Guest sizing · Step 8 of 9

Engineering assumptions

Leave a field empty to use the default. Values linked to a knowledge-base rule are Official only when the rule is approved; otherwise they show Validate . A project override is recorded as Customer . Administrators maintain organisation-wide defaults under Admin › Assumption defaults.

Compute & density 24 parameters

Parameter	Effective value	Evidence	Project override
Multi-session users per physical core - light	18 users/core	Validate CTX-WKL-037	users/core
Multi-session users per physical core - medium	10 users/core	Validate CTX-WKL-038	users/core
Multi-session users per physical core - heavy	6 users/core	Validate CTX-WKL-039	users/core
Single-session desktops per physical core - light (Windows 10)	6 desktops/core	Validate CTX-WKL-032	desktops/core
Single-session desktops per physical core - light (Windows 11)	5 desktops/core	Validate CTX-WKL-033	desktops/core
Single-session desktops per physical core - medium (Windows 10)	5 desktops/core	Validate CTX-WKL-034	desktops/core
Single-session desktops per physical core - medium (Windows 11)	4 desktops/core	Validate CTX-WKL-035	desktops/core
Single-session desktops per physical core - heavy (Windows 10/11)	2 desktops/core	Validate CTX-WKL-036	desktops/core
vCPU-to-physical-core ratio - multi-session (Virtual Apps)	2 :1	Validate CTX-WKL-003	:1
vCPU-to-physical-core ratio - single-session (only for the "vCPU ratio" CPU method) Engineering assumption. The VDI Handbook density method implies its own ratio per workload; this value applies only when a pool uses the vCPU-ratio CPU method.	8 :1	Assumption	:1
vCPU-to-physical-core ratio - GPU workloads	2 :1	Assumption	:1
vCPU-to-physical-core ratio - infrastructure VMs	2 :1	Assumption	:1
Windows 10/11 multi-session density reduction vs Windows Server - task (light) workers	19 %	Validate CTX-WKL-040	%
Windows 10/11 multi-session density reduction vs Windows Server - knowledge (medium/heavy) workers	32 %	Validate CTX-WKL-040	%
Density reduction when Microsoft Teams runs in multi-session sessions Citrix test result (WS2019, VDA 2106, knowledge worker). Treat as an indicative planning factor.	40 %	Validate CTX-WKL-081	%
Density reduction when Session Recording records all sessions	5 %	Validate CTX-INF-075	%
Density factor - antivirus not optimized (AV)	0.25 factor	Validate CTX-WKL-041	factor
Density factor - real-time monitoring agent (Mon)	0.15 factor	Validate CTX-WKL-041	factor
Density factor - Microsoft Office 365 (Off)	0.25 factor	Validate CTX-WKL-041	factor
Density factor - hyper-threading (HT, subtracted) The VDI Handbook formula subtracts HT while its table labels HT as a density decrease. The sign is ambiguous on the source page; leave HT adjustment off unless validated.	0.2 factor	Validate CTX-WKL-041	factor
VDA spare capacity (powered-on buffer) Informed by the Autoscale default capacity buffer (10%). Used here as a sizing buffer, which is an engineering decision.	10 %	Assumption CTX-WKL-095	%
Multi-session VDA vCPU per VM (default) The VDI Handbook density table uses an 8 vCPU / 32 GB VM specification.	8 vCPU	Assumption	vCPU
Maximum sessions per multi-session VDA (load management cap)	250 sessions	Validate CTX-WKL-094	sessions
Persistent desktops powered on at peak	100 %	Assumption	%

