Link to home
Start Free TrialLog in
Avatar of Bolt Burdon
Bolt BurdonFlag for United Kingdom of Great Britain and Northern Ireland

asked on

Help troubleshooting a Virtual Server

Hi,

We have one particular virtual server that hosts our case management system. The system is based on a progress database, which clients connect to via a 32-bit ODBC connection. All of a sudden, two weeks ago, the performance of this system dropped dramatically. Nothing has changed, like literally nothing.


The server has 32 gb of ram but never goes above 23gb usage, the processor sits around 35% usage. Pinging the server from within the office, we get an instant reply of 1 ms and remote replies of 10–30 ms. The server doesn't back up during the day but does replicate to another VM, which is also on the same site.


1.) Its been suggested that we look at the HDD read / write performance

 2.) Create a dedicated hyperv NIC with server i/o enabled and move the case management system onto that.


Points one and two I am not sure how to do it or where to start, so any help would be great.


I also wonder if and how I could use Wireshark to diagnose a network bottleneck.




Avatar of Seth Simmons
Seth Simmons
Flag of United States of America image

All of a sudden, two weeks ago, the performance of this system dropped dramatically. 


have you looked at i/o or system load?  any hardware issues?  have you restarted (host and/or guest)?  any other processes running at that time causing a lot of resource usage?

ASKER CERTIFIED SOLUTION
Avatar of giltjr
giltjr
Flag of United States of America image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of Bolt Burdon

ASKER

@Seth How would I look at the i/o or system load? 

Avatar of skullnobrains
skullnobrains

io load is available from the performance monitor or task manager in windows. on unices, you can use iotop, sysstat, gdisk ( solaris and bsd only ) and a few others.


wireshark will not or not easily help with network congestion. but you can look at black lines in the output which do indicate issues.


network congestion are better monitored using for example snmp counters and information from netstat of sysctl.


the first place to look might be deadlocks in the database.

Hi!

1) Its been suggested that we look at the HDD read / write performance

For a start, how about running disk diagnostics?


2) Create a dedicated hyperv NIC with server i/o enabled and move the case management system onto that.

You would need to "Enable virtual machine queue (VMQ)" and "Enable IPsec task offloading" options on the dedicated NIC before connecting the NIC to a separate virtual switch.