Per-Connection Ingress MPS Control

The Per-Connection Ingress MPS Control (PCIMC) feature limits to a configurable level the per-Connection ingress message rate of each connection. Correctly configured message rate controls ensure that a single Connection cannot use the majority of the resources. (No limiting is done by PCIMC for the egress message rate.)

When a connection has data in its socket receive buffer, the data is read from the socket, but is not processed unless a process request is granted. Each time a request is granted, a single message is processed and another request must be granted before the next message can be processed. If a process request is not granted, the message is discarded.

When a process request occurs, if the connection has not exceeded its reserved capacity, the request is granted. If the reserved capacity has been exceeded and the maximum capacity is not yet reached, then the request is granted only if MP shared capacity has not been exceeded.

Regardless of a process request being granted for each message, the connection reads as fast as it is capable of, one message at a time, until one of the following occurs:

There is no more data in the socket receive buffer to read
The thread is swapped out by the operating system
The batch read limit has been reached

The Per-Connection Ingress MPS Control feature:

Is always available in the system
Applies to per-Connection MPS control for Connections that are statically assigned to MP servers through configuration and do not move from one MP to another
Is applied in the Diameter Transport Function, which is used by all DSR Applications

Capacity management for this feature can be logically separated into:

Management of the ability of the MP server to process ingress Diameter messages - how the MP server's resources are distributed to configured Connections
Management of the ability of a given Connection to process ingress Diameter messages - how each Connection behaves given its configured reserved and maximum ingress MPS settings

Per Connection Capacity Management

Per-Connection Ingress MPS Control allocates a DA-MP's ingress message processing capacity among the Diameter Peer Connections that it hosts. Each Peer Connection is allocated, through user-configuration, a Reserved Ingress MPS message processing capacity and a Maximum Ingress MPS message processing capacity.

The Reserved capacity for a Connection is available for exclusive use by the Connection. The capacity between a Connection's Reserved and Maximum capacity is shared with other Connections hosted by the DA-MP.

The DA-MP reads messages arriving from a Peer Connection and attempts to process them as long as Reserved or shared ingress message capacity is available for the Connection. When no Reserved or Shared ingress message capacity is available for a Connection, the DA-MP discards the messages without generation of any response to the Peer.

Per Connection Ingress Message Coloring

In addition to enforcing ingress message rate limits on a per Connection basis, Per-Connection Ingress MPS Control colors ingress messages based on the Reserved and average ingress message rates. Message color can be used at other traffic shedding points in the DSR, such as DA-MP Overload Control.

Traffic from under-utilized Connections is marked green by Per-Connection Ingress Message Controls, while traffic from over-utilized Connections is marked yellow. Traffic discarded by PCIMC due to capacity exhaustion (per-Connection or shared) is marked red and is not considered for any subsequent processing.

The following items describe the numbered items in Figure 1:

When the Connection's average ingress MPS rate is equal to or below its configured Reserved Ingress MPS, all messages processed by the Connection are colored green.
When the Connection's average ingress MPS rate is above its configured Reserved Ingress MPS, all messages processed by the Connection are colored yellow.
Note: If the Connection's Reserved Ingress MPS is 0, all the messages processed by the Connection are colored yellow.

Per Connection Message Coloring

Per MP Server Capacity Management

MPS rates and thresholds are used to manage ingress message MPS as it relates to the MP server as a whole.

A DA-MP has two configured ingress message rate capacity limits:

Engineered Ingress MPS: Maximum ingress message rate that a DA-MP will support without overload.
This value provides a limit to the total Reserved Ingress MPS of all Diameter Connections assigned to the DA-MP. The value is displayed on the Diameter > Configuration > DA-MPs > MP Profiles GUI page for the MP Profile assigned to the DA-MP.
MP Engineered Maximum Ingress MPS: A configurable ingress MPS limit that dictates the maximum rate at which the DA-MP will attempt to process messages from all Diameter Connections.
This value may be greater than the MP Engineered Ingress MPS.

The DA MP monitors its MPS rate and limits the rate to an MP Engineered Ingress MPS value. If the MP Engineered Ingress MPS rate is exceeded, overload can occur and ingress messages are discarded (due to MP Ingress MPS Limiting and MP Congestion Controls).

Diameter Configuration for Per Connection Ingress MPS Control

Each Diameter Connection is associated with a Capacity Configuration Set that includes the following configurable elements:

Reserved Ingress MPS

Defines the capacity reserved exclusively for the Connection and not available for use by other Connections

The Reserved Ingress MPS cannot exceed the configured Maximum Ingress MPS for a given Connection. A Connection can be configured with a zero Reserved Ingress MPS value; such Connections will not reserve message processing capacity.

The Reserved Ingress MPS for a Connection cannot be used by any other Connection, regardless of the load offered to other Connections.

If the Reserved Ingress MPS Capacity is set to a non-zero value, that value times the number of Connections using that Capacity Configuration Set on a given MP server must not be allowed to exceed the MP Maximum Reserved Ingress MPS (which is equal to the MP Engineered Ingress MPS - the highest MPS rate at which the MP server can process ingress Diameter messages).

