Hello!
I have a problem with the bandwidth of my network connection
Initially, I had a network configured with a bandwidth of 100 Mbps. And for a long time now, my network has been completely clogged every couple of seconds, but in general, this did not affect the performance of the system or did not affect it very noticeably
Over time, the situation worsened and the network became so clogged that the main server ceased to be detected in the network at all, which caused a transition to the backup server
Moreover, when the server worked alone without a backup, the network was not loaded for some time, but after about a day, the network was completely clogged again and the backup server stopped responding. When connecting the main and backup servers, the network became clogged much faster, in just a couple of hours. We were forced to use a backup server without the main one or the main one without a backup one
It was decided to increase the network bandwidth to 1 Gbps
We found that during data exchange between servers, the exchange rate reaches 400-500 Mbps approximately every couple of seconds, then drops again to 4-5 Mbps and after a couple of seconds again reaches 400-500 Mbps
At the moment, everything has been working correctly for more than a day with such network characteristics
I don’t quite understand what the main server can give to the backup server in such quantities every few seconds
We poll about 150 devices via modbus tcp, two opc da clients, about 20-25 vision clients are open at the same time, cold backup is configured
Which segment of your network is showing these speeds? Is it the vision clients connection? Is it the gateway network? Is it the modbus segment? I haven’t seen a server with 100mbit port on it since 2009, so I’m assuming you mean something other than the gateway network?
Both servers have gigabit ports, but the network was initially configured on an old 100 Mbit/s switch. When we noticed that our bandwidth was completely full, we rebuilt it on a 1 Gbit/s switch. My main question is: is this amount of data transferred between the main and backup server normal for our number of devices and clients, or did we configure something incorrectly?
Have you looked at your performance and logs in Ignition to see if anything is showing up there corresponding to these spikes? Are you doing any historical logging on these servers that's trying to keep them in sync also?
The spikes look to be occurring at regular intervals. In addition to checking the logging like @michael.flagler mentioned, check your client and gateway event scripts, specifically for any timer or scheduled scripts. Perhaps there is something there unexpectedly sucking up a lot of bandwidth.
we don't have gateway scripts, but we do have client scripts. One of them is a script with a 1000 millisecond timer: if there is a critical unacknowledged alarm, then turn on the soundtrack and several simple scripts that execute queries to our database. we also have an alarm journal and an alarm status table. we also have a relatively complex window for displaying trends: each user has their own unique set of penses (20-30), which is determined by a query to the database.
In general, it seems like nothing that needs to be transferred in such volumes to the backup server
we don't have gateway scripts, but we do have client scripts
Definitely try to find what's happening during those spikes. In resource Manager look at what process is spiking the network. If it is Ignition then you'll have to dig further into what all you have Ignition doing because there's definitely something going on. Are you sure none of the database queries are pulling more data than necessary?
Is your backup server constantly doing full resyncs? (That can happen when you accidentally get a module mismatch or some corrupted configuration.) You may want IA to look over your shoulder (via a support ticket).
This is definitely ignition, and if we analyze tcp connections, then 90-95% of this is data transfer to the IP address of our backup server
We will try to analyze our sql queries again, but it seems that the entire project is really sent to the backup server, judging by the network load, the weight is approximately similar to our project, about 90 MB