Comments (3)
  • Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

1 LMD commented Permalink

Hi Anthony, <br /> Thank you for sharing information about cluster size. <br /> I do not totaly agree with "There should be exactly two FRs for every cluster ". <br /> There is some situations where having more than 2 could be a good pratice. <br /> For example, a cluster who Qmgrs span on two datacenters, and with a kind oh HA between Qmgrs on different DC. <br /> I you put one FR in each DC, when one DC is offline, only 1 FR is left. <br /> And if during this time the server hosting this FR go down, your cluster will be is trouble. <br /> In this situation I thing having 4 FR (2 on each DC) carefully interconnected is a better approch than having only 2. <div>&nbsp;</div> Any comments ?

2 Anthony Beardsmore commented Permalink

Hi Luc-Michel, thanks for reading and hope it provides some useful pointers. Although there are very rare occasions where more two FRs genuinely adds value, in the vast majority of cases exactly two will be best practice. In the situation you describe for example, if any two of the four became unavailable, the cluster might not be in the consistent state you expect - accepting the slightly lower redundancy of a single FR in each datacenter would probably be preferable. This often catches people out which is why I mention I want to cover this in a seperate post another day. For now there is a little more information in the 'Best Practices' section of the Infocenter if it helps - http://publib.boulder.ibm.com/infocenter/wmqv7/v7r1/index.jsp?topic=%2Fcom.ibm.mq.doc%2Fzl00270_.htm -- Regards, Anthony

3 Anthony Beardsmore commented Permalink

As promised, see https://ibm.biz/Bdx7WR for a fuller discussion on having more than 2 full repositories.