Scale limits for functions
Concurrent session limits
NSP supports a combined concurrent session limit of 125. The following table defines any concurrent session limits of NSP functions within the global NSP wide session limit.
Table 5-6: Concurrent session limits for NSP functions
Function |
Maximum number of concurrent sessions |
---|---|
Analytics reports |
10 |
Concurrent NE sessions on NFMP managed nodes |
100 |
Concurrent NE sessions on MDM managed nodes |
100 |
Scale limits for Kafka event notifications
Kafka event notifications supports a maximum of 200 OSS subscriptions through NBI notification. Within that global limit, Alarm Management supports a maximum of 5 OSS subscriptions.
Scale limits for Telemetry
Scale Limits for MDM Telemetry (multi-vendor SNMP)
MDM Telemetry data collection is limited by the maximum OSS subscriptions supported by Kafka. The maximum number of Telemetry notifications per second is 1500 per active MDM instance, where one Telemetry record (a collection of statistics counters) update equals one Telemetry notification.
Scale Limits for Cloud Native Telemetry (gNMI)
Cloud Native Telemetry data collection is limited by the maximum OSS subscriptions supported by Kafka. The maximum number of Telemetry notifications per second is 1500 per CN gNMI Collector instance, where one Telemetry record (a collection of statistics counters) update equals one Telemetry notification.
Telemetry data collection is also limited to a maximum of 2,000 NEs per CN gNMI Collector instance.
Scale Limits for Cloud Native Telemetry (accounting file processing)
NSP processing of NE accounting files is performed by Accounting Processor instances. Multiple Accounting Processors can be deployed to gain higher throughput based on network size. Each Accounting Processor instance can support up to 2000 NEs, and can support a file output rate of 6200 records/s with database and kafka rates of 1500 records/s.
Typically local storage can support up to two Accounting Processor instances depending on the available system resources. To gain higher throughput, customer provided network storage should be used.
Maximum scale of 140 million accounting records every 15 minutes is achieved with 25 Accounting Processor instances subject to NSP system capacity.
Scale Limits for combined MDM and Cloud Native Telemetry persistence
The maximum number of rows uploaded to the database per minute is 90 000 per combined active MDM Instance and CN gNMI Collector instance, where one row equals one Telemetry record. This limit applies for Postgres and Auxiliary database storage.
If NSP is deployed with multiple active MDM and/or CN Telemetry Instances, the maximum collective upload rate to a Postgres database is 180 000 rows per minute. When Telemetry data is stored in the Auxiliary Database, the upload rate scales horizontally with more active MDM and/or CN Telemetry Instances. network activity, database activity and latency can also affect database upload rates.
Event timeline limits for managed NEs and services
Some applications make use of historical data for managed NEs and services. The amount of historical data is limited according to the mediation component and database storage.
NFM-P managed NEs and services have a default event timeline of 1 week for oracle database storage and can be configured to a maximum of 1 month. For Auxiliary Database storage the event timeline can be increased to a maximum of 1 year.
For NSP, MDM managed NEs and services have an event timeline of 1 week for Postgres database storage. Auxiliary Database storage is not supported for MDM managed NEs and services.
Scale limits for alarms
The following table defines the alarm limits for NSP:
Key dimension |
Maximum number of alarms |
---|---|
Historical alarms from non-NFM-P systems (eg. WS-NOC , MDM, NSP) |
10 million |
Active alarms from NFM-P, WS-NOC , and/or MDM-managed nodes |
500 thousand |
Note: Alarm limits describe the aggregate number of alarms that can be handled by NSP but do not supersede individual datasource limits.
The following table defines the performance limits for alarms:
Key dimension |
Rate |
---|---|
Sustained alarm rate (combined from all sources) (see Note) |
200/second |
Concurrent event notification subscriptions limit |
5 |
Note: Alarm rate describes the aggregate volume that can be handled by NSP but does not supersede individual datasource limits.
