3.3.2 Profile Resource Requirements for Synthetic Feed
The following table displays the profile resource requirements when Synthetic FEED is "100K" MPS.
Note:
Set the value of the parameter TCP_CONN_INSPECTOR_FIXED_DELAY_MS to "60000" in the custom-value.yaml file.
Table 3-6 Resource Requirements
| OCNADD Service | 15K NRF Profile | 55K SCP Profile | 30K SEPP Profile | 100K Profile (SCP:55K NRF:15K SEPP:30K) | 100K SCP Profile | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
Replication Factor = 1 or 2 (when SSD is enabled) Message Size = 3500 FEED Type = SYNTHETIC |
vCPU | Total Replica |
Memory Required (Gi) |
Topic Partitions | vCPU | Total Replica |
Memory Required (Gi) |
Topic Partitions | vCPU | Total Replica |
Memory Required (Gi) |
Topic Partitions | vCPU | Total Replica |
Memory Required (Gi) |
Topic Partitions | vCPU | Total Replica |
Memory Required (Gi) |
Topic Partitions |
| OCNADD microservices resource requirements: | ||||||||||||||||||||
| ocnaddconfiguration | 1 | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - |
| ocnaddalarm | 1 | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - |
| ocnaddadmin | 1 | 1 | - | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - |
| ocnaddhealthmonitoring | 1 | 1 | 1 | - | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - | 1 | 1 | 1 | - | |
| ocnaddgui | 2 | 1 | 1 | - | 2 | 1 | 1 | - | 2 | 1 | 1 | - | 2 | 1 | 1 | - | 2 | 1 | 1 | - |
| ocnadduirouter | 2 | 1 | 1 | - | 2 | 1 | 1 | - | 2 | 1 | 1 | - | 2 | 1 | 1 | - | 2 | 1 | 1 | - |
| zookeeper | 1 | 3 | 1 | - | 1 | 3 | 1 | - | 1 | 3 | 1 | - | 1 | 3 | 2 | - | 1 | 3 | 1 | - |
| ocnaddkafka | 2 | 3 | 24 | - | 3 | 3 | 24 | - | 3 | 3 | 24 | - | 6 | 4 | 48 | - | 6 | 4 | 48 | - |
| Replicated Feed | ||||||||||||||||||||
| ocnaddkafka | 2 | 3 | 24 | - | 4 | 3 | 48 | - | 4 | 3 | 24 | - | 10 | 4 | 96 | - | 10 | 4 | 96 | - |
| WEIGHTED-LB is ON/OFF, L3L4 Mapping is ON/OFF and FILTER is OFF | ||||||||||||||||||||
| ocnaddscpaggregation | - | - | - | - | 2 | 2 | 2 | SCP=12 | - | - | - | - | 2 | 3 | 2 | SCP=18
(Each instance 6 partitions) |
2 | 4 | 2 | SCP=24
(Each instance 6 partitions) |
| ocnaddnrfaggregation | 2 | 1 | 2 | NRF=6 | - | - | - | - | - | - | - | - | 2 | 1 | 2 | NRF=6
(Each instance 6 partitions) |
- | - | - | - |
| ocnaddseppaggregation | - | - | - | - | - | - | - | - | 2 | 2 | 2 | SEPP=12 | 2 | 2 | 2 | SEPP=12
(Each instance 6 partitions) |
- | - | - | - |
| ocnaddadapter | 3 | 1 | 3 | MAIN=6 | 3 | 3 | 3 | MAIN=18 | 3 | 2 | 3 | MAIN=12 | 3 | 9 | 4 | MAIN=54
(Each instance 6 partitions) |
3 | 9 | 4 | MAIN=54
(Each instance 6 partitions) |
| WEIGHTED-LB is ON/OFF, L3L4 Mapping is ON/OFF and FILTER is ON | ||||||||||||||||||||
| Ingress Filter is ON | ||||||||||||||||||||
| ocnaddscpaggregation | - | - | - | - | 2 | 4 | 2 | SCP=18 | - | - | - | - | 2 | 4 | 2 | SCP=24
(Each instance 6 partitions) |
2 | 5 | 2 | SCP=30
(Each instance 6 partitions) |
| ocnaddnrfaggregation | 2 | 2 | 2 | NRF=6 | - | - | - | - | - | - | - | - | 2 | 2 | 2 | NRF=12
(Each instance 6 partitions) |
- | - | - | - |
| ocnaddseppaggregation | - | - | - | - | - | - | - | - | 2 | 3 | 2 | SEPP=12 | 2 | 3 | 2 | SEPP=18
(Each instance 6 partitions) |
- | - | - | - |
| Egress Filter is ON: | ||||||||||||||||||||
| ocnaddadapter | 3 | 2 | 3 | MAIN=6 | 3 | 4 | 3 | MAIN=24 | 3 | 3 | 3 | MAIN=18 | 3 | 9 | 4 | MAIN=54
(Each instance 6 partitions) |
3 | 9 | 4 | MAIN=54
(Each instance 6 partitions) |
Note:
- When advanced features such as Ingress filtering, Egress filtering, L3-L4 and Weighted load balancing are enabled simultaneously, the resource requirement for "ocnaddadapter" service may vary at higher throughput.
- Resource requirement may vary when Filter is ON based on % data allowed after filtering and number of filter condition with values configured.
- Resource requirement may vary when L3L4 is ON and size of global l3l4 configuration is huge.
- Five or six instances of Kafka brokers might be required when running RF=2, based on the setup performance. The end-to-end latency might increase when DISK I/O is slow.
- For DISK I/O, see Disk Throughput Requirements.
- For Kafka PVC-Storage, see Kafka PVC Storage Requirements.