Data collection

Objective

• In this article, we will look into Data collection.
• We will look at the terminology used by Accedian to collect data.
• We will have some screen captures showing the key performance indicator (KPI) from the Nids, Nodes and Sensor control.
• We will look into the Skylight orchestrator configuration to send data to Skylight Analytics.
• We will see where the data came from.
• We will briefly look at SNMP polling and SNMP traps.

Where is the data coming from

This section covers the manner Accedian and non-Accedian devices collect the Key performance indicator (KPI) and send it to Skylight Analytics.

We need to understand the terminology between Skylight orchestrator and Skylight analytics to see where data comes from. We will review the following diagram and several screen captures.

Skylight Orchestrator

• A Network Element is a device that generates performance metrics.
• A Collection Matrix display the session of a device that is poll for data.
• Statistics have the Key performance indicators (KPIs).

Skylight Analytics

• A Monitor session is a device that generates performance metrics.
• A Session is a device collecting metrics that will be poll for data.
• Performance Metrics have the Key performance indicators (KPIs).

Accedian Skylight orchestrator collects performance monitoring data from all its manage and non-manage devices to Skylight Analytics. These metrics collections come from:

• Nids. GE,TE, GT, GX.
• Nodes. LT, LTs, LX-10Gig
• Sensors control. (aks VCX)
• Skylight modules. SFP compute and modules (aks nanos and ants)
• Other vendor applications reflecting frames or packets.

Accedian Skylight orchestrator streams the collect performance monitoring data from all its manage devices to Skylight Analytics in a CSV file format.

For other non-Accedian types of devices such as Cisco, Ciena, Huawei, etc, the Simple Network Management Protocol (SNMP) will provide discovery of sessions and devices.
Skylight Analytics will use rules based on discovery to determine which sessions, devices, and interfaces to visualize.
The Roadrunner instance will do the polling and send the files to Analytics.

Capture data

• Packet capture is a networking term for intercepting a data packet crossing a specific point in a data network.
• Once a packet is captured in real-time, it is stored for a period of time to be analyzed and then either be downloaded, archived or discarded.
• Packet sniffers work by intercepting and logging network traffic via the wired or wireless network interface on a host machine.
• Once the raw packet data is captured, Skylight Analytics analyzes it and presents it in a readable form so that the person using Analytics can visualize it.

Sessions samples

The following are some examples of the data collected from the Accedian devices via the Skylight orchestrator.

TWAMP protocol session

The Skylight orchestrator TWAMP sample session from a Sensor control

The information circled in this screen capture, presents the actual values that will be streamed to Skylight Analytics.

Connectivity Fault Management (CFM) session

Accedian Skylight Element LT. CFM sample session

Accedian Skylight Element LT CFM sample session MEP statistics.

Accedian Skylight Element LT CFM sample session DMM statistics.

Accedian Skylight Element LT CFM sample Packet loss

TWAMP protocol session

Accedian Skylight Element LT. TWAMP sample session.

The information underlined in this screen capture, presents the actual values that will be streamed to Skylight Analytics.

The KPI are outlined in red

Performance Assurance Agent (PAA) session

Accedian Skylight Element LT. Performance Assurance Agent (PAA) sample session.
PAA offers the capability to measure QoS performance metrics at the IP and Ethernet layers.

*The KPI are the columns boxed in gold. *

Skylight orchestrator setup to stream csv files to Analytics

• Skylight orchestrator send PM data (CSV files) to Skylight analytics
• The Metrics Collection option must be enabled on Skylight orchestrator
• Go to the Collection > Server Configuration tab:

SNMP Trap & Polling

SNMP is based on a command center and agent model.
You will have a command center, and at the various remote locations, you will have agents.
The agents job is to collect all kinds of information, even information about themselves like system alarms and report all this information to the command center.
They talk between them via SNMP trap messages; the way it works is whenever something happens, the agent immediately sends a trap to the command center.
The problem with this method is that if an agent does not send a trap, there is no way for the command center to know if the agent is still alive.

Another way used to talk between them is by SNMP polling.
The polling loop allows the command center to send a request to the agents at a regular interval. Therefore, if an agent is offline, you will not get a message back.

Which is the best method?
You need to decide based on your network and your operation. The best decision is to choose a hybrid system like Skylight Analytic that supports SNMP trap and polling; you will be getting data from the best of both worlds.