increase download speed, DICOM Export

[JC]

Hello, I need your help please. When downloading a .ZIP from the "DICOM Export" module, the download speed goes from 1.1 Mbps to 1.8 Mbps .... I need to increase that speed as much as possible. I have performed tests within the same local network and from outside but the result is the same, even install on another server DCM4CHEE-WEB3 and the result is the same, I do not know if something should be configured on the server, JBOSS, DCM4CHEE, etc. .

Server: CentOS7 See: DCM4CHEE-WEB: 3.0.5 Hoping you can help me, I say goodbye. regards

[Jon Ander Zuccaro]

Beyond using lossy compression, there is not much that you can do.

What is your hardware situation?

What are the reading speeds of your hard drive? Consider using a smaller but faster solid state drive as your online storage file system.

What is the nominal speed of your network interface card?

sudo ethtool eth0 | grep Speed




What is the actual speed of your whole network? It doesn't matter if your server has a 10Gb card if your routers, switches, client machines and even your cables can't handle it.

Use a tool like iperf to determine the actual connection speed between your server and client machines:

https://askubuntu.com/questions/7976/how-do-you-test-the-network-speed-between-two-boxes

Any kind of software optimization that you could do will crash with your hardware limits. I see you have 2K+ instances, so these are probably huge MR or CT studies.

Suerte!

[JC]

Dear, forget to say that downloading the .ZIP from the same server speed is the same, so I insist that it must be some configuration parameter ...

On the other hand at the time of receiving / sending DICOM files from the modes / work stations, I have no problems and achieve speeds over 60 Mbps

I hope you can help me

Thank you!

regards

[Jon Ander Zuccaro]

I understand... the way the download Zip Servlet works is by creating a zip stream to which it starts writing all files (images) from the study. It first performs a query against the database to obtain all paths from the files, and then reads them and writes them to the stream. It basically creates the Zip file "on the fly" if you will, so what you are seeing is not related to the download speed and has more to do with the processing power and read speed on the server's hard drive.

So, for a study that contains a large amount of images the bottleneck is really the 2K+ disk reads that it needs to perform... you are not downloading a single existing zip file of X size, the server is actually reading all images and creating a zip file "in real time".

Can you do the following test please: From a workstation, try to time how long it takes you to download a large study using a DICOM query/retrieve command, and I mean to download the whole thing: all 2K images, and then try to time how long it takes to download the same study as a Zip file from the GUI on the same machine.

Concentrate on how long it actually takes to download each complete study, not the stated speed.

In my experience, specially over the Internet, for large studies, the Zip file even though it is created in real time has always been  faster, but your case may be different, if it is actually taking WAY longer to download the Zip file then something is going on with your installation.

[JC]

Dear, thank you for the clarification of the operation! it is clearer and has much logic what you tell me. I will perform the tests ...

The download of the .ZIP files is for specific cases outside of working hours (the doctor reports emergencies ..)

Regards!