Performance & Load Estimation

On large archive workloads, roughly 2x faster. In benchmark testing, MetaDefender Cluster processed a large CAB dataset at 2,120,000 objects per hour in 175 minutes, compared to 1,050,000 objects per hour in 353 minutes for a load-balanced deployment — about half the processing time. The advantage comes from distributing archive contents, so on non-archive files the two perform nearly the same.

Archive load sharing splits the nested contents of a single archive across multiple MetaDefender Core™ instances, so one large archive is extracted and scanned in parallel instead of being pinned to a single instance. It is the mechanism behind the largest measured speed gains — a load balancer routes each file whole to one instance, leaving the rest of the pool idle.

Throughput rises and processing time drops close to linearly. In benchmark testing on a non-archive dataset, scaling from 4 to 12 MetaDefender Core instances raised throughput from 1,000,000 to 2,850,000 objects per hour and cut the run from 120 minutes to 42 — roughly 2.9x the throughput in about a third of the time. Archive-heavy workloads gain more.

It depends on your file volume, the share of archives in your traffic, and your throughput target. Because archive contents are distributed across instances, archive-heavy workloads benefit from more instances than document-only workloads at the same file count. Share your volumes, archive mix, and throughput targets with OPSWAT for a load estimate sized to your workload.

In benchmark testing, MetaDefender Cluster sustained roughly 1.9 to 2.1 million processed objects per hour in the load-balancer comparison scenarios, and up to 4,000,000 objects per hour on a large-archive workload scaled to 12 MetaDefender Core instances. Actual throughput depends on file mix, archive depth, the number of instances, and the engines enabled.

Only marginally. In benchmark testing on 2,000,000 non-archive files, MetaDefender Cluster processed 1,928,000 objects per hour versus 1,913,000 for a load-balanced deployment — a difference under one percent, because there are no archive contents to distribute. MetaDefender Cluster's advantage appears on archive-heavy workloads and when scaling instance count.

MetaDefender Cluster reroutes that instance's workload to the remaining healthy instances, so an individual failure reduces available capacity rather than halting scanning. Incoming files are balanced across all healthy MetaDefender Core instances continuously, so work never queues behind an instance that is offline or idle.