Memory 18 parameters

Parameter	Effective value	Evidence	Project override
Multi-session vRAM per user - light (WS2022)	256 MB	Validate CTX-WKL-011	MB
Multi-session vRAM per user - medium (WS2022)	640 MB	Validate CTX-WKL-012	MB
Multi-session vRAM per user - heavy (WS2022)	1024 MB	Validate CTX-WKL-013	MB
Multi-session VDA base memory (OS + agents) per VM	4 GB	Assumption	GB
Single-session vRAM per desktop - light (Win10/11)	4 GB	Validate CTX-WKL-008	GB
Single-session vRAM per desktop - medium (Win10/11)	8 GB	Validate CTX-WKL-009	GB
Single-session vRAM per desktop - heavy (Win10/11)	16 GB	Validate CTX-WKL-010	GB
MCS/PVS RAM cache per desktop - light (for scale)	128 MB	Validate CTX-WKL-020	MB
MCS/PVS RAM cache per desktop - light (for experience)	256 MB	Validate CTX-WKL-020	MB
MCS/PVS RAM cache per desktop - medium (for scale)	256 MB	Validate CTX-WKL-021	MB
MCS/PVS RAM cache per desktop - medium (for experience)	512 MB	Validate CTX-WKL-021	MB
MCS/PVS RAM cache per desktop - heavy (for scale)	512 MB	Validate CTX-WKL-022	MB
MCS/PVS RAM cache per desktop - heavy (for experience)	1024 MB	Validate CTX-WKL-022	MB
MCS/PVS RAM cache per multi-session VM - light	4 GB	Validate CTX-WKL-023	GB
MCS/PVS RAM cache per multi-session VM - medium	8 GB	Validate CTX-WKL-024	GB
MCS/PVS RAM cache per multi-session VM - heavy	10 GB	Validate CTX-WKL-025	GB
Single-session RAM cache sizing column	scale mode	Assumption
Additional per-user memory with Microsoft Teams (multi-session)	20 %	Validate CTX-WKL-082	%

Storage capacity 23 parameters

Parameter	Effective value	Evidence	Project override
MCS differencing / PVS write-cache disk per desktop - light	10 GB	Validate CTX-WKL-014	GB
MCS differencing / PVS write-cache disk per desktop - medium	15 GB	Validate CTX-WKL-015	GB
MCS differencing / PVS write-cache disk per desktop - heavy	20 GB	Validate CTX-WKL-016	GB
MCS differencing / PVS write-cache disk per multi-session VM - light	20 GB	Validate CTX-WKL-017	GB
MCS differencing / PVS write-cache disk per multi-session VM - medium	60 GB	Validate CTX-WKL-018	GB
MCS differencing / PVS write-cache disk per multi-session VM - heavy	60 GB	Validate CTX-WKL-019	GB
MCS identity disk per VM	16 MB	Validate CTX-WKL-057	MB
PVS vDisk store allowance over image size	20 %	Validate CTX-WKL-063	%
Image versions retained per master image (MCS snapshots / PVS versions)	3 versions	Assumption	versions
User profile size per named user (default if not measured)	2 GB	Assumption	GB
Profile Management container default capacity (reference limit)	50 GB	Validate CTX-WKL-071	GB
Additional per-user storage with Microsoft Teams	1024 MB	Validate CTX-WKL-082	MB
User data / home drive per named user (default if not provided)	5 GB	Assumption	GB
OS/system disk per infrastructure VM	100 GB	Assumption	GB
Logs and monitoring data allowance	200 GB	Assumption	GB
SQL transaction log, tempdb and local backup allowance over database size	100 %	Assumption	%
Configuration Logging database allowance (MCS environments) Citrix states 30-40 MB for non-MCS environments and that MCS environments can easily exceed this (CTX-INF-053). The MCS allowance is an engineering assumption.	2 GB	Assumption	GB
Storage growth allowance (beyond user growth)	10 %	Assumption	%
Free-space reserve kept on datastores / storage pool	20 %	Assumption	%
Data reduction ratio claimed (deduplication/compression) 1.0 means no data reduction is assumed. Only raise this with platform evidence.	1 :1	Assumption	:1
Backup target capacity multiplier over protected data	1.5 x	Assumption	x
Three-tier: one hot spare per N drives	24 drives	Assumption	drives
Three-tier: vendor-managed protection overhead (raw/usable)	1.25 x	Assumption	x

