Proposed Projects Done
Week 2 of work placement goes around in more detail with the traffic through put. By using PRTG dashboard maps, I created a well visualized network of NPIL. All the Juniper switches are spread out and visually connected with the correct switch as with the original network design.
The idea of making this diagram is to know how traffic that goes through a switch and to help diagnose which area causes the jam. It had been notified to me that the design of this network has a high risk of network bottle neck as it is serially chainned to the core. Some even have 4-5 chains in them. Thus by knowing each of bandwidth that goes through every switch, the network administrator could easily detect and diagnose which system causes the problem.
This project is quite straight forward by mapping out the switches as by the design. However, as it had been quite a while since it had been updated, There are some additional switches which are new ones and are not in the map at all. Simply put, it was just not documented yet, just deployed. By making this map, I indirectly contributed in sorting out some found issues on the switch network such as incomplete switch map, unusual ports and physical issues.
In the creation of this map, there are several major issues which could make the project not working properly. The first one is the issue with PRTG, in which its gauge system which takes the highest value that had gone through a sensor to be the maximum value of the gauge. This was an issue for us because in some time, even from a 1 Gbit/s maximum uploink speed on the switch port, the maximum value reached 200 Gbit/s. Which if we put a warning and error limit, the gauge will look fully red all the way but not in a warning state. It is not that it causes an error, it is just not visually suitable for the job. The map was supposed to help the network management visually. With that, I had to create a custom map object in PRTG to display last value of a sensor in numbers only. In addition to that, a traffic light is displayed as well along with each switch to display its bandwidth condition.
The second issue was port issue, we found several port which does not display the amount of traffic it is supposed to be. After a few investigation, it seems that there was a change on this port that goes to the core switch a year ago. Without changing the labels, it was a misinformation shared of through that port. There are another part where labels were an issue as it doesn’t provide the correct information about the traffic that goes through. These issues were not able to be done within the IT team knowledge. Thus, we contacted CCL network engineer to provide detail on the issues at hand.
After having a few issues on downed APs in the network, I proposed a new map out of my original project which is a map of AP-switch network. This would help identify where the location of an AP is when it is failing and easily check which switch it is connected to and ultimately it’s location. Previously, they had to get back to a document and search through which is connected to which. With this map, it makes things easy as it also displays the condition of the AP.
As all the proposed project provided by my supervisor has been almost done now, he proposed a new one on making a documentation on the procedures of changing a failing switch when in an emergency. It is an interesting project to look through as I would need to learn more on how Juniper switches work and what are actually the procedures to do it especially with minimal down time as possible as this is a 24/7 running factory.
Marcellus's Blog 