September 29, 2014

Improving Computer Security for the CCDMD Servers

Did you know that more than a hundred sites are hosted on CCDMD servers? This encompasses all the sites and web applications produced by CCDMD from our call for proposals:

  • CEPO
  • The World in Images
  • 1000 Images on the Tip of my Tongue
  • A Listening Companion
  • Infants and Toddlers
  • Foundations of Effective Tutoring
  • Etc.

In addition, the CCDMD offers hosting for several partners of the college network:

As part of its strategic plan, the CCDMD has set a goal of improving service and data security and so this year, major work in the CCDMD server room was undertaken to do so. Here is an opportunity for you to better understand what is happening under the hood of the CCDMD online.

This year’s goals included the establishment of a virtual server as well as adding an additional air conditioner in the server room. In this article, we will focus on the addition of the air conditioner.

Cooling a server room is as important as the servers themselves. Without cooling, the room would quickly become hellishly hot and the hardware would overheat quickly and break down. This is what happened to us in the summer of 2012, which prompted us to take the bull by the horns in order to improve this situation. We only had one air conditioner then, and it broke down during the holidays.

Despite the complexity of carrying out such work, everything went very smoothly, and the service was interrupted for only 10 hours in total, which in our context is a major success. To this end, we wish to acknowledge the outstanding work of our two CCDMD technicians, Sylvain Gagnon and Jonathan Mark Lapointe, as well as the Physical Resources Service at College de Maisonneuve. Financially, this work required an investment of more than $ 45,000 and was funded in part by a grant from the Canada-Quebec Entente.

Other initiatives are or will be put forward to ensure optimum accessibility for the CCDMD websites while protecting data security:

  • The increase our bandwidth by the RISQ will enable larger traffic capacity.
  • An upcoming subscription to a new service offered by RISQ will protect against denial of service attacks.
  • The purchase of four new servers in the fall to improve the performance of some of our high traffic sites involving a relay server, ready to take over quickly in the case of the failure of another physical server.

How a New Air Conditioner Was Installed in the CCDMD Server Room

  • Temporary relocation of all servers. 220 volt service needed to be installed in this room. Reconnecting servers. Cutting web services between 8am and 12.30pm.
  • Simultaneously, work was being done on the roof: scan of the concrete slab to prevent damage to existing pipes or wires and drilling to allow for the passage of piping, fan installation using a crane, drainage of air conditioner coolant, electrical work, plumbing, delivery of the new air conditioner to the server room, moving, installation and commissioning.
  • Two days later, returning the servers to the server room. Web services cut between 8am and 2:30pm. Delay due to a problem getting servers started before restoring the internet connection.
  • Maintenance on the air conditioner 01 (Old). Replacement of components to correct a constantly too fast roof fan.
  • Finding error messages on the air conditioner 01 and a breakdown. Service call to the supplier.
  • Changing the pressure regulator, finding that the problem persists and changing a piece of pipe that contains a faulty valve. The faulty valve was improperly installed upon initial installation of the air conditioner. The air conditioner fan is cleaned and the problem seems resolved.
  • Problem this time with the new air conditioner. The device leaks due to welds that have been compromised by the heat. The welds are repaired.

Temporary relocation of all servers (photo credit: Philippe Bonneau)

Fan installation using a crane (photo credit: Philippe Bonneau)

Returning the servers to the server room (photo credit: Philippe Bonneau)

These few incidents demonstrate that redundancy for air conditioners is an essential. Normally the units will share the cooling load which reduces wear. Now, when one fails, the second is there to take over. Mission accomplished!

Notify of

0 Commentaires
Inline Feedbacks
View all comments