Home » Databases » Sybase » ASE » Webservices and Sybase ASE/Replication Server Monitoring?

Webservices and Sybase ASE/Replication Server Monitoring?

At work, we have three monitoring systems, BMC Patrol (gawd I hate it), Nimsoft Nimbus and a custom built Perl/Web monitoring system.

  • BMC Patrol for Sybase is a heavy weight agent that sits on your server and consumes roughly about 15 to 25% cpu resources because of the methods they utilize to obtain the statistics. It is not easy to extend and often captures not enough of the data you care about and not enough of the stuff you do. The worst part of BMC Patrol is the front end. It is a Java applet that requires a specific version of java to be installed (and no other!) and Internet Explorer 6. The Java applet is buggy and can crash IE, forcing you to reboot. BMC is working on a new front end based on the AJAX technology suite, from what I’m told.
    • Luckily this monstrosity is watched by another group and is being phased out in favor of the following two systems.
  • Nimsoft’s Nimbus is primarily a System Administrator’s monitoring tool where monitoring of machines, routers, web servers are the primary focus. They do have agents for the major DBMSs that provide minimal monitoring coverage of the database.
    • The Nimbus front end comes in three flavors:
      • Windows application that allows for drag and drop counters, graphs, pie charts, etc. An excellent tool if you’re using a windows desktop.
      • Web front end. Design the web page using the Windows application. Very easy to use.
      • Web Services. Nimbus allows you to pull data directly from the Nimbus repository server using web services (SOAP based). This allows for custom monitoring solutions that could include complex rules that must exist (transaction log is 90% in tempdb with application X *NOT* consuming most of the space) before the DBA is contacted.. Another option would be for automated maintenance (e.g. reorg rebuild) to occur on a table if a certain job ran or if the data is badly fragmented in a certain definition of fragmentation. (You do know that there are many types of fragmentation that can occur within ASE don’t you?)
    • The main reason for this minimalist approach is two fold from what I can tell:
      • Nimsoft doesn’t have performance & tuning experts on staff that cover the nitty gritty details that a good DBA often requires to tune the DBMS.
      • Performance & Tuning is such a subjective art, Nimsoft leaves it up to the DBA to extend the Nimbus DBMS agents by telling it what to look for.
  • The third monitoring solution, which we rely heavily upon, is a set of spiders that crawl the various Sybase ASE servers obtaining specific information and storing the information into a central repository. We have Mason/mod_perl based web pages that pull information from the repositories giving us ‘at a glance’ monitoring. The benefit of this method is that we are only pulling the data we need, when we need it and most important of all, we control exactly HOW it is obtained.
    • For example, there are several ways to determine what processes are running within ASE:
      • Exec sp_who
      • MDA tables
      • Sysprocesses
      • SYBMON: who
    • Which method we use depends on the situation and the quality& quantity of information we require. If you don’t know what Sybmon is, don’t worry about it, it is a tool built into the dataserver/sqlsrvr.exe binary that is used by Sybase to obtain access to the memory structures of ASE. It is invaluable for those people that know what to look for but because of the various NDAs, we can’t share any of that information with the public. (Note that I am NOT a Sybase employee – I was one many moons ago but I got better 😉
Share Button

Leave a Reply

Your email address will not be published. Required fields are marked *

*
*

Facebook login by WP-FB-AutoConnect