Lindsay Hill automation, networking, product management

ScienceLogic Architecture Overview

This is a basic overview of the ScienceLogic EM7 system architecture, describing the various components, their functions, and how they can be combined or split across multiple systems. I’ve been deploying ScienceLogic systems recently, so this blog acts as my own notes/summary.

(Read more...)

War Stories: Loops that Permanently Broke the Network

(Read more...)

HP NNMi Licensing Changes

HP has made some changes to the way Network Node Manager (NNMi) is licensed. These changes are a Good Thing, as they address two key pain points - license calculation, and access to extra features.

(Read more...)

Too Many Communities

I have come to realise that I’ve tried to take part in too many communities, and it’s getting me down. Too many forums filled with too much noise, too many people not doing basic research. Too much time on my part to scroll through various RSS feeds & email lists. Time to re-evaluate what I’m doing, and why.

(Read more...)

nxlog - Convert any text file to Syslog

Recently I’ve been converting a network from an agent-based monitoring system to an agentless system. One challenge was handling custom application logfiles. Most agent-based monitoring systems make it very easy to add logfile encapsulators, to read a custom file, and react to specific patterns. But now I needed to convert the logfile into syslog, so it could be directed to my NMS, which would parse the syslogs, and raise alerts as required.

(Read more...)

The HP SDN APP Store - It Might Just Work

HP has been laying out their SDN vision over the last few months. They want to develop a complete SDN ecosystem, including an Open Standards-based network that can integrate with common overlay models, development resources to help 3rd parties to develop applications, test & validation tools, right through to an “App Store” that will deliver applications to end-users.

(Read more...)

Network Monitoring Design Choices - Polling Distribution

Network monitoring systems poll devices periodically, and then calculate average rates across the poll interval. Selecting the right polling interval is important - too frequently, and low-end devices can’t keep up. Too long an interval, and we might miss spikes, will lose detail, and will increase our time to detect & respond to incidents. We may choose to sample some specific values, or devices, more frequently - e.g. 1-minute status polling of our core devices, 5-minute polling for edge devices. But what about the collector - how does it decide when to run those polls?

(Read more...)

DHCPv6 on Juniper SRX-110 - Progress

Last year I posted about my frustrations with getting the DHCPv6 client working on a Juniper SRX-110. I am pleased to report that Juniper has now released 12.1X46-D10.2, which resolves at least some of my issues. I couldn’t find it documented in the Release Notes anywhere, but the parser now allows you to have client-ia-type ia-pd without also requiring client-ia-type ia-na

(Read more...)

SDN App Stores - My Network is not a Phone

I’ve been thinking a lot about HP’s SDN App Store approach. In case you missed it, HP’s vision is to deliver an App Store for SDN applications, similar to Apple’s App Store. They want to have all SDN applications available in one place, with centralised purchasing and distribution. It will include both HP and 3rd Party-developed applications. HP says:

(Read more...)

SiteHost - Top-Notch Service

We needed a hosting provider that could deliver a platform to run a Virtual Appliance on. The issue was that we needed a reasonably high level of RAM, and we really wanted something based in New Zealand. Most IaaS providers are quite expensive if you want a high memory instance, and/or something that will run 24/7. We worked out that the most cost-effective option was to lease a server, install ESXi on it, then run our virtual appliance on that.

(Read more...)