Process Mapping Logo

Process Mapping - Forums

Sharing 19 years of knowledge and experience

 
Metastorm BPM forums
Sign up Latest Topics
 
 
 


 
Poll Results
 
 Best option for high availability
 Run load balancer with 2 servers 0 0%
 Keep 1 server on standby and change dns 0 0%
 Other - Please post below 0 0%
Total votes: 0   Please or sign up to vote.


Reply
  Author   Comment  
JoeOmerta

Veteran
Registered:
Posts: 210
Reply with quote  #1 
I have 2 web servers both running engines.  Currently they're setup with a round robin dns.  I was wondering if anyone here has setup 2 or more servers on a hardware load balancer/appliance.  I know the general population says multiple engines/servers aren't needed, but this is mainly for a high availability site.  0 downtime is ideal where i work but i understand that's not always possible. 

I'm thinking of 2 options...

Run both servers all the time through a hardware load balancer, only run microsoft updates on 1 system at a time incase of failure.

OR

Run 1 server at a time with the second server on stand by and if one has any issues just change the dns entry until the original server is fixed.

Thoughts?
0
BMellert

Guru
Registered:
Posts: 688
Reply with quote  #2 
Not casting a vote since I can't confirm the best approach as we have only tried the one.

However, we also have a two engine/web server configuation with a hardware load balancer front-end for about 8 years without issue.  We have this configuration in case of fail over -- which thankfully we've never had to prove.  However, our round-robin balancing routine also affinities a user.  It hasn't been an issue thus far, but the affinity logic seems to favor one machine for users over another.  (I could be wrong about that though.  Same configution for V7 and V9, though we don't have much migrated to V9 yet to see how well that continues to work.)

(We are considering seapraet web layers for Asia and Europe since performance is poor for them, but that may be the future -- if it proves to enhance performance for them -- not what we've been using for 8 years.)
0
JoeOmerta

Veteran
Registered:
Posts: 210
Reply with quote  #3 
I was also thinking of just making a clone in VMWare on standby.  My concern about that is if the primary server goes down, i can shut it down, spin up the second one.  But in order to work on the original one, i'd need a different hostname/ip to prevent conflicts.

For your hardware load balancer, are you balancing by ports, services, etc...

If ports, which ports are you monitoring, if services, are you just monitoring the engine?  Does it monitor user load?

Thanks!
0
BMellert

Guru
Registered:
Posts: 688
Reply with quote  #4 
Our network guys have it balancing by simple round-robin -- by user connection I think is how it would be referred to.  I should talk to them to check if the service is running before directing users as we've had, on rare occasion, one service stall/crash on one server, but the server itself was running without issue.

Our current balance approach didn't detect the service being down since the service was up, so kept directing users to that server.  Was fun trying to figure out why some users were having issues while others were not until I figure out it was the service on one of the matchines.
0
Previous Topic | Next Topic
Print
Reply

Quick Navigation:


Create your own forum with Website Toolbox!