ESET PRIVATE Scanning Solution – Table of Contents

Scaling and Scanning Performance

ESET PRIVATE Static Scanning Engine scales horizontally by increasing replicas/pods/tasks to handle higher scanning throughput, with each scanner instance supporting up to 8 concurrent scanning threads (simultaneous users).

Scaling Model

Scaling depends on the number of scanner replicas to process incoming scan requests from multiple agents/users. A single scanner replica can handle up to 8 simultaneous scanning threads (concurrent users/agents). Higher throughput can be achieved by deploying additional replicas.

Performance Factors

Scanning performance may vary based on file type characteristics:

File Type

Relative Performance (depending on the deployment setup)

Comment

Small files (<1MB)

~ milliseconds

Minimal CPU/memory; highest throughput

Documents (PDF, DOCX)

~ milliseconds

Standard processing

Archives (ZIP, RAR)

~ seconds

Extraction + nested scanning overhead

APKs

~ seconds

DEX analysis + signature matching

Executables (EXE)

~ seconds-minutes

Behavioral analysis + heuristics

Large media (videos)

~ seconds-minutes

Byte-by-byte scanning;

Optimized performance is achieved by subsequent scanning of small parts of the entire file, transfer of the entire file is typically not needed, only in selected cases.

Variables affecting performance:

File entropy/complexity

Nested archive depth

Cloud reputation lookups

CPU/memory allocation

Sizing Recommendations

Due to workload variability, ESET recommends running a Proof of Concept (PoC) with representative file types and volumes:

1.Deploy single Scanner replica using on-prem/EKS/ECS guides (see dedicated sections). Typical resource allocation per component is 1vCPU, 2GB memory.

2.Run Scanner Agent with realistic workload (rotate the scanned content to keep utilizing the scanner threads up to 8 active users)

3.Monitor metrics:

a.Scanner scanning performance

b.CPU utilization \ memory usage

c.Time from agent submit -> scan complete

Single Scanner instance scans files with a reference throughput up to ~ 1GB of processed data within one minute.