Performance

Industry-leading performance

Overview

• Test suite of 20 business files ranging from 0.05 MB to 17 MB in size
• Mean file size = 3.74 MB
• File types include: PowerPoint, Video, Excel, Word, Image, PDF, Audio
• The test was run against v2.6.1 of halo

AKS

Cluster configuration assumes specific memory and compute allocations for containers. Production performance will always depend on size and complexities of real world files. Configurations can be optimised to favour throughput or file processing speeds.

For these tests the following requests and limits were set on the pods.

Small

Based on a single node (1 x DS5_v2) in a Kubernetes cluster comprising of:

• 10 Engines
• 16 virtual cores
• 56 GB Memory

Medium

Based on 5 node (5 x DS4_v2) in a Kubernetes cluster comprising of:

• 25 Engines
• 8 virtual cores per node 40 total
• 28 GB Memory per node 140 total

Large

Based on 5 nodes (5 x DS5_v2) in a Kubernetes cluster comprising of:

• 50 Engines
• 16 virtual cores per node 80 total
• 56 GB Memory per node 280 total

OVA

For these tests the following requests and limits were set on the pods.

Small

Based on a small VM (F16s_v2) running the OVA. comprising of:

• 3 Engines
• 16 virtual cores
• 32 GB Memory

Large

Based on a large VM (F32s_v2) running the OVA.

• 13 Engines
• 32 virtual cores
• 64 GB Memory

OVA vs Kubernetes Performance Differences

• The OVA uses RKE with the Kubernetes control plane also using resources on the machine.
• Multi node clusters need to use a file share for storing files. IO with this share will incur an overhead.
• While it limits the scaleability, the OVA will inherently have faster network speeds between services due to all being on the same machine.