We have seen it happen again andagain, first with the Death Star, then with the Death Star II, and most recently, with the Starkiller Base. Precision shots by a pilot equipped with the right knowledge could destroy a weaponized base a thousand times bigger than his fighter craft. If you know what we’re talking about, chances are you’re a Star Wars fan, or have at least watched the series and understood the plot.
Medical Treatment Vs Death Star Destruction
So what do medical treatment and the destruction of Death Stars have in common? For one, it saves lives. For another, how insightful knowledge lies at the heart of such noble activity. But you see, insights do not just show themselves. They are the refined product of data analysis, supported by a base of powerful infrastructure.
To offer a better idea, let’s turn our attention to Sydney Adventist Hospital. Commonly referred to as the San, the facility is New South Wales’ largest private hospital, with some 53,000 patients and 180,000 outpatients treated annually.
Doctors here use a picture archive and communication system (PACS) as an essential diagnostic tool for patients. Since 2004, the average PACS study size for each patient has doubled from 32 megabytes to 76 megabytes with the number of cases exploding from 1,457 to over 64,000 studies a year.
“It is difficult to assess future storage requirements in a hospital environment because we can’t predict what diagnostic modalities will be brought on board,” says John Hoang, Senior Systems Engineer at Sydney Adventist Hospital.
At the San, specialist departments do their own research, attend conferences, and decide what tools will make a difference to patient outcomes. This means that the storage environment needs to be able to accommodate new tools as they are onboarded.
One can only imagine the kind of storage requirements needed to support such data expansion. And more than that, the speed required to scour and retrieve the right data files.
Imagine if you were an X-Wing fighter pilot taking heavy enemy fire and you had just one fly pass to take a single shot to destroy the Death Star. Chances are you’d need to know where to make the shot, and you’d need to know it really quick.
Similarly, time waits for no man in the medical world. Sometimes doctors only have one shot to treat a patient, and they need all the insights they can glean from patient diagnostic data. And they need it fast.
Turning Uncertainty into Certain Certainty
In addition to the San’s existing EMC VNX unified storage solution, a PACS storage environment has been deployed with twoclusters at two separate sites with 85 terabytes of storage. This has been further enhanced by to provide easy-to-manage replication of data between the two sites, which is critical to the hospital’s agile infrastructure, ensuring all nodes in the Isilon cluster concurrently send and receive data during replication jobs.
“The PACS system is more resilient because more storage is handled on a multi-node architecture. If we lose one node, we still have two nodes online so specialists can continue to retrieve images. In this sense the technology has paid for itself – we simply don’t need to worry about outages or disruption to services due to storage limitations anymore,” explains Hoang.
Another piece of good news is when data volumes increase, capacity can be seamlessly scaled to ensure performance is consistent. An Isilon X-Series cluster can be easily brought online within minutes, with a single cluster possessing the ability to scale from a few terabytes to more than 50 petabytes. Now that’s over 200 gigabytes per second of throughput. From another perspective, a patient’s entire PACS history file could be downloaded even before a doctor finishes saying, “Do you feel pain here?”
A Legacy Forged Today
A long time ago in a galaxy far, far away…
Perhaps in a century or more, technology and medical historians would look back at this period as a defining moment where data represented infinite possibilities for human health. Who would have thought something as simple yet complex at the same time, such as data storage and retrieval, could play such a pivotal role?