TechOnTip Weblog

Run book for Technocrats

Archive for the ‘Clustering’ Category

Microsoft iSCSI Target Cluster – Building Walkthrough

Posted by Brajesh Panda on May 3, 2011

In one of my last post I have discussed about Microsoft iSCSI Target Software 3.3. You can use this software to convert your (old) servers especially where you can populate a bunch of high capacity SATA/SAS HDD to convert the same to an IP SAN. Well this kind of scenario may be applicable for LAB/SMB business units where we don’t want to buy/spend lots of money for a specialized storage box. Well these days a lot of low cost SAS based DAS or IP-SANs are available in market with specialized storage features. So for usability you have to decide where you can use this kind of setup. Lets build a highly available storage, kind of active passive iscsi storage controllers ;-). Hope you are not getting confused with Building a MS Failover Cluser using MS iSCSI Target Software.

My Design Decisions & Blueprint

In my environment I used to have a Dell MD3000 SAS based direct storage box. I can connect 4 Servers directly without any path redundancy to this storage box & 2 Servers with 2-redundant path. But I have bunch of old (10) Dell Poweredge servers in my lab (generated out of dc consolidation project), I would like to repurpose them in my LAB & for some low priority low production purpose in my headquarter. Well none of my servers or devices is having warranty any longer. Renewing warranty or buying new hardware is out of budget. So I thought let’s do it this way…..

  • Apart from 1st 2 HDD I have pulled out all HDDs from Dell PE Servers & Populated remaining empty slots of MD 3000i. And made sure these HDDs are compatible with the storage box. Easy way of doing is install MD 3000 storage manager software & verify from the console software. I have discussed the installation & MD 3000 config step in one of my old post. It is here.
  • On server front I have decided
    • Two Node Windows Failover cluster with complete redundant path
    • I will utilize this cluster for Two Roles
      • Hyper-V Failover Cluster Role;
        With one Clustered Shared Volume to Host Less Critical LOB Application Virtual Machine
      • Ms iSCSI Target Cluster Role;
        with multiple clustered LUNs to Host virtual disks. These virtual disks will be presented to other Stand Alone Poweredge Servers through iSCSI Target using iSCSI technique. After wards I can use the same disk to host multiple virtual machines or I can use the same disk as a pass-through disk to your existing VMs. I prefer them to use as pass-through disks. . If I could use these Disks to host multiple Virtual Machine VHDs, there must be a lot of virtual translations i.e. VHDs inside -> VHDs inside -> NTFS partition.

Note: In next version of iSCSI Target I would like to see direct use of physical disk not VHDs inside NTFS partition. I could have also used those 8 Stand Alone Poweredge Servers to again create a hyper-v cluster using exported iSCSI volumes. But I never thought of that as a requirement because I will be using them in my test lab. Even if one server fails I will re-assign iSCSI exported virtual disk (LUN) to another node & start VMs over there. We can design all components in a better way to provide more performance like one VHD in one LUN, redundant network connections etc.

  • So to start with I have carved out 3 LUNs from the MD3000 & assigned to Node1 & Node2
    • Quorum LUN:         1GB
    • CSV LUN:        550GB
    • iSCSI Target LUN:    550GB
  • Network Design is one of the key components to achieve this; so I have decided on below NICs for Failover Cluster Nodes. I am flexible because it is a LAB & low priority production env.
    • Management & Virtual Machine Network: 1 NIC/node with X.X.X.X subnet
    • Cluster Communication & Live Migration Network: 1 NIC/node with Y.Y.Y.Y subnet
    • iSCSI Target Network: 1 NIC/node with Z.Z.Z.Z subnet (this is the adapter through which iSCSI Traffic will be coming in from Standalone LAB Hyper-v Server’s iSCSI initiators)

Read the rest of this entry »


Posted in Clustering, DELL MD3000i, Hyper-V, IP-SAN, WindowsStorageServer2008 | Tagged: , , | 6 Comments »

Best Practices for Exchange 2007 CCR Stretched Cluster

Posted by Brajesh Panda on March 8, 2010

In course of my SI Projects I found these are should be the best practices for a Exchange CCR Stretched Cluster. Let me know if anything I should include in my list.

Decision Making Factors

  • Is automated failover is required in unplanned site outage?
  • How frequently you are facing situation like complete site disaster?
  • Where is the major user base located? Is at Prod Site or DR Site?
  • Do we have enough bandwidth between these two sites?


AD Sites

  • Both CCR Hosts must be in same AD Site

    N.B:    Typically, Active Directory communication is controlled based on the AD Site design. Since both datacenters are using the same AD site, there is a risk that servers and clients in 1 datacenter may be communicating to domain controllers in the secondary datacenter. Cross authentication may leads slow response in Directory Server Access!

    –    Plan two domain controllers in each site


  • Encase of Windows 2003 Server we have stretch the subnets across datacenters
  • Windows 2008 Stretch cluster can span across different subnets; however we have to consider below points;
    • Latency must not be more than 50ms
    • Two different subnets should be available for Cluster Private & Public Communication
    • Read the rest of this entry »

Posted in Clustering, Exchange2007, MsClustering, MsFailOverClustering | Tagged: , , , | 1 Comment »

Before you start installing SQL 2005 Cluster

Posted by Brajesh Panda on March 8, 2010

Here is some best practices you should aware of before you start installing Windows Cluster. I will be keep updating this list….. Let me know if I am missing something most important.

Applicable Product: Windows Cluster & SQL 2005

  • Make sure you have installed Advanced/Enterprise/Data Center Edition of Windows OS on all servers
  • Deployment User account is added to local admin groups
  • Deployment User account must have all Active Directory permissions to create Computer accounts for Virtual SQL Node
  • If you’re IT policy is preventing you to get computer account creation rights for deployment user account please go through KB articles how cluster computer accounts can be staged before you start installation
  • Verify Task Scheduler is running on all cluster nodes (For SQL Clustering)
  • Verify Windows Cryptographic Service is running on all cluster nodes (For SQL Clustering)
  • Verify NTLM Security Provider Service is running on all cluster nodes (For SQL Clustering)
  • Verify Remote Registry Service is running without any error on all cluster nodes (For SQL Clustering)
  • Verify you can access media location from other nodes by using UNC path (For SQL Clustering)
  • Apart from SQL Clustering you have to install software individually on all other nodes. Hence make sure all media you are using are of same version & cluster aware.
  1. SQL Clustering use Task Scheduler to install the SQL software on all nodes at the same time.
  2. Apart from SQL Standard Edition, you need Enterprise Edition software for Clustering. SQL Standard edition support only 2 Node clusters
  • Make sure you have the right software media
  • Better to keep handy SMSTrace tool to verify Installation logs encase of any problem
  • File Systems must be formatted as NTFS. Read the rest of this entry »

Posted in Clustering, MsClustering, MsFailOverClustering | Tagged: , | Leave a Comment »

%d bloggers like this: