Data Center Improvements

We've been working to improve the performance and reliability of our data centers in the Sliger and Dirac buildings.  This effort includes a major overhaul of our network and a consolidation of resources into the Sliger Data Center.

Over the past several months and continuing into 2018, we have been making major changes to our data center configuration.  The goal of this effort is to improve reliability and managability at the physical level.  You may have been affected by some of the HPC Maintenance we conducted this past fall.  That maintenance was part of this effort.

Our aims are as follows:

  1. Consolidate more HPC nodes and other critical RCC systems in the Sliger Data Center
  2. Replace older network switches and routers with newer, faster equipment
  3. Re-configure network topology for greater performance and manageability

RCC equipment resides in two datacenters; one in the Dirac Science Library, and the other in the Sliger Data Center in Innovation Park. These two datacenters are connected via a dedicated network tunnel over the FSU campus network.

Hurricane Hermine hit the Tallahassee in September, 2016, followed by Hurricane Irma in 2017.  These two weather events, the first of their intensity in decades, demonstrated the need to fortify our physical environment against future events.  All of the damage and residual issues that occurred after each storm occurred in our Dirac data center.  So, in 2017, we decided to consolidate the majority of our servers and other equipment in Sliger.  The Sliger data center provides a more resilient environment for the growing cluster.

We are also taking this opportunity to make major upgrades to our network infrastructure.  To this end, we have replaced our primary IP network switches with newer equipment, and we have reconfigured our topology for better maintainability.  While these changes may not seem immediately evident to end-users, they will result in faster performance and higher uptime.

In Fall 2017, we completed the consolidation and network upgrades of all owner-based HPC nodes in Sliger.  We also moved the Spear System, login nodes, and some important admistrative servers from Dirac to Sliger.

In the coming weeks, we will continue this project by doing the following:

  1. Moving the Virtualization Cluster (VM "SKY") System to Sliger
  2. Moving additional HPC nodes to Sliger
  3. Moving more administrative systems to Sliger
  4. Completing remaining network upgrades

We believe these upgrades, together with our new storage system, will provide a robust and reliable platform for computationally-intensive research at FSU going forward.

We'll continue to post updates as we progress with this project.  Thanks very much for all of your patience during the periodic maintenance windows.