By Balaji Venkatraman, Director, Product Line Management, HP Network Management Solutions
The most basic requirement of any fault and event management product is to continuously monitor the network infrastructure for the relevant parameters and store it in the database. Once this data is stored, it can be used to generate alerts and reports in real-time. These reports including capacity planning and sophisticated analysis reports (e.g. Root Cause Analysis) are then used to aid with forensic analysis.
A central question in the design and development of such a product is: how do we architect the network discovery (data collection) capability and how do we store that in the database for efficient retrieval?
In this brief article, I will focus on NNMi’s patented, Spiral Discovery process, which is responsible for continuous Layer-2 and Layer-3 discovery. As a product manager of the industry-leading HP Network Node Manager i (NNMi), I am always trying to see how I can articulate the value of a feature to solve customer problem. I feel that the impact of the Spiral Discovery process is too important to keep to myself!
In Spiral Discovery, discovery and inventory data collection occur in stages where each stage provides information that is used to drive subsequent stages. This figure illustrates the process:
HP NNMi's discovery process
NNMi’s Spiral discovery engine avoids any episodic or batch discovery and allows NNMi to track dynamic changes in near real-time. Spiral Discovery uses a variety of MIBs and protocols to reduce overhead and manage multivendor environments. This means that NNMi can discover network topology changes, and use that information to drive additional value added analysis.
The network administrator configures the discovery process by defining the starting points for discovery. The administrator also configures additional parameters such as:
System object ID ranges
Exceptions and limits to be added to the NNMi database
When a device is identified, Spiral Discovery collects the basic information to understand what kind of device it is. That information, in turn, tells it what additional device inventory information is available. Details are collected, augmenting the inventory with additional information and connections that continue the process. In addition to automatic discovery, NNMi also allows for connection editing and subnet connection rules to override the link-level discovery protocol information when needed.
To minimize management overhead and continuous management of the product, NNMi’s internal throttling mechanism limits the amount of outstanding requests to any device. NNMi discovery runs until it completes the initial discovery, and then goes into "hint mode" until the next discovery is triggered.
So what are the benefits of Spiral Discovery?
1. Discovery and inventory information is collected in stages, thereby avoiding bursty load on the network and network devices and improving NNMi’s ability to track dynamic changes in “real-time”. This capability enables NNMi to scale to the largest Enterprise/xSP networks; for example: 100k interfaces, each with an average for 10-polled elements supported in a single NNMi instance)
2. Spiral discovery is dynamic. It continuously discovers the existence of network devices so network changes are quickly detected and incorporated into topology automatically. This capability enables NNMi’s topology to always be up-to-date, enabling Network Operators (Net OPS) to identify and troubleshoot problems in real-time as they happen.
3. Spiral Discovery is extensive. It collects detailed information about each device, including component health such as CPU, memory and fans. Spiral Delivery maintains the detailed inventory information in the NNMi database. It determines how devices relate to each other in network topology and topology overlays like virtual private networks (VPNs) and virtual LANs (VLANs). It also discovers and understands routing cluster protocols (HSRP, VRRP) as well as port and link aggregation. This capability enables NNMi to have a complete and accurate L2- and L3-model of the network.
4. Spiral Discovery is automatically incorporated into Root Cause Analysis. When things change in your environment, your topology accurately reflects those changes in near real-time. The NNMi casual engine reacts to topology changes as they occur for a quick analysis of root cause for incidents. This capability enables NNMi to be adaptive, self-learning and keep up with the changes network.
5. Spiral Discovery is extensible. It is designed so that HP can easily support new devices with NNMi as they become a part of your network in the future. Add-on iSPI products (NNMi Smart Plug Ins) not only add content, performance and analysis detail for things like MPLS, IP Multicast, and IP Telephony, but they also extend core NNMi discovery in those areas.
Mark is currently employed by Hewlett-Packard's Professional Services (PS) organization and serves as a Lead Solutions Consultant for HP's Business Service Management (BSM) set of HP software products. Mark is responsible for HP Software architecture and solution design for HP's comprehensive IT Performance Management solutions. Mark works with some of HP's largest customers within the continental United States. Mark comes from a strong background of understanding an organizations business needs for monitoring modern day business applications. Mark has held many different and challenging positions and responsibilities over the years at Hewlett-Packard Company including roles as a Senior Product Marketing Manager for HP's Network Management Center of products and has held various software presales roles including Dedicated Network Consultant, District Presales Leader and Solutions Consultant. Mark understands the many challenges that face network operators and corporations today and can craft management solutions from the smallest to the largest of enterprises. Mark enjoys working with leading edge technology as that engineered by HP Software.