Maximum Ingress MPS

Defines the maximum rate in ingress Diameter messages per second that the Connection is allowed to process

The Maximum Ingress MPS must be greater than or equal to the Reserved Ingress MPS. Any difference between the Maximum Ingress MPS and the Reserved Ingress MPS represents MP server resources that are shared among Connections that are using the same Capacity Configuration Set.

The configured Maximum Ingress MPS of a Connection cannot exceed the Engineered Ingress MPS of the Connection (the ingress MPS that a Connection can process at a sustained rate without errors). If the Connection has Reserved Ingress MPS, the configured Maximum Ingress MPS must be greater than or equal to the Reserved Ingress MPS. All Connections must have a non-zero configured Maximum Ingress MPS; otherwise they would not be allowed to process traffic at all. (The Maximum Ingress MPS value in the Default Capacity Configuration Set is non-zero.)

The sum of the Maximum Ingress MPS configured for all Connections on the MP server can exceed the MP Engineered Ingress MPS - the highest MPS rate at which the MP server can process ingress Diameter messages.

Ingress MPS Minor Alarm Threshold

Defines the percent of the Connection's Maximum Ingress MPS at which a Minor alarm is triggered

The Ingress MPS Minor Alarm Threshold value must be less than the Ingress MPS Major Alarm Threshold value

Ingress MPS Major Alarm Threshold

Defines the percent of the Connection's Maximum Ingress MPS at which a Major alarm is triggered.

The Ingress MPS Major Alarm Threshold must be greater than the Ingress MPS Minor Alarm Threshold.

Reserved Ingress MPS Abatement Time

Defines the minimum time (in ms) that a Connection's ingress message rate must remain less than or equal to Reserved Ingress MPS, after exceeding Reserved Ingress MPS, in order to revert the ingress traffic color from Yellow to Green.

A Default Capacity Configuration Set is provided; additional Capacity Configuration Sets can be configured. The Default Capacity Configuration Set is used for a Connection if no other Capacity Configuration Set is assigned to the Connection. The elements of the Default Capacity Configuration Set have the following default values:

Reserved Ingress MPS - zero MPS
Maximum Ingress MPS - value equal to the Engineered Ingress MPS for the Connection
Ingress MPS Minor Alarm Threshold - 50% of the configured Maximum Ingress MPS
Ingress MPS Major Alarm Threshold - 80% of the configured Maximum Ingress MPS
Reserved Ingress MPS Abatement Time - 2000 ms

Maintenance and Monitoring for Per Connection Ingress MPS Control

The Per Connection Ingress MPS Control feature provides the following maintenance and monitoring information:

Alarms and measurements to assist the network operator to detect and avoid possible capacity issues related to messaging rates
The ability to view the average ingress Diameter MPS for each Connection

The Per Connection Ingress MPS Control feature uses the following GUI information:

The Diameter > Configuration > Connections GUI page specifies which Capacity Configuration Set the Connection uses.
The Diameter > Configuration > Configuration Sets > Capacity Configuration Sets GUI pages provide elements for configuring Capacity Configuration Sets.
The Diameter > Maintenance > Connections GUI page reports KPI 10500 for average ingress MPS for each Diameter Connection (Ingress Msgs Per Second).
For each Connection, the MP server maintains the average number of ingress Diameter messages per second read from the socket. This is the rate at which ingress Diameter messages are read from the socket, not the rate at which ingress Diameter messages arrive at the socket. There is no efficient means to know the rate at which messages actually arrive.

The average ingress message rate is a running average (approximately 30 seconds). This provides a good picture of the level of ingress messages being read by each Connection while avoiding erratic readings caused by short duration spikes in the rate.

Connection Alarm

The Per Connection Ingress MPS Control feature provides a Connection alarm with two severities to alert the network operator when the average ingress MPS rate goes above the configured thresholds for percentage of the configured maximum ingress MPS for the Connection.

The Connection Ingress MPS Alarm is a per Connection alarm that can be configured in a Connection's Capacity Configuration Set to trigger at a Minor and Major capacity threshold.

The Minor alarm is asserted when the 30-second average MPS rate exceeds the configured Ingress MPS Minor Alarm Threshold value for the Connection. The Minor alarm is cleared when the 30-second average MPS rate falls 5% below the Ingress MPS Minor Alarm Threshold value configured for the Connection.

The Major alarm is asserted when the 30-second average MPS rate exceeds the Ingress MPS Major Alarm Threshold value configured for the Connection. The Major alarm is converted to a Minor alarm when the 30-second average MPS rate falls 5% below the Ingress MPS Major Alarm Threshold value configured for the Connection.

An alarm cannot be abated until an abatement time delay has expired. For example, if a Minor alarm is asserted, the alarm cannot be cleared until the abatement time delay has expired and the average ingress MPS for the Connection is 5% below the Minor alarm assert percentage. The abatement time delay is 15 seconds.

The alarm abatement time delay affects only clearing of alarms, not asserting of alarms. Therefore, it is possible to transition rapidly from a Minor alarm to a Major alarm.

Limitations

PCIMC relies on a configured relationship between a Connection and the MP server that will host the Connection.
PCIMC does not prevent the total Reserved Ingress MPS of all Connections hosted by a DA MP from exceeding the MP Maximum Reserved MPS capacity when the MP Maximum Reserved MPS capacity is reduced after Connections are already configured with non-zero Reserved Ingress MPS.