System Requirements

Network Connectivity

Depending on the configured options, Apstra Flow requires various TCP and UDP ports to receive flow records, retrieve data for enrichment, and store data in your chosen data platform. To allow communication on TCP and UDP ports, you must configure any host or network firewalls to allow traffic to pass through as described in the following sections:

Listening for Flow Data
Accessing Enrichment Data

Listening for Flow Data

You can configure the Apstra Flow collector to listen for incoming flow record packets on one or more UDP ports. Table 1 shows the collector's UDP default ports.

Table 1: Apstra Flow Collector Default Ports
Protocol	Port	Direction	Description
UDP	9995	in	Apstra Flow default port
UDP	2055	in	Netflow standard port
UDP	4739	in	IPFIX standard port
UDP	6343	in	sFlow standard port

While a variety of ports can be used to listen for flow record packets, the specific ports which must allowed are those for which the collector is configured using EF_FLOW_SERVER_UDP_PORT.

Accessing Enrichment Data

The Apstra Flow collector can enrich flow records with additional information. The following sections show the various enrichment options and corresponding allowed ports.

DNS
SNMP
OpenSearch
VM Sizing

DNS

Required when EF_PROCESSOR_ENRICH_IPADDR_DNS_ENABLE is true.

Table 2: DNS Allowed Port
Protocol	Port	Direction	Description
UDP	53	out	DNS

SNMP

Required when EF_PROCESSOR_ENRICH_NETIF_SNMP_ENABLE is true.

Table 3: SNMP Allowed Port
Protocol	Port	Direction	Description
UDP	161	out	Network interface attributes through SNMP.

OpenSearch

Specify one of the TCP allowed ports when EF_OUTPUT_OPENSEARCH_ENABLE is true.

Table 4: OpenSearch Allowed Ports
Protocol	Port	Direction	Description
TCP	9200	out	OpenSearch REST API.
TCP	5601	out	OpenSearch dashboards, GUI and API.

VM Sizing

We conducted tests for VM sizing using Apstra Flow OVA on ESXi 8.0 (Table 5). VM sizes and storage results are listed in Table 6 .

Note:

Other workloads were not active on the system during testing. "Noisy neighbors" or other resource contention could negatively impact the results in production environments.

Table 5: Components Tested for VM Sizing
Component	Description
CPU	AMD EPYC 7702 (64-core Zen2), locked at 2.9GHz to avoid thermal throttling.
Memory	256GB DDR4 3200MT/s, all 8 memory channels populated for maximum memory bandwidth.
Storage (Direct-SSD)	4TB SATA SSD
Storage (NFS)	8x HDD (RAID10) with NVMe read/write cache through 10Gbe

Table 6: VM Sizing and Storage Results
VM Size	CPU and Memory Sizing	Ingest Capacity
Default (Medium) VM	CPU: 16vCPUs Memory: 64 GB	Direct-SSD: Recommended ingest: 12,000 records per second. Burst ingest: up to 40,000 flow records per second. NFS Recommended ingest: 10,000 records per second. Burst ingest: up to 40,000 flow records per second.
Small VM	CPU: 8vCPUs Memory: 32 GB	Direct-SSD: Recommended ingest: 6,500 records per second. Burst ingest: Up to 21,500 flow records per second. NFS Recommended ingest: 5,500 records per second. Burst ingest: up to 21,500 flow records per second.
Large VM	CPU: 24vCPUs Memory: 64 GB	Direct-SSD: Recommended ingest: 16,000 records per second. Burst ingest: up to 53,000 flow records per second. NFS Recommended ingest: 13,000 records per second. Burst ingest: up to 53,000 flow records per second.
X-Large VM (custom)	Contact your Juniper sales representative for guidance on creating a cluster for a custom deployment.	Greater than 15,000 FPS.

Licensing

The Apstra Flow collector operates with the integration of a Juniper Apstra license. Premium license holders benefit from enhanced features and an elevated flow rate capacity. In contrast, users with the basic license have a constraint of up to 500 flows per second. For information about activating your license, see the Juniper Licensing User Guide.

ON THIS PAGE