Details

    • Type: Improvement
    • Status: Open
    • Priority: Major
    • Resolution: Unresolved
    • Affects Version/s: None
    • Fix Version/s: None
    • Labels:

      Activity

      Hide
      menaka Menaka Madushanka added a comment -

      Thank you very much James. I'll go through them.

      Show
      menaka Menaka Madushanka added a comment - Thank you very much James. I'll go through them.
      Hide
      jamestaylor James Taylor added a comment -

      Hello Menaka Madushanka - welcome. I'm glad to hear you're interested. Take a look at the sub items. The main technologies involved are HTrace, HBase, and Phoenix.

      Show
      jamestaylor James Taylor added a comment - Hello Menaka Madushanka - welcome. I'm glad to hear you're interested. Take a look at the sub items. The main technologies involved are HTrace, HBase, and Phoenix.
      Hide
      menaka Menaka Madushanka added a comment - - edited

      Hello,

      I'm Menaka Madushanka a final year Computer Engineering undergraduate at Faculty of Engineering University of Peradeniya. I found myself that this project idea is very interesting for me. So I decided to work on this project.
      So, I'll be very grateful if you can give me some details about the idea and the technologies involved.

      Thanks and Regards
      Menaka Madushanka

      Show
      menaka Menaka Madushanka added a comment - - edited Hello, I'm Menaka Madushanka a final year Computer Engineering undergraduate at Faculty of Engineering University of Peradeniya. I found myself that this project idea is very interesting for me. So I decided to work on this project. So, I'll be very grateful if you can give me some details about the idea and the technologies involved. Thanks and Regards Menaka Madushanka
      Hide
      jamestaylor James Taylor added a comment -

      PHOENIX-1120 is for tracing the full life cycle of a statement, including client and all servers involved. PHOENIX-1452 is for metrics collection on a single box, focused on the client as I read it. I think they're different, but I think as we learn what metrics we want to collect, both those JIRAs will be effected.

      Show
      jamestaylor James Taylor added a comment - PHOENIX-1120 is for tracing the full life cycle of a statement, including client and all servers involved. PHOENIX-1452 is for metrics collection on a single box, focused on the client as I read it. I think they're different, but I think as we learn what metrics we want to collect, both those JIRAs will be effected.
      Hide
      jfernando_sfdc Jan Fernando added a comment -

      James Taylor I updated PHOENIX-1452 the other night with more detailed information on the kind of client-side metrics I could use to help me understand Phoenix resource utilization and performance in both production and performance environments. I think this overlaps with PHOENIX-1120. Perhaps we merge the two?

      Show
      jfernando_sfdc Jan Fernando added a comment - James Taylor I updated PHOENIX-1452 the other night with more detailed information on the kind of client-side metrics I could use to help me understand Phoenix resource utilization and performance in both production and performance environments. I think this overlaps with PHOENIX-1120 . Perhaps we merge the two?
      Hide
      jamestaylor James Taylor added a comment -

      My initial take on the priorities of this are:

      1. PHOENIX-1197 Measure the performance impact of enabling tracing. If tracing puts too high a load on a cluster, that determines whether or not we can keep in on in production.
      2. PHOENIX-1519 Determine the usefulness of tracing. Are we capturing enough information to be useful? Is it structured in a such a way to lend itself to analysis?
      3. PHOENIX-1433, PHOENIX-1191, PHOENIX-1518, PHOENIX-1115 Provide a SQL command to turn tracing on/off. These would get us to the point where we could enable tracing for a single connection. If tracing produces information about other queries running, perhaps that's ok if a) the load is low enough, and b) it's easy to query the tracing table to get the information you're after.
      4. PHOENIX-1241 Add typing to trace annotations. This allows the information captured in the tracing table to be "typed". Without this, doing any reasonable querying over the data (i.e. analysis) will be difficult.
      5. PHOENIX-1120 Provide additional metrics beyond wall clock time. What additional information should we attempt to capture with tracing?
      Show
      jamestaylor James Taylor added a comment - My initial take on the priorities of this are: PHOENIX-1197 Measure the performance impact of enabling tracing. If tracing puts too high a load on a cluster, that determines whether or not we can keep in on in production. PHOENIX-1519 Determine the usefulness of tracing. Are we capturing enough information to be useful? Is it structured in a such a way to lend itself to analysis? PHOENIX-1433 , PHOENIX-1191 , PHOENIX-1518 , PHOENIX-1115 Provide a SQL command to turn tracing on/off. These would get us to the point where we could enable tracing for a single connection. If tracing produces information about other queries running, perhaps that's ok if a) the load is low enough, and b) it's easy to query the tracing table to get the information you're after. PHOENIX-1241 Add typing to trace annotations. This allows the information captured in the tracing table to be "typed". Without this, doing any reasonable querying over the data (i.e. analysis) will be difficult. PHOENIX-1120 Provide additional metrics beyond wall clock time. What additional information should we attempt to capture with tracing?

        People

        • Assignee:
          Unassigned
          Reporter:
          jamestaylor James Taylor
        • Votes:
          0 Vote for this issue
          Watchers:
          6 Start watching this issue

          Dates

          • Created:
            Updated:

            Development