How does the Byte code instrumentation work?
I am note sure that I am venturing into trade secret area here. If I am, let me know....
A bank have bought 17 JAVA agents. There was a huge debate about Collectors. In essence, they want to know how the agent and collector interact.
If we say that "dynaTrace Collector is responsible for the instrumentation of the dynaTrace Agent" does that imply we send the java code in plain text to the collector, where it is instrumented and then back to the agent? As this might be a security risk if these are sent around on their network. If the collector is on the other side of a firewall (yes I know but they are trying to save on collector machines as they have 84 DMZ's) would this expose their code?
Also, should the firewall be opened bidirectionally or only from agent to collector and collector to server?
hi Anton
yes, the Java classes are sent as they are (i.e. the Java bytecode, NOT source code) to the Collector and back. if they don't want the classes to leave the machine/network, they'd have to put collectors on those machines/in those networks.
opening the firewall only from agent -> collector -> server should be enough, as the TCP connections are always established in this direction.
hope this helps,
Christian
