03-Mentor Server Architecture

Lesson ObjectivesBy the end of this lesson you will be able to describe the main Mentor Server components and data flow By the end of this lesson you will be able to describe the main

Mentor Server

Mentor Server Optimization

Lesson Objectives

By the end of this lesson you will be able

to describe the main Mentor Server

components and data flow

By the end of this lesson you will be able

to describe the main Mentor Server

components and data flow

Main Topics

• Introduce the Mentor server and roles

• Describe the main building blocks

• Review the enterprise server architecture

• Describe the data process flow

• Discuss the client-server interface

Mentor Server

• By automating these processes, the client significantly reduces the

process effort, and shortens the optimization cycles, leaving the

engineer with more time for analysis and decision making

• The enterprise configuration allows the Mentor server to manage

medium and large networks under a single centralized system

• The enhanced load-balancing capability allows efficient usage of

system machines, and redundancy in cases of component failure

The Mentor server automates routine tasks, relieving the RF

engineer of manual tasks (such as network data collection, network

modeling, analysis, and optimization etc.)

Mentor Server – Roles

• Physical network configuration data collection

• Log, KPI, and mobile-measurement collection

• Collected data processing

• Data model creation

• Template-based scheduling of analysis and optimization tasks

• Network element and cluster definitions

Clients and Applications

• Desktop Clients

• Forte – RF optimization for GSM/GPRS/EDGE networks.

• Mentor – GSM/UMTS/HSDPA/LTE/CDMA/EVDO network

analysis and optimization

• Mentor Server Admin – server management console for

automated data collection and processing, and security settings

• Web Applications

• Insight – data-collection analytics tool.

Main Server Building Blocks

• Processing Servers

• Front-End (FE) Server (Windows) – manages connected client

requests, and controls and operates other agents

• Agents (Windows) – perform raw data collection, network model

creation, and calculation tasks

• Data Servers

• Oracle Database – stores server configurations

• Call Database (run by a Linux machine) – stores the calls in a

generic structure, created based on the collected raw data

The following are the main Mentor Server building blocks:

Server Deployment

• Simple Server – central server handling all server-related tasks

• Includes a local agent, which is usually used for running

calculation tasks (analysis and optimization), and shares the

same file storage as the main FE server

• Enterprise Server – distributed system that includes one or more

remote agents, which are installed near the data devices and that

handle the collection and model creation

• Each remote agent has its own local file storage

• Scalable, supports large networks

The Mentor server can be deployed in two main methods,

according to the project requirements:

Enterprise Server – Front-End Clustering

• The server supports FE server-clustering functionality, where the

system allocates a serving host from its group of existing FE servers

• Only a single FE, the Master Server, manages the agents If it

crashes, another FE server takes its place

• Task state (queues containing the current status of the tasks) is

saved in a shared memory for all FEs, using Terracotta, enabling

any FE to be the master at any time

• Benefits

• Provides load-balancing and resilience in the system

• Provides high-availability to support user requests and agent

management

Enterprise Server – Agents Group

• Each network element (usually RNC/BSC) collection and modeling

task runs on a single defined agent

• The raw data is saved on the agent’s storage (grouped according

to date and hours), and is later used for the model run

• Agents Group – several agents can use the same storage, which

means that a network element can run on any one of them

• Calculations can run on any agent

Enterprise Server Architecture

Mentor Desktop Client

Forte Desktop Client

Insight Web Application

Call DB

Server Configuration

File System Cellular Data

FE Servers & Agents Agents Data Devices

Data Devices Agents

Logical Entities – NEs and Clusters

• A Network Element (NE) is the smallest building block of a network

defined in the server (e.g BSC, RNC)

• It contains configuration data (coordinates, antenna heights,

carrier definitions, etc.)

• A Cluster is a set of one or more NEs, and represents the view of the

engineer user

• A NE can be contained in multiple clusters

• Every user is permitted to view / edit one or more clusters

• Each recurring calculation defined in the server is related to a

specific cluster

High-Level Process Flow

• Data Collection and Processing

• Logs, KPIs, and topology files for every network element are

periodically collected from the data devices by the agents.

• The agents process and parse the collected data into a generic

and unified format

• For every NE, the agents periodically generate and save

network models in daily / NE resolution

• Call Database

• The generic and unified format calls (txt files) are loaded into the

Call DB by a daemon process using SQL loader.

High-Level Process Flow (cont.)

• Calculations

• Engineers can schedule recurring analysis, APD, and

optimization calculations to run on the server (through the Mentor

client or the Admin console)

Data Processing Workflow

Log & KPI Collection

Log & KPI Collection

Physical Collection

Physical Collection

Log Processing

Log Processing NE Modeling

Engineering Tasks

Engineering Tasks

Client-Server Interface

• Data Import

• Mentor users download the network configuration and model

• Optionally, to support call-related features (e.g virtual drive tests)

they can also import call data from the call database

• Manual network-configuration changes made by engineers in the

Mentor client (new sites, missing data) are updated to a specific

cluster in the server (actual changes are delegated to the NEs)

• Engineers can schedule recurring analysis and optimization tasks

Client-Server Interface – Import Workflow

Server Mentor Client

User requests cluster

data for several days

Data is merged from NE/Daily level to Cluster/Period level

Download via FTP/http

Client-Server Interface – Submit Workflow

Server Mentor Client

User performs changes in the

local environment.

User clicks “Submit”.

Calculates differences

Submits changes User accepts/rejects

Client-Server Interface (cont.)

• When connected to the server, engineers receive pop-up

messages with updates of the configuration of their cluster:

• Notifications

• Event-based notification messages are sent to users or by email:

• Etc.

Client-Server Interface – Sync Workflow

Server Mentor Client

Client requests differences (in the background)

Calculates differences

Apply changes on local User accepts/rejects

Lesson Summary – Q&A

• What are the main components comprising the enterprise

server?

• Describe the data collection and processing flow.