I am having a discussion about the merits of sending data in various ways across an AppServer. Tom Bascom did some measurements on this thread https://community.progress.com/community_groups/openedge_development/f/19/t/17102.aspx?pi20882=3 that showed the relationship of the various payload sizes for TT, JSON, and XML packets, but I am wondering if anyone has done any measurement of the timing involved. I.e., if one sends a TT, the AppServer has to spend some time converting the TT to the transmission format, then there is time on the wire, and then there is conversion at the received end into the native form for the client in question. My guess is that the bulk of the end to end time is spent in the actual transmission, but I am wondering if there is any data.
Thread created by Thomas Mercer-HurshI am having a discussion about the merits of sending data in various ways across an AppServer. Tom Bascom did some measurements on this thread https://community.progress.com/community_groups/openedge_development/f/19/t/17102.aspx?pi20882=3 that showed the relationship of the various payload sizes for TT, JSON, and XML packets, but I am wondering if anyone has done any measurement of the timing involved. I.e., if one sends a TT, the AppServer has to spend some time converting the TT to the transmission format, then there is time on the wire, and then there is conversion at the received end into the native form for the client in question. My guess is that the bulk of the end to end time is spent in the actual transmission, but I am wondering if there is any data.
Stop receiving emails on this subject.Flag this post as spam/abuse.
-- Tom Bascom 603 396 4886 tom@greenfieldtech.com
I remember that thread. As I recall, the compression helped little in-house, but was meaningful over the slower WAN connection.
But, that is not the point I am looking for here. My "opponent" comes from an RTE world where every last bit of performance is critical and the code is in things like C. He thinks that since the code for AppServer sending a TT over the wire is necessarily generic, that it is going to be poorly performing. Of course, doing one's own conversion to a RAW in ABL would be massively slower because it would be thousands of ABL statements and I doubt the RAW would be much smaller than the TT packet based on your results. But, if I can show that the conversion of the TT to transmission is some small percentage of the end to end time since the bulk of the time is in the transmission, then I can show that the performance is either not poor as imagined or that at any rate it is fast enough to be essentially irrelevant.
Reply by Thomas Mercer-HurshI remember that thread. As I recall, the compression helped little in-house, but was meaningful over the slower WAN connection.
But, that is not the point I am looking for here. My "opponent" comes from an RTE world where every last bit of performance is critical and the code is in things like C. He thinks that since the code for AppServer sending a TT over the wire is necessarily generic, that it is going to be poorly performing. Of course, doing one's own conversion to a RAW in ABL would be massively slower because it would be thousands of ABL statements and I doubt the RAW would be much smaller than the TT packet based on your results. But, if I can show that the conversion of the TT to transmission is some small percentage of the end to end time since the bulk of the time is in the transmission, then I can show that the performance is either not poor as imagined or that at any rate it is fast enough to be essentially irrelevant.
Stop receiving emails on this subject.Flag this post as spam/abuse.
-- Tom Bascom 603 396 4886 tom@greenfieldtech.com