Hospital network conditions for testing purpose

[yuvraj-narewade]

Hello Everyone,

I want to test my medical device which connects to the cloud sever through LAN and currently through our office lab network. I want to perform test by simulating the hospital network for various Downlink/Uplink speeds, latency, jitter and packet loss. Does Cisco have this information for normal, busy and worst case network conditions? 

Please share a link if there is required information. Your help will be greatly appreciated. 

Regards

Yuvraj

[Enes Simnica]

gDay @yuvraj-narewade Cisco doesn’t publish standard hospital network conditions, since performance varies a lot between facilities. However my G, u can simulate typical ranges by using network emulation tools, and applying ur own test profiles for bandwidth, latency, jitter, and or for packet loss......

And from my experience G, most hospitals operate on stable Lans, but u can model normal, busy, and worst-case scenarios based on industry guidance....

hope it helps, and pEACE!

-Enes

[yuvraj-narewade]

Thank you @Enes Simnica.

[Enes Simnica]

also @yuvraj-narewade check these links (also in these links u can find some labs and stuff....): 

https://www.cisco.com/c/en/us/solutions/design-zone.html

https://www.scribd.com/document/522219039/Exp-3-Design

https://www.researchgate.net/publication/380596990_Smart_Hospital_Network_Enterprise_Design_for_Medicine_City_Hospital_via_Packet_Tracer

https://medium.com/@akun.ferdiyanto/designing-a-scalable-hospital-network-using-cisco-packet-tracer-7da2f832fcf0

[yuvraj-narewade]

I will check these links out. Thanks!

[Ben Weber · Accepted solution]

@Enes Simnica makes a whole heap of great points, and has lots of good resources there.

I would like to add that, if possible, you should try to configure your local LAN switch with the same configuration as the hospital access switch that your device will patch into. Does the hospital use MPLS? Do they use multigigabit switches, or are they using older switching equipment? Do they use NAC?

You should also be testing the traffic that your device will generate, as this could influence the speed of the network immensely. Where is the traffic going - is most of it going back to the cloud server, or is it required to integrate with other clinical services located at site? 

If you would like to share more information on your specific use case (maybe via PM), I'd be more than happy to help (having worked on network integration for multiple hospital networks previously).

[Joseph W. Doherty]

BTW, although @Ben Weber poses questions like:

Does the hospital use MPLS? Do they use multigigabit switches, or are they using older switching equipment?

Possibly a much more relevant question is whether QoS is possible?  Without it, supporting needed service levels is a game of chance.  Better/faster equipment often just improves the odds in your favor, but they are usually still not 100%.  QoS, though, can often insure 100%, even on what may appear to be inadequate hardware capacity.

Personally, I don't have any experience working with a hospital networks (laugh, except for the time I installed the first ever LAN network at the Veterinary Hospital of the University of Pennsylvania back in '83, but I wouldn't count that), but I do have "some" network experience supporting VoIP, Video Conferencing, video streaming, etc., on small networks, to huge networks, local on-site to international networks, from fractional T1s/E1s to 100g.  Such apps, were often sensitive to bandwidth, latency, jitter and packet loss.

Also, BTW, understand, QoS, alone, is not a cure all for everything.  It may guarantee bandwidth, but it has to be there to guarantee.  Or, it can do wonders mitigating with latency and jitter, but nothing for distance based latency (like hosts, literally, on opposite sides of the world).

Besides figuring out what your device's service needs are, you may need to be able to recommend how to meet such needs using QoS (much as commonly done if adding VoIP to a network).

[yuvraj-narewade]

Hi @Joseph W. Doherty, some hospitals may have MPLS, older one may not have it. QoS might be implemented in big hospitals, but may not be in small hospitals or clinics. I am planning to use a WAN Emulator which I discussed with @Ben Weber offline. WAN Emulator will allow me to test various speeds, latency, jitter and packet loss. However, I am not sure what should be the typical (normal), busy and worst case values for these parameters. 

[Joseph W. Doherty]

Again, cannot say there's any particular expectations for speed, latency, etc., including what might be considered normal or busy.

Some often rate individual link loading "busy", or not, by some value of average link utilization.  Often, I've seen links running at 30%, or less, not considered busy, while links running at about 70%, or above, considered busy.  However, in my experience, average link utilization, alone, often isn't a good indication whether there's sufficient bandwidth to meet application SLAs.

In my experience, I have seen network applications still work, although often perceived working (very) poorly, with up to a 10% drop rate and/or one second latency.  Also again, I suspect it's up to you to determine what service level your device needs to work at all to up to working very well.  For example, if doing VoIP it has specifications for its network service need along with MOS scoring.

BTW, on the subject of MPLS, from an application service need view point, about the only thing it offers, that's often useful, MPLS SPs often can provide rudimentary QoS support, although sometimes with an additional fee.

[Joseph W. Doherty]

I don't believe Cisco, or anyone else, can provide a specifications for "normal, busy and worst case network conditions".  What you can sometimes find are application requirements concerning "speeds, latency, jitter and packet loss".  I.e., it's up to the network whether they can meet the application requirements.

For your testing purposes, a network impairment emulation tool should allow you to discover your medical device's service level requirements for its acceptable performance.

[Ben Weber]

@yuvraj-narewade Joseph makes some excellent points in both of these replies, especially regarding QoS, you should consider whether QoS is implemented at site and whether the traffic that your device will be generating/receiving requires a specific QoS configuration.