_count_per_profile=30)
provider = full_benchmark_provider(
    root_seed=2026052713,
    problem_count_per_profile=30,
)

Esses catálogos de grau de evidência de campanha-completa são respaldados-por-citações, caracterizados, ligados-por-hash, e ainda marcados como evidência de calibração até que os gates de campanha, divulgação, e publicação promovam afirmações específicas.