Features
MySolvers is a browser-based discrete-event simulator for queueing systems. Build models visually, watch them run live, then run them in batches for results you can put a confidence interval on — all without installing anything.
Visual Model Builder
Design queueing systems on an interactive canvas. Place components, drag a connection from one to the next, and rename, copy or delete them one at a time or as a selection. Zoom with the wheel or a pinch to work with models of any size.
Nine Core Components
Build any queueing topology from nine primitives: a Generator that produces entities at a configurable arrival rate, a FIFO Queue with a configurable capacity, a Server that processes entities at a configurable service rate, a Delay that holds every entity for a latency, a Router that divides traffic between branches, a Shared Server that serves everything it holds at once and slows under load, and a Resource Pool of a limited number of units that entities take at an Acquire, hold across several steps and return at a Release.
Five Distributions
Choose how arrival, service and delay times vary around their mean: Exponential, Deterministic, Uniform, Erlang or Lognormal. The defaults — Poisson arrivals and exponential service times — give you the classical M/M/c building block for telecom, manufacturing, and service-system models, and delays start out fixed, for transport and lead times.
Traffic Profiles
Make a Generator's arrival rate change over the run: a start rate followed by stages of a duration and a target rate, in the manner of k6's ramping arrival rate. Start from a Ramp, Step, Spike, Wave, Soak or Stress preset, preview the rate over time as you edit, and combine the profile with any distribution. Requests arrive on schedule whatever the system does.
Live Simulation
The discrete-event engine runs in a background Web Worker so the UI stays responsive. Dots travel along every connection as entities cross it — red where a full queue turns them away — and speed runs from 0.25× all the way to 16×.
Live Inline Analytics
Charts appear directly on the canvas next to each component as soon as the simulation runs: queue length, server state breakdowns (waiting / busy / blocked), entities in flight, the share each router branch receives and entity rates, all read over a window of the last 10 seconds, 30 seconds or minute.
End-to-End Cycle Time
Every entity is tracked from its generator to the exit. The whole model's completed and dropped entities, share lost, and mean, median, 95th percentile and maximum cycle time stay on screen throughout a live run.
Mid-Run Parameter Tuning
Change arrival rate, service rate, queue capacity, latency or a router's policy and weights while the simulation is live. The engine applies your updates instantly, so you can explore what-if scenarios without restarting.
Batch Runs with Confidence Intervals
Run a model up to 2,000 times at full speed and read every figure — throughput, loss, cycle time percentiles, each component's utilisation, and the arrivals and drops in each stage of a traffic profile — as a mean with a 95% confidence interval. Stop early or add ten more runs whenever you like.
Warm-Up Detection
Discard the start-up transient so results describe the system in steady state, and find how much to discard from a few pilot runs plotted with Welch's method instead of guessing.
Reproducible Runs
Every batch starts from a seed you can copy. Run the same model with the same seed and you get the same results, so two versions of a model can be compared on equal terms.
Goodness-of-Fit Testing
See the cycle time of every entity across all runs as one distribution, set it against an exponential or normal distribution of your choosing, and test the fit with Kolmogorov–Smirnov.
Model Library
Start from a prepared model, ready to run: the classic queues — M/M/1, M/M/c and M/M/1/K — a server pool calling a shared database, two paths sharing a connection pool, and that pool under a spike of orders. Each opens from the toolbar or from its own address.
Traffic Routing
Split traffic with a Router: round-robin, at random, or by a weight you set for each branch. The rule sits on the canvas as its own node, and the share each branch actually received is reported live and after a batch.
Model Save & Load
Export your model to a file and reload it in any browser session. Auto-save keeps your work safe between visits without any extra steps.
Everything above is free to use today, with no account and no limit on the size of a model — and it stays free when a paid tier arrives.
See what is coming next