| || |
Hello, I have couple of questiosn about two different monitors: Oracle and network
1) What are the measurements are generally monitored for Oracle piece - please be specific!
2) Is there any online documentation on Network delay monitor? Again, what are the parameters generally monitord?
Thank you for response!!
Re: LR Monitor
I cannot comment on the oracle monitor as I have not used it much, but I can provide some insight on the network monitor. There are two pieces that provide input to the network monitor.
The first item is RTT. RTT (Round Trip Time) is a measurement obtained during the three-way handshake of an IP session. The connection record passed during an IP session initiation is always the same length, therefore it is easy to make a calculatuion for delay and throughput. You have bytes from the connection sequence and you have the amount of time required to complete the connection. Bytes/time is throughput.
As a modifier to this information traceroute information is also collected. Traceroute provides information on the paths take by traffic when connecting from client to server. You need to identify hosts that will be used to collect this network information. As a note about TraceRoute, it is based on ICMP protocol (like PING) as as such has the lowest priority of any of the IP family protocols. Routers can discard them and hosts can ignore them. Have you ever seen only one in four PINGs have a response, well ICMP was the reason. Also a lot of enterprises have begun to turn off ICMP on critical hosts in order to prevent Denial of Service attacks/Pings of Death from impacting the critical host and or network segment. Be sure and coordinate your work with the network/infrastructure administrator of your org or you may wind up with no results for your logging.
RTT can degrade as the network gets busier and busier but it is a very small piece of information with which to make a large decision. It also tends to break the model of "like a user" as the user does not just send one very small byte flow with a business process, but many flows back and forth. The size and number of flows can impact effective client throughput dramitically. If have seen one situation recently where the maximal throughput obtainable per client was 64K/client due to the continuous turn around of small byte flows being sent and received.
There are better network testing solutions, unfortunately they do not generally integrate well with loadrunner. One exception is Antara, which offers a hardware solution called the "flamethrower" for testing at layers 2-6 of the OSI model.
I do not have personal hands on with the Antara product, but what brings it to mind is a recent press release noting Antara's relationship with Mercury through the Open Testing Architecture (OTA) program
Anyone else have any thoughts on the Oracle Monitor?
iTest Solutions, Inc
Replace ineffective offshore contracts, LoadRunnerByTheHour
. Starting @ $19.95/hr USD.
Put us to the test, skilled expertise is less expensive than you might imagine.
Twitter: @LoadRunnerBTH @PerfBytes
Re: LR Monitor
Thank you James!