Subscriber Policy Broker

OVERVIEW
Subscriber Policy Broker
Overview

Subscriber Policy Broker: Overview

The Subscriber Policy Broker (SPB) is a requisite component of Sandvine’s policy control platform deployment, and
fulfills two key functions:

Persistent Data Store Provides long-term storage for the Sandvine deployment. To maximize
performance and efficiency, storage is split between two separate databases:

• A subscriber database that stores subscriber profile information (e.g.,

subscriber attributes, session mapping information, location, etc.)

• A statistics database that enables long-term historical reporting

Subscriber Provisioning The SPB enables subscriber-aware policy control by making subscriber
information, session details, and related subscriber attributes available to the
Sandvine Policy Traffic Switch (PTS) and the Sandvine Service Delivery Engine
(SDE) in real time, both via notifications and a query interface, through a high-
performance, in-memory database.

Different Versions for Different Needs

The SPB is available in two product versions, the SPB Data Manager and the SPB Insight Engine.

SPB Data Manager The SPB Data Manager is well-suited for relatively smaller policy control
deployments that focus on straightforward reporting use cases using simple
reporting structures (e.g., basic per subscriber reporting based on element IDs).

SPB Insight Engine The SPB Insight Engine is designed for relatively larger deployments, and for
deployments in which the SPB is used in conjunction with big data systems or in
which it fulfills the role of a data lake.

The Insight Engine uses a columnar database designed for analytical and
exploratory use cases that depend on storing and retrieving granular information
with segmentation and complex data structures (e.g., reporting per-subscriber,
per-location, per-protocol).

The Insight Engine scales elastically, acting as one virtual cluster, and includes
a number of features specifically to enable natively big data functions and
interoperability.

Both SPB versions are completely virtualized, and run on Red Hat Enterprise Linux (RHEL) or Community Enterprise
Operating System (CentOS).

Please speak with a Sandvine representative to determine which version of the SPB is best for your needs.

2
 Subscriber Policy Broker
Overview

High Performance, High Availability

Performance and availability top the list of demands when it comes to databases, and the SPB delivers both through
a combination of four closely related technology features.

Clustering Clustering contributes to optimizing SPB performance and availability by combining many
SPBs together to scale performance and storage.

For both the Data Manager and the Insight Engine, clustering is an important element of
high availability and fault tolerance, and allows operators to segment network locations into
individual data homes.

In the Insight Engine, clustering enhances performance by enabling a practically limitless

number of nodes to operate as a single giant instance, allowing for RAID-like striping and
massive parallel processing for fast reporting.

To simplify deployment and management, clusters include automatic load balancing and
automatic discovery of new elements.

Roles SPB roles segment SPB functions (e.g., statistics storage, subscriber session management,
subscriber profile management) to enable independent scaling, redundancy, and
performance optimizations based on network needs.

Hierarchies Hierarchies allow operators to segment network reporting from local regions all the way up
to comprehensive site-wide visibility. In this manner, operators can optimize different parts
of an SPB cluster based upon specific needs and scale (e.g., smaller databases for smaller
regions, reporting up into a larger database for the entire network).

High Availability A range of technologies ensures the SPB is a reliable, always-on data store; these features
ensure fault management, in general, and minimized downtime during upgrade procedures.

Other SPB Features

Beyond those features described above, the SPB includes a number of additional features designed to simplify
deployment, to ensure smooth operation, and to enable meaningful network insight.

The table below lists some of these features, and indicates to what SPB roles the features apply.

Subscriber
Session Data Insight
Feature Description
and Profile Manager Engine
Manager
Top Talkers Automatically identifies the network’s top N
 
subscribers, in terms of bandwidth usage.
Web Services API Simplifies integration with operational support
systems and other external systems via a
  
SOAP or HTTP REST API to set and/or retrieve
data from the system.
Subscriber Attributes are type-value pairs
Attributes such as servicelevel=“gold”,
tethered=“true”, browser=“chrome”,

and masqueradeflag=“true”; attributes are
used as conditions in SandScript policy and to
enable deep business intelligence reporting.

3
Subscriber Policy Broker
Overview

Subscriber
Session Data Insight
Feature Description
and Profile Manager Engine
Manager
Subscriber Enables historical tracking of subscriber-to-IP
Attribute Archiver address mapping and of subscriber attribute
values over time.

Once the attributes are archived, they can be 

used in queries that join against subscriber
statistics, further enhancing reporting and/or
insights into subscriber trends.
Inactive Subscriber Optimizes performance by automatically
Management managing subscribers who have not been
provisioned for a configurable number
of days. Perform inactive subscriber
management in these ways:

• Automatically unassign all the IP

assignments from a subscriber after
the configured threshold of the number
of days of inactivity (default 30 days) is

reached.
• Automatically change the status of a
subscriber to inactive after the configured
threshold of the number of days of
inactivity (default 180 days) is reached.
• Automatically delete a subscriber after
the configured threshold of the number
of days of inactivity (default 240 days) is
reached.
Attribute Indexing Adds indexing capabilities to one or more
subscriber or session attribute definitions,
enabling efficient retrieval of the collection of
subscribers with particular attribute definition
values.   

This feature allows for use of non-identifying

keys to retrieve subscriber information by
outside systems.
Element Classifiers Element classifiers represent unique
measurements across one or more
dimensions that can be written to the SPB by
Sandvine elements, enabling powerful and
unique reporting and insights.  
This capability allows operators to create
their own custom report dimensions within
SandScript to extend well beyond the data
logged by default.
Auto-Discovery Allows data from new elements to be added
automatically into the database schema so
that no additional configuration (beyond   
pointing the element to the database) is
required.

4
 Subscriber Policy Broker
Overview

Subscriber
Session Data Insight
Feature Description
and Profile Manager Engine
Manager
Zero Provides self-diagnostics and automated
Administration clean-up of the SPB database without the
need for additional tools (e.g., no need for
re-indexing or refreshing of data).   

Additionally, the installation self-tunes based

upon available resources and defined roles.
Advanced Replaces slower disk I/O with fast CPU cycles
Compression through aggressive compression, reducing
the disk space needed for storage relative to 
traditional databases and allowing smaller
disks to be used for storage.
Columnnar Storage Analytical column store intelligently organizes
and Execution data on the disk, delivering extremely high
data query performance by reducing the

amount of data needed to be retrieved for
advanced queries and lowering the I/O load on
disks.
Distributed Query Takes advantage of massive parallel
Execution processing for faster query execution and

reporting by leveraging all nodes within the
SPB cluster.
Top N Aggregation Joins and aggregates encoded data for
efficient execution over billions of records, no

matter how many nodes are within a cluster or
group of clusters.
Elastic Data Store Automatically rebalances data when nodes
are added, removed, replaced, repaired, etc.

on-the-fly, and clones databases from one
cluster to another.
Database continues to support query requests
Zero-Downtime while rebalancing is in progress (e.g., during

Database upgrades or when taking down a node for
maintenance).

5
This document is published by Sandvine Incorporated ULC©. All rights reserved. No part of this publication may be reproduced, copied or transmitted in any form
or by any means, or stored in a retrieval system of any nature, without the prior permission of Sandvine Incorporated ULC. Although the greatest care has been
taken in the preparation and compilation of this whitepaper, no liability or responsibility of any kind, including responsibility for negligence is accepted by Sandvine
Incorporated ULC.

© Copyright: Sandvine Incorporated ULC 2017

408 Albert St. Waterloo, Ontario Canada N2L 3V3

www.sandvine.com