Storage performance 22 parameters

Parameter	Effective value	Evidence	Project override
Steady-state IOPS per desktop user - light, with RAM cache	1 IOPS	Validate CTX-WKL-026	IOPS
Steady-state IOPS per desktop user - light, without RAM cache	12 IOPS	Validate CTX-WKL-026	IOPS
Steady-state IOPS per desktop user - medium, with RAM cache	1 IOPS	Validate CTX-WKL-027	IOPS
Steady-state IOPS per desktop user - medium, without RAM cache	20 IOPS	Validate CTX-WKL-027	IOPS
Steady-state IOPS per desktop user - heavy, with RAM cache	3 IOPS	Validate CTX-WKL-028	IOPS
Steady-state IOPS per desktop user - heavy, without RAM cache	35 IOPS	Validate CTX-WKL-028	IOPS
Steady-state IOPS per multi-session user - light, with RAM cache	0.5 IOPS	Validate CTX-WKL-029	IOPS
Steady-state IOPS per multi-session user - light, without RAM cache	4 IOPS	Validate CTX-WKL-029	IOPS
Steady-state IOPS per multi-session user - medium, with RAM cache	1 IOPS	Validate CTX-WKL-030	IOPS
Steady-state IOPS per multi-session user - medium, without RAM cache	6 IOPS	Validate CTX-WKL-030	IOPS
Steady-state IOPS per multi-session user - heavy, with RAM cache	1 IOPS	Validate CTX-WKL-031	IOPS
Steady-state IOPS per multi-session user - heavy, without RAM cache	8 IOPS	Validate CTX-WKL-031	IOPS
Steady-state read percentage	20 %	Assumption	%
Boot/logon storm read percentage	80 %	Assumption	%
IOPS per VM while booting No official Citrix figure exists (recorded as not found). Replace with measured values when available.	100 IOPS	Assumption	IOPS
Additional IOPS per session during logon No official Citrix figure exists (recorded as not found). Replace with measured values when available.	50 IOPS	Assumption	IOPS
Average I/O size	16 KB	Assumption	KB
Storage performance headroom	30 %	Assumption	%
Three-tier estimate: IOPS per NVMe drive	100000 IOPS	Assumption	IOPS
Three-tier estimate: IOPS per SAS/SATA SSD	20000 IOPS	Assumption	IOPS
Three-tier estimate: IOPS per 10K HDD	140 IOPS	Assumption	IOPS
Three-tier estimate: IOPS per 7.2K HDD	80 IOPS	Assumption	IOPS

Network 18 parameters

Parameter	Effective value	Evidence	Project override
HDX bandwidth per session - Office workload (Citrix measured average)	22.14 kbps	Validate CTX-WKL-074	kbps
HDX bandwidth per session - rich web browsing (Citrix measured average)	478.38 kbps	Validate CTX-WKL-075	kbps
HDX bandwidth per session - 1080p video without BCR (Citrix measured average)	1663.47 kbps	Validate CTX-WKL-077	kbps
Mixed bandwidth profile - share of Office activity	70 %	Assumption	%
Mixed bandwidth profile - share of web activity	25 %	Assumption	%
Mixed bandwidth profile - share of video activity	5 %	Assumption	%
HDX bandwidth per GPU (3D) session No official per-session figure for 3D workloads was found; measure in a proof of concept.	5000 kbps	Assumption	kbps
Average user activity ratio (for average bandwidth)	60 %	Validate CTX-WKL-005	%
Teams optimized audio per user (each way)	90 kbps	Validate CTX-WKL-078	kbps
Teams optimized video per user (each way)	700 kbps	Validate CTX-WKL-079	kbps
Users in Teams meetings concurrently at peak	20 %	Assumption	%
Steady profile/user-data (SMB) traffic per session	100 kbps	Assumption	kbps
PVS streaming bandwidth per 500 target devices	1 Gbps	Validate CTX-WKL-050	Gbps
Management traffic allowance per host	1 Gbps	Assumption	Gbps
Live-migration (vMotion) burst allowance per host	10 Gbps	Assumption	Gbps
ToR downlink-to-uplink oversubscription target	3 :1	Assumption	:1
ToR uplink port speed	100 Gbps	Assumption	Gbps
Inter-switch (peer) link ports per ToR switch	2 ports	Assumption	ports

