Autonomic Networking: a system that governs itself
The Ipanema's Autonomic Networking system is at the heart of Ipanema’s ability to connect the network infrastructure to the enterprise’s business goals. Self-learning, self-adapting and self-healing, Ipanema is the first and only system to deliver fully automated Application Visibility, QoS & Control, WAN Optimization and Dynamic WAN selection over the global network, whatever its size and transport technology.
Now is the time for autonomic networking
Enterprise networking means tens-of-thousands of users, thousands of sites and hundreds of applications. The only constant is change, since traffic patterns shift with new initiatives (data center consolidation, desktop virtualization), new applications (peer-to-peer, voice, video and tele-presence, social networking) and real-time users’ activity. Moreover, non-critical traffic increases exponentially, threatening business applications and putting the network and IT budget under pressure.
Traditional traffic management tools - called “policy-based management" - no longer fit in with this increasing complexity. They rely on site-per-site, partial information which are manually processed by engineers who then must make their best guesses for appropriate policies for each device. The tools are cumbersome to configure and not agile enough. Configurations are not optimal and cannot adapt to the ever changing traffic patterns that flow over the network. As a result, the performance delivered to end-user does not match expectations.
The time has come for autonomic networking, which enables automated, second-by-second decision making by distributed components that exchange information quickly and accurately, while re-focusing the efforts of the networking team on the more important tasks of defining overall applications performance objectives (APOs) and controlling their achievement with reporting and SLAs.
Autonomic networking is a fundamental step forward, bringing simple management to the constantly evolving complexity of enterprise network and application environments. The Ipanema System is leading the way in the commercial implementation of autonomic networking to the enterprise WANs, introducing unique functionalities such as:
- Carrier-class central management software allied to distributed cooperative devices
- Global performance management based on high-level application performance objectives
- Complete visibility of network traffic and application performance - from global to fully detailed views
- A very cost effective deployment scheme
Autonomic networking architecture: carrier-class central management software allied to cooperative devices
Ipanema's Autonomic Networking system uses carrier-class central management software - called IMSS for enterprises or SALSA for Telcos. The software automates back office functions such as global configuration and supervision, as well as measurement collection and consolidation. It provides a portal for entering application performance objectives and accessing reports.
Dedicated cooperative devices called ip|engines are typically positioned in-line between the access router (CPE) and the LAN. It is not necessary to install ip|engines at every site, only on main sites like data centers and headquarters. Bidirectional application flows are managed cooperatively between ip|engines, according to network performance and APOs.
Unequipped sites are “tele-managed” by software agents called virtual ip|engines, which are automatically made available in physical ip|engines according to network traffic characteristics.
Ipanema's Autonomic Networking system provides a carrier-class solution:
- Supports complex networks, including asymmetric routing, redundant configurations, sites with multiple VPNs, hybrid networks, etc.
- Highly resilient at device level - reliable design incorporates an embedded hardware failsafe
- Highly resilient at the system level, using self-healing capabilities to ensure service continuity in case of device failure
- Compatible with any transport technology, including MPLS classes of service, frame relay, Internet, etc.
- Extremely easy to grow with a centralized and global management model and configuration-less distributed devices
Global performance management from high-level Application Performance Objectives (APOs)
Ipanema's Autonomic Networking system manages application performance using APOs that are globally defined in central management software and communicated to all available ip|engines. The ip|engines form a distributed system that constantly evolves to satisfy and even exceed these objectives.
ip|engines perform a continuous real-time analysis of the application traffic and performance. Using the APOs, they continuously calculate the optimal way to handle all traffic. This includes complex situations like full traffic mesh and elastic flows, deducing the local policies that should be applied locally and enforcing them in real-time.
Ipanema's Autonomic Networking system removes single point of failure vulnerability:
- If the central management server goes down, the system continues to function correctly on the network as there is no change in the APOs
- In the rare eventuality of an ip|engine failure, the other ip|engines would automatically collaborate to manage the corresponding traffic on behalf of the faulty one
Complete visibility of network traffic and application performance
Information about site to site application traffic is transmitted by ip|engines to the central management software. Measures are then collected and aggregated each and every minute. The ip|engines are all synchronized within a millisecond, building a multi-dimensional table of all application flows on the network. This global table contains detailed information about each flow, including application identity, volume, source, destination and quality. The table can be used for long-term reporting, real-time traffic monitoring, network troubleshooting and alarming.
Historical views of applications, sites, etc. can be analyzed for any time period - from minutes to years. Custom reports focus on specific areas of the network, providing a very flexible and complete visibility on applications behavior over the entire network.
A cost effective deployment scheme
The Ipanema Autonomic Networking System fits globally and cost-effectively into any network topology:
- In star or hub-and-spoke topologies, only the data center needs to be equipped, the other sites being tele-managed.
- In multi-star topologies, where branches communicate simultaneously with multiple data centers, only the different data centers need to be equipped. The cooperating ip|engines manage meshed situations without the need for devices in branches, reducing the cost and complexity of application traffic management in large networks. Where there are high-delay or bandwidth-limited sites, additional ip|engines can be deployed to provide acceleration of application response times and additional virtual bandwidth.
- In fully meshed topologies, where branches are communicating with any and all network sites, each branch needs to be equipped. The cooperation between ip|engines ensures total control of application performance for all protocols (including TCP and UDP).
