Using a single Ethernet port on each AO server

There is a pending request from LBTO staff to reduce the number of Ethernet interfaces used by the AO workstations.

This should be feasible, despite the problems quoted in NetConfiguration, due to the capabilities of network switches used at LBTO. A set of preliminary tests should anyway be performed in order to verify that a modification in networking configuration does not trigger unexpected problems.

There are two aspects to be considered:

  1. Insufficient throughput or increased latency for the communication of diagnostic data currently supported by a direct cable connection between two ethernet interfaces on the adsec and wfs servers.
  2. Unwanted network traffic reaching the BCUs.

Evaluation of network traffic

In order to provide better data for the evaluation of aspect 1 we have activated on the two AO servers a procedure to gather data on etherent traffic on each involved NIC. We suggest to gather data during actual use of the AO system so that we can have reliable measures of the actual traffic. In the meanwhile we are gathering the same data on the AO servers used at Arcetri Solar tower for a comparison.

Unwanted traffic on BCUs

In order to verify that the BCUs are not receiving unwanted (and harmful) traffic, we suggest to make a modification to the current network configuration on the two AO servers, by assigning the different IP numbers currently used for different interfaces to the same physical interface. After that we must do the following checks:

  1. Verify the basic functionality of the system (i.e.: all required network connections can be established). If not, a reconfiguration of VLAN settings may be needed.
  2. Verify the (possible) spurious traffic on BCU ports. This point is somewhat triky because putting a sniffer in place of a single BCU does only provide information when the system is not operational. It would be far more significant to enable packet mirroring from the switch ports dedicated to the BCUs to a spare port on the AdSec switch and put a sniffer on that port during some (daytime) activity.

Both points require support from LBTO staff for the reconfiguration of Ethernet switches.

Logistics

The evaluation of network traffic is currently ongoing, we only need to wait for a period when the FLAO system is actually operational, and them some offline time for data analysis.

For the BCU tests we need some time (daytime) during the commissioning activity to modify the configuration and let the system go for a while while sniffing on port (we need at least two people, one of them up in the observing chamber) and the support of the LBTO network engineer for the configuration of switches. The full test should require 2-3 hours to complete.

Topic revision: r2 - 02 Mar 2011, LucaFini
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback