UNSOLVED
D
dave_omnibizsol
1 Rookie
•
5 Posts
0
22638
January 9th, 2008 05:00
Active/Active Clustering on MD3000 and/or MD3000i
We are interested in setting up Active/Active clustering or load balancing for some web applications.
We would need the servers to have concurrent access to the same drive(s)/files.
Our applications constantly are updating their files and those changes need to be reflected on the other node(s) as close to real time as possible.
Is it possible to setup MD3000 and/or MD3000i to have the same drive/data accessible from multiple servers? If so what is entailed and where can I find some documentation?
We currently have 2 Active passive clusters setup with MD3000's but I would like to look into the possibility of utilizing both servers and potentially adding additional servers to scale as needed in the future. Any info would be GREATLY appreciated.
dave_omnibizsol
1 Rookie
•
5 Posts
637
0
Posted January 9th, 2008 13:00
Windows 2003 Server Enterprise.With MSCS. NLB, or something else, are main interest is allowing us to get use out of the 2nd server, while allowing for redundancy, we are open to options.dining_philosop
60 Posts
637
0
Posted January 13th, 2008 15:00
MD3000 and 3000i support the ability to expose a virtual disk to multiple servers so that they have concurrent access. You should check the hardware owner's manual for information on how this can be done - this done by creating a host group and adding the hosts to the host group and assigning the virtual disks to the host group. That is the simple part. Now you want to ensure that the multiple (more than 2) servers that can access the same data do not end up over-writing each other. This is where host based software like clustering comes in - cluster software managed and coordinates access. MSCS/Oracle RAC/Linux HA etc are examples and your web application may have built-in clustering capabilities.dining_philosop
60 Posts
637
0
Posted January 14th, 2008 12:00
The controller has no issue in providing concurrent access if you map a virtual disk to multiple servers. The issue is that operating systems expect that they have exclusive access to the data that they see on attached disks. Unless you have an application (cluster) that is coordinating access between multiple servers, you could end up with unpredictable results. eg: What happens if a server tries to read data from a file that is being written by another server. You could have the case you may get inconsistent data. As I mentioned earlier the storage controller can be set up to have multiple host access data on it. It is up to app design to ensure that accesses from multiple app servers are coordinated so that data on the disk is consistent.dave_omnibizsol
1 Rookie
•
5 Posts
637
0
Posted January 14th, 2008 12:00
That's the thing, we already have some MSCS Clusters with a shared virtual disk, we want concurrent access so that is the app on server A makes a change to a file on the shared 'drive' that that change is seen by the app on the other server(s), and vice-versa, kinda like if they were all pointing to UNCs or mapped drives, any server can make a change and the next time any server reads that file they will read the modified data.dave_omnibizsol
1 Rookie
•
5 Posts
637
0
Posted January 16th, 2008 06:00
Will Veritas clustering work with the MD3000(i) ?tomjacobchirayi
4 Posts
637
0
Posted October 30th, 2011 05:00
I would like to know if the same scenario is possible for two hosts which are not clustered. The Scenario is like this. Both the servers are running Windows Web Server 2008 R2. Both these servers are being added to a host group and this host group is assigned the virtual disk. Now, the Same drive is visible in both the servers. Please note that these two servers are not clustered. But, the drive is visible in both the hosts. But, the changes made in one server is not replicated to the second server. Is there any way to achieve this? MSCS is not playing any role here. Is this possible?
Dev Mgr
6 Operator
•
9308 Posts
•
3 Points
637
0
Posted January 9th, 2008 12:00
The reason I say this is that one of the more common cluster solutions, like Microsoft's Cluster Services (MSCS), will not permit 2 servers to access the same disk at the same time (irrelevant of the storage solution).
In an MSCS you can only do an active/active by having 3 or more virtual disks/LUNs (1 for the Quorum and 2 for data) and use 2 resource groups and put each resource group on it's own server. This does mean that each server has it's own disk space (not actively shared). The cluster part allows the cluster group to fail over to the other cluster node in case of an intended or unintended outage.