Server Clustering Explained

Posted on 2011-10-27
Last Modified: 2012-05-12
We have a centralised print server management system for 250 network printers.  We have had this on two virtual server (using vmware) and are suffering from poor performance due to the limitations we have on processing.  One server had sql server on it, while the other had the print queues and print management software.

They have recommended we have a server cluster and one node has the SQL server while the other has the print queues and management software.

Can you please explain server clustering and the hardware required and  how it would help?
Question by:WNottsC
    LVL 7

    Expert Comment

    From what I just read, it sounds like you have 2 separate physical servers, both running a VMWare Host OS.  In those VMWare hosts, you have a Virtual Windows server.  One Virtual Windows Server is your print server, the other Virtual Windows Server on the other physical machine.

    Now, someone (who are 'they'?) is recommending an 'active/active' clustering strategy between those 2.  One cluster actively will host the Windows print server, the other hosts SQL Server.  

    IF this is accurate, and please let us know if it is not and clarify, then my suggestion would be to ask 'them' if they ever actually did Active/Active clustering.  Do they realize that in fact, you have to have enough cpu/ram on each node to handle BOTH servers at peak usage?  If your print server fails, it will failover to the SQL Instance.  Your print server is already pegged at high CPU usage anyways, and if this failover happens during peak hours, will your SQL server be able to handle the additional work load?  So far, this sounds like a nightmare.  Not only will 1 service go down, but when it fails over, then both will go down.  Not good.  If you are set in stone to do Active/Active clustering, you need a 3rd node as a passive one at a minimum.  Not good architecture it seems.

    Please give us an update. Thanks.

    Author Comment

    I am sorry to be so unclear I am trying to get my head round it at the moment.

    A little more information:-

    We have an external print company in to help implement there Print Management software, as part of this we are having to make sure all our printers are networked (as you would expect) and that they are all using the HP Universal PS Print Driver.  This causes enough of a headache in itself, using the PS driver causes documents that used to print with the PCL driver to take along time to spool and at times not print at all.  These I think are a combination of printer server resources and lack of memory on the printers them selves.

    We are currently using VMware to manage out virtual machines.  We have 4 host servers at the moment each with 2 quad core processes and 64GB of memory.  These resources are then shared out among our 40 or so virtual machines.  We have a combination of web, SQL, files, and Application servers.

    As mentioned we currently have one virtual machine which has the SQL database for their software and another virtual machine for the management software and the print queues.  

    The company we have providing the print management software have suggested that the poor performance we are getting where the CPU usages of the 2nd virtual machines (print monitoring and print queues) are maxing out at 100% will be fixed by moving towards two physical machines in a cluster.  They have not talked about the type of cluster (active/Active or Active/ passive) just that one node would be the database server and one would hold their management software and I assume the Print Queues.
    LVL 7

    Accepted Solution

    Hi WNottsC,

    No problem, I fall into the same habits of forgetting some info myself when asking for help.  

    So if they are saying 'create a cluster using only physical servers, put sql server on 1 node of the cluster and the print management software on the other', then they mean an 'active/active' configuration.  

    If the SQL Server has no resource issues, and there's good ways of checking if there are or not, then there's no need to put it on the new physical cluster.  If the SQL Server is not resource intensive at all, then you could without worrying too much, but it sucks if you do, because you have to pay lisencing for both nodes and get enterprise (SQL Std works with 2 node active/passive clustering only, need enterprise for more and sql std clustering is full of troubles) which will cost tons of money.  

    I'd say this, if I got everything right:
    do a proof of concept.  Test their app on non-vm hardware, see how well their app works.  If it works very well then do active/passive clustering, save the licensing costs.   Does this company have some hardware they can throw your way to test it on?  Perhaps your company has some server they can use to test.

    Author Comment

    We are an education so luckily for us the licences are dirt cheap :)

    I am not sure on the SQLl server and if it is having any resource issues it may be fine to leave where it is and as you said have the print queues and software on an active/passive setup

    Thanks for all your help


    Write Comment

    Please enter a first name

    Please enter a last name

    We will never share this with anyone.

    Featured Post

    Looking for New Ways to Advertise?

    Engage with tech pros in our community with native advertising, as a Vendor Expert, and more.

    Requirements: root access via SSH, telnet, or other.. Alternately, access from the server administrator to run a counter-strike server, and the proper access rights to do so. Enough free disk space (and allowed to use this much, eg disk quota): 6…
    The 6120xp switches seem to have a bug when you create a fiber port channel when you have a UCS fabric interconnects talking to them.  If you follow the Cisco guide for the UCS, the FC Port channel will never come up and it will say that there are n…
    Get a first impression of how PRTG looks and learn how it works.   This video is a short introduction to PRTG, as an initial overview or as a quick start for new PRTG users.
    This video gives you a great overview about bandwidth monitoring with SNMP and WMI with our network monitoring solution PRTG Network Monitor ( If you're looking for how to monitor bandwidth using netflow or packet s…

    779 members asked questions and received personalized solutions in the past 7 days.

    Join the community of 500,000 technology professionals and ask your questions.

    Join & Ask a Question

    Need Help in Real-Time?

    Connect with top rated Experts

    18 Experts available now in Live!

    Get 1:1 Help Now