The following table defines the squelching limits for alarms:
Key dimension |
Maximum number of objects |
---|---|
Port squelching |
1000 ports |
Network element squelching |
1000 network elements |
Resource group squelching |
250 000 ports and/or network elements combined |
Note: Because the maximum size for a port group is currently 100k (100 000) ports, multiple resource groups are needed to achieve the 250k squelching limit.
Network Health Overview
Network Map
The number of NEs and links managed in the network may affect performance and topology rendering time.
Multi-layer maps support a recommended maximum of 4000 objects. Users should expect the following multi-layer map loading times with different numbers of NEs.
-
For 250 NEs (125 physical links), approximately six seconds for the initial page loading and four seconds to reload.
-
For 500 NEs (250 physical links), approximately nine seconds for the initial page loading and six seconds to reload.
-
For 2000 NEs (1000 physical links), approximately 50 seconds for the initial page loading and 28 seconds to reload.
Link Utilization Map
The Link Utilization map view has limits on the number of supported endpoints and links that can subscribe for stats simultaneously. The following table lists the recommended maximum number of links on the current operational view for different NE types. These are not absolute maximum values but safe recommended limits based on product testing.
Link Type |
Maximum |
---|---|
7750 SR physical link |
500 |
7705 SAR / 7210 SAS physical link |
200 |
7750 SR LAG link |
160 |
7705 SAR / 7210 SAS LAG link |
60 |
MD-OAM Scale and Performance
Scale limits and performance metrics are based on testing in a lab environment with an enhanced NSP deployment.
Table 5-7: Scale and performance limits - isolated per metric type
Performance Metric |
Scale Limit (records/time) |
Scale Limit (tests/time interval) |
---|---|---|
gNMI collection rate |
2.7m records/15min |
30 000 tests @ 1 metric/10s |
oam-pm accounting collection rate |
2.4m records/15min |
36 000 tests @ 22 bins, 15min measured interval |
combined CRUD rate 1 |
48 000 CRUD actions/15min |
24 000 service validation workflows/hour |
Notes:
Combined CRUD is a term used to define an aggregation of CRUD (create/update/delete) operations in the setup/execution/retrieval/deletion of service validation tests.
Table 5-8: Scale and performance limits - combined activity
Performance Metric |
Scale Limit (records/time) |
Scale Limit (tests/time interval) |
---|---|---|
gNMI collection rate + oam-pm accounting collection rate |
3.8m records/15min |
20 000 tests @ 1metric/10s + 30 000 tests @ 22bins, 15m measured interval |
gNMI collection rate + oam-pm accounting collection rate + combined CRUD rate 1 |
2.2m records/15min + 24 000 CRUD actions/15min |
12 000 tests @ 1metric/10s + 16 000 tests @ 22bins, 15m measured interval + 12 000 service validation workflows/hr |
Notes:
Combined CRUD is a term used to define an aggregation of CRUD (create/update/delete) operations in the setup/execution/retrieval/deletion of service validation tests.
The calculation of records per test type can be determined with the following table.
Table 5-9: Records per Test Result
Test Type |
Record Count |
Results class info |
---|---|---|
cfm-dmm (accounting) |
24 to 93 per measurement interval |
3: /base/oampm-accounting/cfm-dmm-session-acc-stats; 3 * # of bins (7-30): /base/oampm-accounting/cfm-dmm-bin-acc-stats |
cfm-dmm (gNMI) 1 |
1 to 6 per sample window |
1 * # of metrics: /base/oam-pm/eth-cfm-delay-streaming |
cfm-lmm (accounting) |
1 per measurement interval |
1: /base/oampm-accounting/cfm-lmm-session-acc-stats |
cfm-loopback |
1 per test execution |
1: /base/oam-result/loopback-result |
cfm-slm (accounting) |
1 per measurement interval |
1: /base/oampm-accounting/cfm-slm-session-acc-stats |
twamp-light (gNMI) 1 |
1 to 6 per sample window |
1 * # of metrics: /base/oam-pm/twamp-light-delay-streaming |
twamp-light delay (accounting) |
24 to 93 per measurement interval |
3: /base/oampm-accounting/twl-session-acc-stats; 3 * # of bins (7-30): /base/oampm-accounting/twl-bin-acc-stats |
twamp-light loss (accounting) |
1 per measurement interval |
1: /base/oampm-accounting/twl-session-loss-acc-stats |
Notes:
Metrics for gNMI collection records include: fd-average (directions forward, backward, round-trip) and ifdv-average (directions forward, backward, round-trip).
