Planning makes perfect for successful upgrades to the site data systems

Cory Stuart, Site Data System and Cyber Security Manager, Argonne National Laboratory
Cory Stuart, Site Data System and Cyber Security Manager, Argonne National Laboratory
Throughout 2016, Cory Stuart, ARM’s Site Data System (SDS) and Cyber Security Manager, and his team at Argonne National Laboratory have been methodically visiting all ARM sites to upgrade the data systems, especially storage capacity. By their last installation—completed at the Eastern North Atlantic (ENA) site in the Azores on August 29—they had their on-site time down to a mere two days.

The secrets to such rapid-response deployments, which minimize data loss and cost, were preplanning, component testing, and up-front configuration at Argonne. Brent Kolasinski of the SDS team coordinated much of the testing. In fact, the entire effort has gone like clockwork because of professional attention to these critical success factors.

Big Data Needs Big Capacity

In early 2015, the ARM SDS team began capacity planning for all sites for the next five years to handle the reconfiguration at the Southern Great Plains (SGP) megasite, increase data storage capacity to deal with heavier data volumes from new radars, and maintain reliability, redundancy, and performance. A final goal was reducing the support effort, especially at remote sites.

Given these factors, the team decided to go seriously big. From 16 terabytes (TB) of capacity at all sites before the upgrade, SGP would be increased to 180 TB, the ARM mobile facilities to 120 or 125 TB, ENA to 120 TB, and the North Slope of Alaska site to 58 TB.

To kick off the upgrade effort, the team installed a pilot version of the new data system in the second ARM Mobile Facility for the ARM Cloud Aerosol Precipitation Experiment (ACAPEX) on the U.S. West Coast, thereby verifying the basic specification. With proof of concept done, the team spent the rest of 2015 "spec’ing" other sites and ordering new equipment.

Haggling paid off, Stuart says suppliers agreed to discounts totaling 50 percent.

Hopscotching the Globe

As the SDS team upgraded each ARM site around the world during 2016, they encountered unexpected challenges in the field. For example, cooling is normally not a problem at ARM’s Arctic sites at Utqiaguik (formerly Barrow) and Oliktok Point, Alaska. However, with the extra heating from the new data systems, it was necessary to add an air conditioner at the last minute. Heat management also posed a problem on tropical Ascension Island, with temperatures reaching a baking 140 F° in the servers. The system again required additional air conditioning and reconfiguration of the computer system.

Capacity in a Can

A rear view is provided of the new data system at the Eastern North Atlantic site in the Azores.
A rear view is provided of the new data system at the Eastern North Atlantic site in the Azores.
As Stuart says about the new data systems in the mobile facilities, “This is seriously cool! It’s easy to forget how unique this is. Basically, we’re running a complete enterprise data center in a seatainer."

"It’s hard to do this even in a purpose-built server room in downtown USA, much less in a steel box deployed to places as remote and potentially hostile as Antarctica or Ascension Island. Other than maybe the military, I’m not aware of any other organizations doing this.”

ARM’s aim in this upgrade is to ensure adequate data capacity for the next five years. If more is required, the current solutions can be grown to meet the need. As hardware ages and becomes too costly to maintain, it will be replaced and upgraded.

For now, mainland facilities can hold six months of data, while mobile facilities can hold nine months. The first and second mobile facilities each now boast a whopping 125 TB of usable disk space, about eight times what they had before.

This “overkill capacity,” as Stuart terms it, is primarily to ensure disaster recovery, but it can also be used to allow for on-site processing. Yet just as auto traffic rapidly fills any new freeway system, SDS expects the same with ARM’s huge new storage systems. So, while data now flows faster than ever, five years could come much quicker than expected.

# # #

The ARM Climate Research Facility is a national scientific user facility funded through the U.S. Department of Energy's Office of Science. The ARM Facility is operated by nine Department of Energy national laboratories including Argonne National Laboratory, which leads ARM's site data system and cyber security management.