Hello and welcome to the video on VX correlated statistics for VX for file data. There are statistics that can be collected in real time. These statistics will help identify the top talkers as we call them in your environment. So you can list statistics by client IP addresses NFS, user or group names by sifts, client names, sifts server IPS.
You can list out statistics by certain volumes. You can filter the output that you get on these, you can sort the outputs as well. There is the ability to control the polling frequency of the statistics that you're looking at and you also have the ability to automate some of these collections. So you can set up policies that will go ahead and kick off once the threshold is reached and run this real time stats for you.
You can even output these statistics to a CS V file in previous generations. It took a long process to identify your top talkers. You had to contact the storage admin, collect some logs, analyze those logs. Most cases you would have to reproduce and collect network traces of the issue that was going on upload the traces to E MC or have the network folks analyze them. And then you would get your identified top talkers and you could then deal with them with the correlated statistics using the server stats command.
We now have to just collect those stats, wait a few minutes, analyze them and then we can stop the top talkers. The automated collection, as I mentioned is constant monitoring. It's going to utilize the NAS alert or D Damon. You can set a policy and specific thresholds to monitor for and when that policy or threshold is reached, it will go ahead and kick off the service stats for you. The out of the box policy alerts are CPU utilization, memory, utilization and NFS operation. You can send email and or SN MP traps, the alerts S and MP tramps or emails are sent when the threshold that you have set is crossed for a certain number of minutes.
Now, this is to help prevent any kind of spikes to just sending alerts does have to cross for that certain number of minutes that you set and then you'll go ahead and get that alert or email or SN MP trap. Now, just to show a demo of some server stats, I want to go ahead and show the servers stats syntax and for more information, you're going to want to refer to the managing statistics on VX guide. So it's gonna have all the information you can do about statistics to monitor and so forth. But I just wanted to show a few here.
So there is the server stats command, you'll run, you'll run it against the data mover. You'll give it a count in an interval and then you'll give it a monitor and then a stat to monitor. This one is going to be NFS standard. Along with NFS standard. You can do a S standard. You see it's going to go ahead and give you out things like the active threads, sifts open files, the share connections, it'll give you the total IOPS that are running.
You can go ahead and run a SIFS dot client. This step path is going to go ahead and list out the SIFS clients that are doing reads and writes to the system. You can see the calls that are being made. Now you do get a summary every time at the end of one of these monitoring stats, you can put a dash termination summary space. No, and you won't see that summary at the end.
This is going to give you the same sort of information as the sivs. But for NFS, so it's gonna give you the client name, the average calls that are being made, read, write, total ops, that kind of stuff. You may notice that you have the names that are resolved to DNS that show up for the specific clients.
If you want to see the IP addresses, you can add a no resolve flag on the end of the command and you'll see it shows up with the IP addresses there. You can also combine statistics. You can see the NFS dash standard and SIF stash standard that I have output here and we combine all of them on the same screen. Now, one place this might be useful if you want to collect all these SS and NFS standard statistics would be in the case of formatting the output to a CSV file and output it to a different location.
You can do that with the format flag with an option of CS V and the file flag, which you then give it a file path and it will go ahead and collect all the statistics and just put them in a file. You'll notice once you format and put out to a file, you don't actually get the stats to the screen as they are put into the file, but you can cut that file and see the information in the CSV format. You can also sort the stats by different options.
This one will do a sort of an NFS client by right calls and you specify the order we're going to do descending instead of ascending and we'll do four lines. Now, we only have two hosts. So it's only going to show two lines. But you can see here that the operations are sorted by the right calls and the host that has more right calls will be on the top of each iteration. So here we'll go ahead and look at the store dot volume stat path for this one.
I'm going to put on the termination summary. No, because we are going to have a bunch of volume information that comes out. So here you can see we can see the total blocks, the red and written blocks and some other information. And these are all for specific D balls and files. You can see the file systems on these D vals as well. And that concludes a little look around the server stats command for some additional information on the statistics. You can take a look at the documentation on E MC online support titled Managing Statistics for VNX. I've also included a couple of knowledge based solutions here. If you want to look at how to gather network information like a network trace on the system itself and that concludes the video on correlated statistics.
Thank you.
