Size buckets & the distribution flag

Every device is generated from a model (a size bucket). The --distribution flag sets how many devices fall into each bucket, so you can shape a fleet from “mostly small access switches” to “all pathological cores”.

The distribution syntax

--distribution takes comma-separated model:weight pairs, where weights are percentages that should sum to 100:

--distribution "sm:40,md:40,lg:15,xl:5"   # the default
--distribution "xl:50,2xl:30,3xl:20"      # a heavy, large-config fleet
--distribution "ciena-6500-tl1:100"       # an all-Ciena optical fleet

The default sm:40,md:40,lg:15,xl:5 approximates a typical enterprise fleet.

Cisco IOS size buckets

These nine buckets all use the cisco_ios driver. Sizes below are approximate; exact bytes depend on the seed. The sm–xl buckets have distinct templates; 2xl–6xl share the xl feature set and scale up the counts.

Bucket	Approx size	Class	Profile highlights
`sm`	~30 KB	Access switch	48 interfaces, basic VLANs/ACLs, no routing
`md`	~150 KB	Aggregation / small router	48 ifaces + 60 sub-ifaces, OSPF, crypto, QoS
`lg`	~700 KB	Core router	96 ifaces, BGP (10 neighbours), 3 OSPF areas, VRFs
`xl`	~3–5 MB	DC core / firewall	192 ifaces, 60 ACLs, 20 BGP, 30 VRFs, big ACLs
`2xl`	~8 MB	Hyperscale edge	256 ifaces, 100 ACLs, 30 BGP, 50 VRFs
`3xl`	~16 MB	Regional spine	384 ifaces, 160 ACLs, 40 BGP, 70 VRFs
`4xl`	~32 MB	Service-provider core	512 ifaces, 240 ACLs, 60 BGP, 100 VRFs
`5xl`	~64 MB	Pathological	768 ifaces, 380 ACLs, 100 BGP, 150 VRFs
`6xl`	~128 MB	Maximum stress	1024 ifaces, 600 ACLs, 160 BGP, 220 VRFs

The larger tiers exist to stress diff engines, parsers, and storage with configs far beyond what most tools are tested against.

Ciena bucket

Bucket	Approx size	Class	Driver
`ciena-6500-tl1`	~1 KB	Ciena 6500 7-slot optical	`ciena_tl1`

This model produces a TL1 shelf inventory served by the Ciena TL1 driver, not a Cisco config. Mix it into a Cisco fleet to test multi-vendor handling:

--distribution "sm:35,md:35,lg:20,ciena-6500-tl1:10"

Sizing disk

Disk usage is dominated by the large buckets. A default 50k fleet (40/40/15/5) is modest, but a fleet skewed to 5xl/6xl can need hundreds of GB. Plan capacity from your distribution — see Prerequisites.