Ideas

HA in Rendering Servers

We need to proceed many renditions and guarantee real high availability. We'd like to set up Meridian to use a Rendering Cluster, meaning more than one render server is working at a time.  Would it be possible to get it?

  • Francesc Bastida
  • Jun 14 2018
  • Attach files
  • Admin
    Rudi Pieters commented
    18 Jun 10:49

    Is the issue high availability or high throughput?

  • Francesc Bastida commented
    18 Jun 11:13

    We consider both missing features.  Meridian seems to capable to continue converting to PDF in the event of failure of Rendering Server. And also, whenever there is a high rendering demand of large drawing, there is no option to increase throughput by adding concurrent servers.

  • Francesc Bastida commented
    18 Jun 11:14

    Mistake "Meridian seems NOT capable"

  • Aaron Swain commented
    05 Jul 03:15

    Hi Francesc Bastida, have you looked at the Enterprise Server Cluster and creating publishing jobs dedicated to each Enterprise Server in the cluster?

    I find that dedicating publishing jobs to a specific Enterprise Server in the cluster works well but only when the publishing load is split evenly between the publishing jobs. Enterprise Server Clustering would really benefit from some kind of internal load balancing functionality, where the publish queue would be intelligently sent to different publishing jobs depending on load of each publishing job.

  • Francesc Bastida commented
    05 Jul 14:00

    Hi Aaron Swain

    My little knowledge about this topic is  that each Vault's  configuration use only one Rendition JOB. And each Rendition Job only use one server of the Enterprise Cluster (only one) . Hence only one server will be handling  documents to PDF (Rendition Job). 

    Your idea seems feasible, maybe I missing something to fully understand. What do you actually mean with Several publishing jobs...?