Device Management
The following limitations apply when using NSP’s Device Management views.
While the Device Management inventory tree is manually traversed, 500 objects are loaded at a time, to a maximum of 2000 objects.
If any logical tree level shows more than 500 objects, that entire logical tree level is refreshed after a manual refresh is clicked, but only the first 500 objects are displayed.
Scale limits of Map Layout and Group Directories
Nokia recommends a maximum of 2000 NEs per region for the Operational map view.
Where IP/optical coordination is deployed, the following scaling limits for Map Layout will apply:
Group directories have the following scaling limits.
-
There is no limit on the number of directories for each directory type (network element, port, service, analytics resource).
Scale limits for NSP Baseline Analytics
The NSP Baseline Analytics can support collection storage in the Postgres database or in the Auxiliary database. Baselines are supported on NFM-P and MDM managed nodes.
Key dimension |
Postgres database storage |
AuxDB storage |
---|---|---|
Number of baselines |
10 000 |
100 000 |
Retention time |
35 days |
403 days |
Collection Interval |
300 seconds |
300 seconds |
Window Duration |
15 minutes |
15 minutes |
Season |
1 week |
1 week |
Note: Reducing the Collection Interval or Window Duration will result in a reduced number of Baselines that can be supported.
Scale limits for NSP Indicators
The NSP Indicators can support collection storage in the Postgres database or in the Auxiliary database. Indicators are supported on NFM-P and MDM managed nodes.
The NSP Indicators can only support up to a total of 20 Indicator rules. The recommended maximum number of resources that can feed an Indicator rule is 2500.
Key dimension |
Postgres database storage |
AuxDB storage |
---|---|---|
Number of resources (number of incoming entities into NSP Indicators) |
10 000 |
50 000 |
Retention time |
35 days |
403 days |
Collection Interval (Complex Indicators) |
300 seconds |
300 seconds |
Collection Interval (Simple Indicators) |
900 seconds |
900 seconds |
Window Duration (Complex Indicators) |
15 minutes |
15 minutes |
Note: Reducing the Collection Interval or Window Duration will result in a reduced number of resources that can be supported.
Flow Collector scale for NAT collection
The Flow Collector BB NAT collection limit is 350,000 records/s when customer retrieves files with native s/ftp application.
Scale limits for Large Scale Operations
The Large Scale Operations feature has scaling limits for framework and for device operations.
The following table summarizes the framework limits.
Key dimension |
Maximum |
---|---|
Number of concurrent LSO executions |
20 |
Number of stored operations (historical and running) |
500 |
Number of operation types |
100 |
Number of targets per operation |
10 000 |
Number of phases per operation type |
10 |
The following table summarizes the NE device operation limits.
Key dimension |
Maximum |
---|---|
Number of nodes for NE backup |
10 000 |
Note: Numbers are based on using enhanced profile disk availability for File Service.
Note: Role Based Access Control will not apply to LSO app user operations in this release.
Scale limits for Zero Touch Provisioning
The following limits apply to Day 0 Zero Touch Provisioning (ZTP):
Key dimension |
Maximum |
---|---|
NE instances created per second |
5 |
Simultaneous downloads from file server |
10 |
ZTP instances in various provisioning states |
1000 |
Scale limits for Generic Mediator
The Generic Mediator application has the following scaling limits:
Key dimension |
Maximum |
---|---|
Concurrent threads |
10 |
Request queue size |
50 |
User Access Control Performance
In an NSP deployment with User Access Control enabled, and more than 10 user groups are defined, in large networks (> 2000 NEs), NSP GUI performance may be affected if the resource groups contain a very large number of equipment and/or service objects.