When PACS Pushes the Limit - Image Storage: The Basics

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon

 With image volumes surging every day, many healthcare facilities are re-evaluating their image storage strategies. And as more sites move toward filmless and paperless environments, the need for more storage space seems to be never-ending. Fortunately, as patient data, imaging studies and archived images increase, the cost of storage is coming down and choices in data storage are expanding.

Hospitals and imaging centers running out of disk space on their PACS are looking for ways to upgrade infrastructure to expand capacity. “We’re seeing growth all over, but specifically in cardiology and medical imaging, as we digitize more, the images are getting bigger and bigger, such as in digital mammography and 64-slice CT,” says Jonathan Goldberg, VP and CIO at St. Peter’s Healthcare Services in Albany, N.Y. “We’re seeing huge growth rates in terms of what the needs are.”

Facilities down the road and across the country are experiencing the same thing. Health Imaging & IT spoke with several that are revisiting their storage strategies. How to save short term? What to save long term? How can I guarantee immediate image access? They provide some answers on quenching the storage need.


Image storage overview



A PACS archive requires two levels of storage: short-term and long-term. The short-term or online storage is a temporary directory that receives files directly from the imaging modality, and typically stores images for 12 to 18 months on fast-access, spinning disk with a retrieval rate of two to five seconds per study on average. Long-term or near-line storage is the more permanent form of storage where images are housed for about two years.

PACS archives typically need networked storage, such as a storage attached network (SAN) — a network that connects storage devices to computers, or network attached storage (NAS) — a network that’s not attached to servers, allowing access to storage using network standard protocols. A SAN is about two and a half times more expensive than NAS because a SAN requires dedicated networks and costly gigabyte switches.


Forming a winning team


St. Peter’s Healthcare revamped its infrastructure in 2004 to manage the organization’s explosive data growth of about five to 10 terabytes per year. The healthcare network comprises 24 facilities, including a 450-bed hospital, specialty clinics, nursing homes and rehabilitation centers. “We were running out of space on our current storage,” Goldberg says, adding that when their Siemens Medical Solutions Sienet PACS was installed in 2004 it relied on an optical disk jukebox, and they were coming to the end of that capacity. “We knew at that growth rate, within four to six months, we’d be out of disk space.”

First, St. Peter’s looked into an expansion module they could simply plug into their current storage, but they soon realized that wasn’t practical in preparing for the future. They ultimately selected EMC Corporation to create a centralized, highly available storage infrastructure. Goldberg said they chose EMC because of the reputable relationship between Siemens and EMC, and the fact that Siemens PACS easily integrated with EMC CLARiiON NAS and EMC Centera content-addressed storage (CAS).

St. Peter’s merged radiology and cardiology PACS images, as well as other application storage onto both the CLARiiON and Centera to archive stores of data and to serve as their primary storage. The CLARiiON CX700 NAS system is designed for environments like St Peter’s that require high capacities (40 terabytes) and maximum application performance. It works best for heavy database applications, as it scales up to 76 TB with a modular pay-as-you-grow approach.

CLARiiON serves as St. Peter’s storage area network, storing all enterprise organizational data, including serving as the disk for their 100-plus virtual servers. Centera provides archive storage for medical images both in radiology and cardiology PACS.

St. Peter’s evolving storage infrastructure has cut data retrieval time for patient-critical from five minutes to only 30 seconds, Goldberg says. Physicians are extremely happy with it. “They like the response time, and they like the fact that they can go from study to study quicker. Now it’s an expectation that we’ve set,” he adds. “To them, it’s invisible. They have no idea it’s there. The only thing they know is that [images are available] a lot quicker.”

The new storage strategy has improved St. Peter’s customer satisfaction, made the infrastructure easier to manage, helped secure information, plus, there’s built-in redundancy, and it’s easily expandable, Goldberg says.


The long-term view


Created in 1995, Iowa Health System (IHS) of Des Moines, Iowa, also required an infrastructure upgrade, but they were looking for storage software that could outlast future technology changes and upgrades. Iowa Health generates more than 1 million imaging exams per year that use about 15 terabytes of primary storage.

There are lots of images to manage for a health system that operates 11 physical hospitals in seven large Iowa communities and Rock Island, Ill. They also support a system of rural hospitals in 14 Iowa communities and partner with physicians and clinics in more than 80 communities in Iowa, western Illinois and eastern Nebraska.

To store these images, IHS opted for the IBM Grid Medical Archive Solution (GMAS) that uses a standard file system instead of a proprietary file system. IHS preferred IBM’s standard system approach because they knew for the long term that they would be able to migrate data regardless of the futures of IBM or other software companies involved.