Citrix infrastructure services 38 parameters

Parameter	Effective value	Evidence	Project override
Active sessions per Delivery Controller (count formula divisor)	10000 sessions	Validate CTX-INF-008	sessions
Minimum Delivery Controllers per zone	2 VMs	Validate CTX-INF-005	VMs
Delivery Controller vCPU (Citrix reference configuration)	4 vCPU	Validate CTX-INF-006	vCPU
Delivery Controller RAM (Citrix reference configuration)	8 GB	Validate CTX-INF-007	GB
Local Host Cache additional cores per Controller (starting point)	2 vCPU	Validate CTX-INF-016	vCPU
Local Host Cache additional RAM per Controller (starting point)	2 GB	Validate CTX-INF-016	GB
StoreFront vCPU (recommended)	4 vCPU	Validate CTX-INF-026	vCPU
StoreFront base RAM	4 GB	Validate CTX-INF-027	GB
StoreFront RAM per available resource per user	700 bytes	Validate CTX-INF-027	bytes
StoreFront minimum RAM (VDI Handbook starting point)	8 GB	Validate CTX-INF-033	GB
StoreFront servers per server group (minimum)	2 VMs	Validate CTX-INF-030	VMs
StoreFront servers per server group (maximum recommended)	6 VMs	Validate CTX-INF-031	VMs
StoreFront CPU uplift with FAS	30 %	Validate CTX-INF-066	%
License Server RAM (minimum)	8 GB	Validate CTX-INF-035	GB
License Server vCPU No official CPU requirement was found (recorded as not found).	2 vCPU	Assumption	vCPU
Director server RAM Citrix minimum is 2 GB RAM (CTX-INF-023); 4 GB is an engineering allowance.	4 GB	Assumption	GB
Director server vCPU	2 vCPU	Assumption	vCPU
Director servers	2 VMs	Assumption	VMs
Monitoring data retention (sessions)	90 days	Validate CTX-INF-024	days
FAS logons per server per hour (4 x 8 vCPU servers for 100,000 users/hour)	25000 logons/h	Validate CTX-INF-064	logons/h
FAS server vCPU	8 vCPU	Validate CTX-INF-064	vCPU
FAS server RAM	8 GB	Validate CTX-INF-064	GB
WEM agents per infrastructure server	1000 agents	Validate CTX-INF-070	agents
WEM users per infrastructure server	3000 users	Validate CTX-INF-070	users
WEM infrastructure server vCPU	4 vCPU	Validate CTX-INF-069	vCPU
WEM infrastructure server RAM	8 GB	Validate CTX-INF-069	GB
WEM infrastructure server disk	80 GB	Validate CTX-INF-069	GB
PVS target devices per server (HA planning maximum)	2000 targets	Validate CTX-WKL-062	targets
Session Recording server RAM	8 GB	Validate CTX-INF-072	GB
Session Recording server processing throughput	5 Mbps	Validate CTX-INF-073	Mbps
Average recording data rate per recorded session	20 kbps	Assumption	kbps
NetScaler VPX vCPU (validate against VPX data sheet)	4 vCPU	Assumption	vCPU
NetScaler VPX RAM (validate against VPX data sheet)	8 GB	Assumption	GB
Profile/user-data file server vCPU	4 vCPU	Assumption	vCPU
Profile/user-data file server RAM	16 GB	Assumption	GB
Hypervisor management appliance vCPU (validate with vendor sizing)	4 vCPU	Assumption	vCPU
Hypervisor management appliance RAM (validate with vendor sizing)	24 GB	Assumption	GB
Hypervisor management appliance disk (validate with vendor sizing)	600 GB	Assumption	GB

Hosts & platform 6 parameters

Parameter	Effective value	Evidence	Project override
Host CPU utilization ceiling at peak	85 %	Assumption	%
Host memory utilization ceiling	90 %	Assumption	%
Physical cores reserved per host for the hypervisor	1 cores	Assumption	cores
Hypervisor memory overhead per host	8 GB	Assumption	GB
Memory channels per CPU socket (balanced DIMM population) Depends on the CPU generation. Validate against the server vendor memory population guide.	8 channels	Assumption	channels
Maximum DIMMs per channel	2 DPC	Assumption	DPC

Scenarios. Every calculation also runs these profiles for comparison:

Conservative: Risk-averse capacity: one additional host of failure tolerance, lower host utilization ceilings and more free storage.
Balanced: Project assumptions as entered.
Performance-oriented: User-experience focus: 25% fewer users per core, RAM cache sized "for experience", 25% more memory per user and more storage performance headroom.