Notifications
Clear all
Topic starter
Does anyone have any input on performance tweaks? My main complaint that I hear is when users are in a patient chart and then "go back to the desktop". I can see that myself when I watch them but I have no idea what is going on behind the scenes and how I would improve that process.
I can guess:
1. Maybe data needs to be pulled from the db to populate the desktop?
2. Maybe the ERX server is being queried and that server is slower (it is in my case)
3. Maybe the Jboss server needs more HP or memory (I don't think it does in my case)
My setup:
1. I had the network blessed by a network guru yesterday. Gigabit thick/fat clients, new workstations. My iperf testing shows 700-900 mbit consistently, Cisco 3560G switching, 4GB etherchannel between each switch
2. Recently upgraded to a Dell chassis/blade setup, VMWare 5.1, 2 x Intel(R) Xeon(R) CPU E5-2660 0 @ 2.20GHz GB blades, 96 GB RAM (Jboss and SQL on separate blades)
3. Network cabling is either fiber or CAT5e certified, tested and professionally installed
4. Workstations are Core2 Duo Core i5 and core i7 processors, 6GB RAM, Windows 7 x 64
Running CQIC as the CCC type content which I have heard in the past that CCC slows things down. We are a tad behind in the version but that will be fixed soon. I do not hold out hope that the CQIC is the problem though.
We are on CPS 10.1.3.300 ( I know there is a newer version but we have to shut down the clinic and sneaker-net the client installs due to lack of MSI support)
Servers are Windows 2008R2 per GE recommendations, up to date with windows updates on servers and workstations.
I think I can rule out infrastructure since I see a performance problem at one specific place in CPS.
Any input is greatly appreciated
Mike Zavolas
Tallahassee Neurological Clinic
Posted : August 6, 2013 3:27 am
Mike,
What is the specs of your JBOSS server? When a user goes from Chart to Desktop, several jboss calls are made to make up the desktop components. If the JBOSS is not able to process those requests quick enough, users will see the spinning wheels of "Loading". The other thing to consider is daily reboot of jboss. Currently, we reboot twice a week and it has helped in performance.
-Amar
Posted : August 17, 2013 8:36 am
Topic starter
abulsara said:
Mike,
What is the specs of your JBOSS server? When a user goes from Chart to Desktop, several jboss calls are made to make up the desktop components. If the JBOSS is not able to process those requests quick enough, users will see the spinning wheels of "Loading". The other thing to consider is daily reboot of jboss. Currently, we reboot twice a week and it has helped in performance.
-Amar
Hay Amar, thanks for responding.
Running VMWare 5.1 vCenter, patches current as of two weeks ago
2 X 8 core CPUs allocated (E5-2660 @ 2.2ghz)
24 GB RAM
Storage is FC connected @ 4gbit, EMC Clariion SAN
rebooted daily @ 4:30am
I recently migrated to a Dell chassis with the blade servers listed above, previous hardware was recommeded GE as "the fix" but we have always had issues when users move from the chart to the desktop. I don't think it is the spinning loading thing as much as it is a delay with an hourglass.
previous hardware: X3450 @ 2.67Ghx Xeon (dual processor, 8 core). I am surprised nobody is admitting at least a minimal increase in performance but maybe it is no improvement at all. I should feel lucky that they aren't saying it is worse 🙂
Interesting comment from a doctor over the weekend. He couldn't get in to his desktop from home so I had him hit a VM instead. He said it was faster when going to desktop from chart. The only difference is the number of switch hops from his usual workstation and the server. He mentioned that he is comparing a usual Saturday to this Saturday so it is not because of less people in the system.
Mike
Posted : August 19, 2013 12:37 am
Topic starter
tnc said:
abulsara said:
Mike,
What is the specs of your JBOSS server? When a user goes from Chart to Desktop, several jboss calls are made to make up the desktop components. If the JBOSS is not able to process those requests quick enough, users will see the spinning wheels of "Loading". The other thing to consider is daily reboot of jboss. Currently, we reboot twice a week and it has helped in performance.
-Amar
Hay Amar, thanks for responding.
Running VMWare 5.1 vCenter, patches current as of two weeks ago
2 X 8 core CPUs allocated (E5-2660 @ 2.2ghz)
24 GB RAM
Storage is FC connected @ 4gbit, EMC Clariion SAN
rebooted daily @ 4:30am
I recently migrated to a Dell chassis with the blade servers listed above, previous hardware was recommeded GE as "the fix" but we have always had issues when users move from the chart to the desktop. I don't think it is the spinning loading thing as much as it is a delay with an hourglass.
previous hardware: X3450 @ 2.67Ghx Xeon (dual processor, 8 core). I am surprised nobody is admitting at least a minimal increase in performance but maybe it is no improvement at all. I should feel lucky that they aren't saying it is worse 🙂
Interesting comment from a doctor over the weekend. He couldn't get in to his desktop from home so I had him hit a VM instead. He said it was faster when going to desktop from chart. The only difference is the number of switch hops from his usual workstation and the server. He mentioned that he is comparing a usual Saturday to this Saturday so it is not because of less people in the system.
Mike
Well, just an update here. It looks like a "global refresh bug" which GE engineers are currently working on a fix for. I would expect this bug to be present in CPS 11 too.
Still working on this with GE Support
Posted : October 2, 2013 4:24 am
We are skipping CPS 11.
Posted : October 2, 2013 12:50 pm
Topic starter
abulsara said:
We are skipping CPS 11.
I wish we could do that.
Posted : October 7, 2013 3:51 am