GMAS is an automated, self-optimizing distributed grid storage solution. It allows a multi-campus hospital to link separate storage systems together while offering redundancy and ensuring multiple copies of data are stored in at least two places. GMAS is Digital Imaging and Communications in Medicine (DICOM) content aware and allows for Hierarchical Storage Management/Information Lifecycle Management based upon a file’s metadata.

In a grid storage architecture, independent storage nodes are linked and governed by common control software. That control layer provides a single management interface and fault tolerance among the nodes, as well as the ability to access either file-level or block-level storage. It also makes it possible to reassign nodes to different functions, such as from online to archival storage, as needs change.

“Organizations like ours are required to keep digital images for many years, in some cases the lifetime of the patient. Technology doesn’t last that long, so we know that every five years or so we’ll need to transfer the media that are holding those images from one technology to another,” says Bob Thompson, director of governance at IHS. If a vendor chooses a proprietary file system, the health system is dependent on that vendor’s expertise and product being available to actively support that environment when it comes time to transfer data, Thompson says. But if that vendor expires, is acquired or withdraws support, then users may be stuck with their current environment and not able to easily transfer images, he adds.

Other reasons that IHS chose GMAS: IBM’s enterprise class (IBM eServer xSeries Intel Servers and TotalStorage DS4000) and its excellent support, Thompson says. “We were looking for compatibility with our GE Healthcare environment. IBM and GE worked very well together to prove that integration and performance,” Thompson says. “We saw excellent support from IBM. We knew that if there was a hardware issue, IBM would be there on the spot to correct it.”

The redundancy — maintaining two copies of the images — is completely automated through GMAS, and IHS achieves that automation at a lower cost than would be the case if a hospital had to purchase conventional technology and labor to get that type of redundancy, Thompson says.

GMAS requires very little labor to manage it, which is a great contrast from Iowa Health’s previous environment where they were using similar hardware but were trying to manage it themselves. “The additional software that GMAS adds to the equation automates the difficult labor involved in managing that environment,” Thompson says.


Ease of administration


William Beaumont Hospital is a two-hospital regional medical center in Royal Oak, Mich., with a staff of 15,000. The medical center faced data growth across its two hospitals, seven medical buildings, a rehabilitation center, primary-care clinics, four extended-care centers, a research institute, and a hospice center. While the hospital used networked storage in many areas, each division and group had its own storage systems.

The hospital needed to accommodate new data without compromising the existing infrastructure. This involved consolidating storage on networked attached storage (NAS) using NetApp clustered FAS960 and R200 systems with Data ONTAP 7G technology, and moving from nine NetApp systems to three NetApp storage clusters and four NetApp NearStore systems. They also use NetApp SnapMirror, SnapRestore, and SnapVault software to back up the database.

Project Manager Jess Carruthers lead the team that deployed NetApp Data ONTAP 7G software with FlexVol technology, which offered the type of flexibility and ease of administration that was needed for Beaumont Hospitals’ new and improved IT infrastructure. Carruthers cites NetApp’s ease of administration and its virtual volume, Snapshot technology, and vaulting as its key qualities.

Currently, William Beaumont Hospitals take archive files and put their archive into another volume, which is mirrored every five minutes. Using Snapshot technology, each evening they put all of their databases in backup mode and create a snapshot at 1 a.m. “Using the 1 a.m. snapshot and archive logs that run every five minutes, we can get within five minutes of the outage,” Carruthers says.

Previously, that type of restoration took days or weeks, says Carruthers. They’ve also used less space with NetApp than with other disk systems because of the concept of virtual volume. The idea is this: The hospitals create a very large aggregate and in that large aggregate they create multiple volumes, one for each application, and they can grow and shrink those volumes on-the-fly based on the requirements for the application at that point in time, he says.

After the migration to this system, storage utilization rose to about 76 percent, reflecting almost a 50 percent increase, according to customer-managed testing conducted by NetApp. “This storage utilization helped us because we don’t have as much disk space waiting around, and we don’t have to buy as much disk,” Carruthers says. “Now we can use all of our disk space for the aggregate and flexible volumes.”


Worth the investment


What all of these facilities have in common is that investing in storage software that integrates easily with current PACS has helped improve data retrieval time and overall storage utilization. St. Peter’s, IHS, and William Beaumont Hospitals have all made advances toward preparing for explosive data growth and image overflow in the future. A word of advice from these technological pioneers: Start thinking about storage options now, before your hospital runs out of disk space.