Server Hardware and Virtualisation
Beeks Analytics supports deployment on virtual machines, provided certain conditions are met:
The virtual CPU architecture matches the supported range of Beeks-certified physical servers
Adequate CPU, memory, and storage resources are allocated
The platform delivers sufficient performance for compute, memory, and I/O workloads
The VMX-Analysis component, including the central analysis server and databases, runs reliably on virtualised platforms.
However, live packet capture and hardware timestamping present significant challenges in virtual environments—even on cloud-hosted bare-metal instances. For these scenarios, we strongly recommend consulting with Beeks Analytics for deployment guidance.
A common and effective deployment pattern is to use physical VMX-Capture appliances on-premise, paired with virtualised VMX-Analysis servers.
Distributed Architecture and Data Stores
As mentioned in Beeks Analytics Architecture, for the most scalable solution, a separate dedicated VMX-Analysis appliance can be deployed. This separate VMX-Analysis appliance can perform the following functions:
In a centralised model, a single VMX-Analysis instance can ingest and correlate data from multiple VMX-Capture appliances—whether local or remote.
In a decentralised model, VMX-Analysis can be co-located with VMX-Capture on an all-in-one server, running VMX-Explorer dashboards against local data.
In combined deployments, ensure sufficient resources are allocated to support both VMX-Analysis and VMX-Capture without contention.
From Beeks Analytics version 13 onwards, the Beeks Analytics deployment includes an optional deployment of QuestDB which can be used for statistics storage and to support VMX-Prism packet capture analysis. A commercial licence (available separately) enables advanced features such as clustered databases and more fully featured access management.
QuestDB provides a performant, open data store aligned with Beeks Analytics’ goals around scalable, real-time observability.
Server Chassis Variants
Beeks Analytics appliances are usually built on Dell PowerEdge servers.
The current generation servers are based on 1U and 2U, single and dual CPU slot chassis as follows.
All of these servers support PERC hardware NVMe RAID. The PERC hardware RAID controller is supplied as standard with the chassis models listed.
Whilst NVMe drives offer the best possible performance, there are limitations of capacity on a single drive and then issues with resiliency when losing any drive, hence RAID is a good option with NVMe where possible.
Software/chipset RAID has proven to be poor performance other than for RAID0/1. The latest hardware RAID on PCIE Gen 5 unlocks more of the underlying performance of the NVMe drives.
SAS or SATA SSD are supported for lower throughput systems.
The BOSS Boot Drive is 500GB RAID1 on all of the chassis variants listed.
Name | Dell Model | Rack U | RAID Controllers | CPU Slots | FH PCI Slots | LP PCI Slots |
|---|---|---|---|---|---|---|
BA11 | R6615 | 1 | 2 x E3.S | 1 | 2 | 0 |
BA12 | R6625 | 1 | 2 x E3.S | 2 | 2 | 0 |
BA21 | R7615 | 2 | 2 x E3.S | 1 | 4 | 2 |
BA22 | R7625 | 2 | 2 x E3.S | 2 | 4 | 2 |
Appliance Unit
On appliances with two CPU sockets and two RAID controllers, the hardware can be logically divided into two Appliance Units (AUs)—each functioning independently.
A typical all-in-one configuration dedicates:
One CPU socket and one RAID controller to VMX-Capture
The other CPU socket and RAID controller to VMX-Analysis
The term Appliance Unit refers to a logical subdivision of a physical server that is independently managed and resourced.
Two appliance units can also perform the same function - for example, you might have two Appliance Units on the same physical server with both AUs running VMX-Capture. This could allow you to achieve multiples of the capture performance of a single AU. In this case, each Appliance Unit can be thought of as 16 CPU cores and 64GB of RAM. This should be considered the minimum size of an Appliance Unit when subdividing a single physical system in this way.
CPU Architecture
Beeks Analytics currently utilise AMD EPYC4 CPU architecture, but EPYC2, EPYC3 are fully supported and were deployed on previous generations.
Beeks Analytics is also supported on Intel 2nd, 3rd and 4th Generation Xeon Scalable CPUs.
The above guidance does not imply that Beeks can support these older processor types in the absence of a compatible operating system. See Operating System . If an older CPU architecture is deployed, it is the client’s responsibility to ensure that it is compatible with an operating system which is supported by Beeks.
Compute Variants
Beeks Analytics appliances are available in standard compute tiers ranging from S to XXL:
Name | Cores | RAM | Max AUs Supported | Chassis Compatibility |
|---|---|---|---|---|
S | 16 | 64GB | 1 | |
M | 32 | 128GB | 2 | |
L | 64 | 256GB | 4 | |
XL | 128 (2x64) | 512GB | 8 | Not compatible with the BA11 or BA21 chassis |
XXL | 192 (2x96) | 1024GB | 12 | Only compatible with the BA22 chassis |
Beeks can consider other CPU variants e.g. 96 cores or 2x128 cores.
Tiers are optimised for cost-performance, with CPU selection based on best available clock speed within each core count bracket. In some high-end cases, 96 cores may offer better price-performance than 128 cores, depending on the workload.
Higher core counts benefit multi-tenant deployments and parallel capture pipelines.
Higher clock speeds are generally preferred for pure capture roles when there is a trade-off.
Beeks Analytics appliances built on AMD EPYC4 processors offer excellent performance, even in single-socket chassis. This enables high-throughput deployments in compact 1U form factors such as the BA11 or BA12.