dcm4che as a legacy archive

[Pete]

Hi All,

I am looking for some guidance from someone with practical experience migrating to, and sustaining a dcm4che system as a legacy archive.

I work with an organization that has about 100TB (1.6m exams) of "dirty" legacy data, on a Dicom archive system that is no longer supported by the vendor, and is on aging hardware that is becoming very difficult to find spares for. The system is not being used for any net new data, however,  it is daily used for research and to complete a patient history.

The organization does not want to migrate data into the new environment because:
a. the data is not clean
b. there is a large cost associated with the new vendor to support that the organization can't afford

I have tinkered with dcm4che in a test environment and it looks like it has great potential as a legacy archive, however, I am wary of some of the pitfalls.

If we were to stand up a dcm4che system, and q/r everything over from the legacy environment:

1. How do we make sure that we scale the servers appropriately to support close to 1000 Q/R's per day?
2. How reliable are the services on an enterprise scale? Q/R over high latent WAN links?
3. What sort of training/support is there available for my team so that they can provide appropriate levels of support? 4. What really is the true cost of supporting a system like this vs having a vendor to rely on that may cost more on the surface?

I would really appreciate any advice or feedback.

Sincerely,

Pete

[gunterze]

You may contact my company for professional services for dcm4che archive 5 by http://web.j4care.com/dcm4chepro

gunter

[rrose...@securerad.com]

I'd never step on gunter's toes, but

A) I can't speak about arc-light but....

1. Not much to scale once the data is there. Dmc4chee (2.18.x) is more memory dependent than processor. Your choice of DB will be important. You don't need a $5,000 server, but a Raspberry Pi won't do either. Finally, your network is the issue more than anything else.

2. Extremely, but latent network meaning what? Since you say WAN, do you mean the server connection to the internet is slow, or whoever is Q/Ring? Also, to what scale? How much does a car cost?

3. Get in touch with Gunter. Other than reading the boards like everyone else, you can choose to pay, or not.

4. Good hardware, with the appropriate OS, with the appropriate knowledge and you'll be fine.

B) Some questions for you to think about......

1. How are you planning on migrating the existing data.

2. You really have 1000 Q/R /day over bad WAN connections?

2. Costs are relative. In the end someone is going to pay. The question is....who?

Good luck,

Rich

[Matthew Hill]

With that much data, you should also ask whether you should go to object storage, either direct usage or through an S3 interface.

Will make managing "bad" drives in the future a whole lot less messy.
Castor and S3 interfaces are available in 2.18.x

--
You received this message because you are subscribed to the Google Groups "dcm4che" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dcm4che+u...@googlegroups.com.
To post to this group, send email to dcm...@googlegroups.com.
Visit this group at https://groups.google.com/group/dcm4che.
For more options, visit https://groups.google.com/d/optout.

[Pete]

Thanks for the feedback!

It might be possible to run off a VM if I can find some room - that might help with the scaling aspect.

What might also save a bit of traffic is if I can integrate a oviyam through an API call to save the overhead of a Q/R - they are currently integrated with a Q/R to an ancient version of eFilm.

There are a couple of toolkits available that may work for the migration - we have an old sweeper tool that could be scripted, or possibly initiate the send from the old system. Have to think through that.

[Pete]

For S3 - for privacy/security here, we need to keep things inside the network unfortunately.

[Matthew Hill]

S3 INTERFACE not S3 the Amazon storage service

-----Original Message-----
From: Pete <pete...@gmail.com>
To: dcm4che <dcm...@googlegroups.com>