Network documentation and monitoring are topics that never lose interest to me.
Each product has its strengths and weaknesses that in some cases are nothing more than the aspects on which the manufacturer has decided to concentrate development more.
Some vendors try to sell the golden bullet for network monitoring, the legendary single pane of glass that is supposed to collect all the information about network configuration and state.
We may get there some day, some products are closer to the final result than others but usually that happens in small niches with design and topology strictly under control like a Data Center.
The reality is much different, many network engineers fight daily with campus networks that look more like a Jungle than what we can find on reddit/cablep0rn.
I am too pragmatic to believe the golden bullet that solves all problems, I find the Unix philosophy more similar to the truth
Make each program do one thing well.
where the solution is made by the integration of different products, each one solving part of the problem with some logic on top that glues together all the elements.
This approach applied to network documentation and monitoring may require some coding skills but the final result would have some advantages including
- no useless features pushed by vendors
- no dependency on a single product
- possibility to replace an element that does not meet the requirements
All this comes to a cost of course, the main one being that development, R&D and QA are on charge of the integrator.
After a positive first impression based on the website I decided to download and try it.
This is not only a matter of style but it really helps me, as a potential customer, to grasp in a few click if the product is suitable for my needs without having to search documentation behind a registration wall or to extract something useful from from documentation created by marketing departments.
What is AKIPS then?
AKIPS installation and setup
The setup is quite simple:
- download and install the VM
- request a 30 days free license to activate full features
- run a discovery / walk of the network
In my case I tested AKIPS on a campus made mostly by Cisco switches, the discovery process took 3 minutes to discover and snmpwalk 214 devices on the smallish sized VM with 1 CPU and 4GB or Ram.During the discovery process the CPU reached 25% use, not bad.
How can it be so fast? The poller architecture is described in detail on AKIPS blog.
Discovery setup includes selection of IP subnets to discover and SNMP cmmunity to use:
Discovery log shows discovered devices:
AKIPS in action
The discovery process uses SNMP MIBs to extract information, the output is very clear, this is a Cisco 3560 switch:
Based on device AKIPS can monitor CPU and port use, temperature, fan status and all the usual values you can get via SNMP - no different than other tools like LibreNMS but with some differences I’ll discuss at the end of the post.
The CPU report is useful to find devices with potential issues. The CPU use on these switches is high, bad software easily fixed by an upgrade:
One of the strongest points of AKIPS are the reports, in the screen shot below there are a few, there are so many that would be hard to show and discuss all of them here:
Reports can be exported in PDF
There are plenty of other feature AKIPS offers, including Switch Port Mapper and Unused interface report (similar to unpatchable).
Last word on features go to alerting where we can find the usual alerting via email with granular customization:
API are a base requirement for any integration today. AKIPS has API, I tested a few with Postman:
Details on AKIPS Programming Guide
The initial impression about AKIPS is positive. The scope of the product is clear, installation is easy, initial setup and configuration is a matter of minutes.
I really like the intuitive interface, not missing all the clutter often present in some software that want to display too much in little screen space.
The big difference is in the performances. The same campus is being monitored by LibreNMS server with twice CPUs and memory and optimizations applied. Gaps in the graphs crealy show that LibreNMS can’t collect all the information in time.
AKIPS gets the same values in less time and barely using the CPU.
I dindn’t tested NetFlow and some other features but I’m planning to do that soon.
The product feels solid and responsive, it’s a good candidate for network monitoring. It doesn’t solve all the problems but this is positive, as I said earlier I prefer a software that focuses on a smaller amount of features well done than something that tries to solve everything and becomes a Hydra quite soon.
HTHcomments powered by Disqus