A PACS Admin’s Life

One man takes on the world of PACS, RIS, dicom, and all that is digital imaging.

Managing Storage In PACS

My organization is just a medium sized private clinic, so our storage requirements are dwarfed by hospitals and health authorities. Still, many IT vendors I talk to are simply in disbelief of how much data we generate, and how quickly it grows. It has been two and a half years since we implimented our first digital clinic. With only one clinic, and basic ultrasound studies, storage growth was pretty mild. Once we added more clinics and more modalities, our patient data skyrocketed. We are currently sitting at 2.5TB, and that is expected to be in the 4-5TB range within a year and a half.
Managing this data is a multi faceted problem.
* First, server speed is an issue. Multiple modalities are each sending data simultaniously to a server for storage. That server is then sending to various viewing stations as a pre-cache so that users have seemless loading of studies. That data is also sent off site for backup purposes. In addition, more then one user may be requesting other patient data from the server.
The local PACS server is constantly reading and writing data, and must be able to do so at a rapid pace.
* The network is perhaps the biggest issue facing us, and likely many other companies. The local network is pretty easy to deal with. Gigabit speed is a commodity, and does a pretty respectable job keeping everything moving. It should not be too long before 10Gb speed is widely available.
The WAN, on the other hand, is the real restraining factor in PACS as I see it. Some larger institutions have no trouble paying for OC-1 or gretter connections for their sites, but it simply is not feasible for us. Our 10Mbit wireless just gets us by, allowing backup to occur, but limiting the types of studies staff are willing to wait for as they transfer across the WAN. Not having faster connectivity between sites also dictates that each site needs its own server, rather than consolodating storage at one location.
* Data security and integrity, knowing that all data is accounted for and protected all of the time, is a massive challenge. With data moving between servers, checks need to be in place to ensure that data was not lost at any point. Software can mitigate this somewhat, but our Conquest PACS can not generate the needed reports, so we are left manually running queries to ensure the data is where it is supposed to be until we develop some automatic methods.
Backups of the data also come into play. With multi-terabyte data sets, it becomes more difficult and expensive to ensure the data is protected. We rely on having multiple online copies of the data in geographically dispearsed locations. This keeps us safe from natural disasters, but does little to protect against virus damage or purposeful sabatoge. However, having data on offline storage introduces a new aspect of concern, data loss or theft.
With an unlimited budget, managing storage, while not exactly easy, is
possible. On a limited budget, it is still possible, it just takes some creativity.

Bookmark to:
Add 'Managing Storage In PACS' to Del.icio.us Add 'Managing Storage In PACS' to digg Add 'Managing Storage In PACS' to FURL Add 'Managing Storage In PACS' to blinklist Add 'Managing Storage In PACS' to My-Tuts Add 'Managing Storage In PACS' to reddit Add 'Managing Storage In PACS' to Feed Me Links! Add 'Managing Storage In PACS' to Technorati Add 'Managing Storage In PACS' to Socializer 

Syndicate

Sponsership Centre

Activity

5 total comments, leave your comment or trackback.
  1. Excellent analysis, Drew. You had asked about Carbonite and retrieval of information from it. Fortunately I haven’t had to restore the disk from there as yet, but once you load (or reload if you have had a catastrophic failure) the software and log in, the image of your stored disk can be accessed just as any other lettered disk. Seems very straightforward.

  2. What vendor are you currently using for PACS?

  3. We are using Conquest software for all the back end PACS management, and eFilm for a viewer. The backbone of our storage is two AIC 48 drive storage chassis. I will be expanding on our systems in a future post. We are trying to get by with maximum gains by minimum costs. Not always a good recipe, but it certainly offers unique issues to overcome.

  4. Drew, thanks for the comments on my blog.

    Assuming, you are using Windows, you may want to look in to built-in snapshot technology to protect against virus/corruption threat.

    Are you currently backing up to a removable media at some point?

    How many and type of modalities are sending data to one server? Are you using same server for receiving new studies, retrieving priors, studies storage and workflow management?

    What percentage of new and prior studies are accessed remotely? Do you use pre-fetching?

    I guess enough questions for now.

    I designed infrastructure for several multi-site PACS incorporating geographically distributed remote clinics (3 to 25 sites, 2TB to 400TB), so I will be happy to brainstorm, if you desire.

    Anil

  5. Anil,

    If, by snapshot technology, you mean Windows Shadow Copy, I did look into this, but it does not seem appropriate for PACS archival. Firstly, it would take tremendous disk space as we grow, further stressing the budget. Also, it is not intended as a backup solution, and as such, restores one file at a time. It would simply be impossible to restore all the study information in a resonable time frame. However, if you mean some other technology, then I would love to hear more.

    Everything in our company does go to tape, EXCEPT pacs data. We simply do not have the means to handle that volume of data. At some point, I would love to get there. Until then, we live knowing that certain specific circumstances could permanently destroy our data. (keeps my boss up at night).

    Each clinic location has one PACS server. It can be communicating with up to 7 or 8 modalities (mostly ultrasound, some nuc med, fluro, and DR). It also handles the 5 or 6 client workstations that want images. Lastly, it sends a copy of everything it receives to a central storage server system.

    I do not have a number, but not too many studies are accessed remotely. There are a few times a tech or rad wants to see previous that do not exist on the local server (either too old, or done at another clinic). No pre fetching is in use at this time. The only thing we do is auto send all cases to the rad’s PC so they can work as quickly as possible.

    I very much appreciate your questions and any comments. Every thing we do is based on what I know. There have certainly been times in the past when I came across some knowledge or technology that drastically improved some aspect of the process.


Leave a Reply


Search

The archives run deep. Feel free to search older content using topic keywords.

  You're new! If you like it here, please subscribe to my feed.      
[Close]