Great video, this is the first video I have seen that went into detail on how to set up the MIO .... very good and thanks for your video. I think I will look for more to watch here on your site.
Good video. Of course, you would be using two VMs from two separate VM hosts. the high-availability and redundancy feature is shot to hell if you do this on a single VM host. We need to be able to completely lose a server and still have the database running. Good classroom presentation!
A storage (third server) is indispensable for this function to work? Can't I use a disk attached in one of the nodes or both nodes to directly save the files? For instance when a user access a shared folder and save fles, it will save in a disk of node 1 and then write the same data in disk of node 2 (I know DFS role does this, but I wanna know if Failover Cluster does something similar so I can learn how to configure it)
Thanks for the video! I am going to be using a Synology NAS for my storage in a RAID10. I am assuming MPIO is only needed if you storage has multiple IP addresses on the network? At this point, each of my servers only has one link and the NAS only has one link (network connection). I am working on the networking part, but want to get the cluster up first. Thanks in advance!
Correct, it's not just the 2 IP's (since you can technically assign 2 IPs to a single interface) but multiple physical links are needed for MPIO, otherwise you can leave it out.
I am really struggling with getting clustering working. I just followed your video and it's great. I got it to work the way you listed. However, I do have a SAN to use. I actually have two SANs. Do I need to use iSCSI Virtual Disks at all or am I added an extra unnecessary step? I am really struggling with this and could use some clarification.
The goal is to get the same storage presented to multiple nodes. iSCSI is just how I connected the storage to the servers in this video. If you had a real SAN you would typically use HBAs/Fiber in place of iSCSI.
No if dev-stor1 goes down, the storage is offline and the files are inaccessible. It is a single point of failure in this case. As long as the storage is up, then either one of the TDB nodes can go down and everything stays up.
hey guys in this scenario what im trying to get is that when u configured this and for testing purpose when u shutdown a host, wil it automatically start up on the other hose as a replica server?
Yes that is the general idea, when one host becomes unavailable the services fail over to the other node and start assuming the storage is still available.
Hello This is Rakib Hossen. I Create A iscsi target On a supermicro Storage server . Storage Size is 60TB after Raid 6. * When I create a iscsi volume for iscsi target that says the size of the volume must be 8MB to 16 TB . Why that. ( however I solve that to extend virtual disk) * After I create iscsi target I give access two client, one Client-1 , Client-2, via iqn. I successfully connect and mount volume booth client There is no issue to connect. ( Booth client and server is windows server 2012-R2 licence version ) But when I update or create file or folder on the mounted volume from client -1 it transfer done but when I see same volume from client -2 there is empty, there is no file or folder. Then I tried to send some file of folder from client -2 it's transfer fine, but I can't see those file from client-1 that I send from client -2, Still see old file that was I sen from Client-1 What is the problem please help me how can i solved it?
In the case of this demo, that is what I did. This works fine for testing or in a lab environment. Ideally all 3 would be on separate hardware but the storage node would still be the single point of failure.
@@robwillisinfo Thanks for the info, I am currently trying to set this up on my own home lab. I have to say you have kind of inspired me with your channel and I am doing the same at the moment. I had one question though, have you ever tried setting up the virtual target to have direct access to the disks? I am currently experimenting with ESXi to give the virtual system so that I can get around the extra layer of the datastore to help increase performance.
That's awesome man! This is mainly the way I have done this other than using an actual SAN, but you can give the vm's direct access to the disk if all the vm's are on the same physical host if that is what you mean. You might see a bit of a speed increase in doing it this way, especially if you are using a hardware raid. Windows software raid is not the fastest. :)
Wow, that's a lot to take in, but very detailed and clear. Awesome video, you the only one that I could find that was cut and dry, to the point.
Added to the useful playlist.. Couldnt figure out how to add virtual storage to the cluster - this helped out a lot. Cheers!
Great video, this is the first video I have seen that went into detail on how to set up the MIO .... very good and thanks for your video. I think I will look for more to watch here on your site.
I thought this was a very nice presentation - thank you for the hard work putting this together.
Good video. Of course, you would be using two VMs from two separate VM hosts. the high-availability and redundancy feature is shot to hell if you do this on a single VM host. We need to be able to completely lose a server and still have the database running. Good classroom presentation!
Very good vid. This is exactly what I have been looking for. Thanks.
Excellent, this is exactly what I was looking for. Thanks!
hi Rob, thank you for your help !
Great video, Keep up the good work
perfect howto, thank you for the effort!
this one is awesome
Really great video. Thanks.
A storage (third server) is indispensable for this function to work? Can't I use a disk attached in one of the nodes or both nodes to directly save the files? For instance when a user access a shared folder and save fles, it will save in a disk of node 1 and then write the same data in disk of node 2 (I know DFS role does this, but I wanna know if Failover Cluster does something similar so I can learn how to configure it)
What If I have only 2 nodes, is it okay to use the disk space of Node 2 and connect the Node 1
using iSCSI connection?
Thanks!
Did you already install clustering at the db1 and db2 servers? If so, how? thanks.
Thanks for the video! I am going to be using a Synology NAS for my storage in a RAID10. I am assuming MPIO is only needed if you storage has multiple IP addresses on the network? At this point, each of my servers only has one link and the NAS only has one link (network connection). I am working on the networking part, but want to get the cluster up first. Thanks in advance!
Correct, it's not just the 2 IP's (since you can technically assign 2 IPs to a single interface) but multiple physical links are needed for MPIO, otherwise you can leave it out.
I am really struggling with getting clustering working. I just followed your video and it's great. I got it to work the way you listed. However, I do have a SAN to use. I actually have two SANs. Do I need to use iSCSI Virtual Disks at all or am I added an extra unnecessary step? I am really struggling with this and could use some clarification.
The goal is to get the same storage presented to multiple nodes. iSCSI is just how I connected the storage to the servers in this video. If you had a real SAN you would typically use HBAs/Fiber in place of iSCSI.
sir i have a question. what if your dev-stor1 will down, what happen to the failover? is the files are still accessible? Thanks
No if dev-stor1 goes down, the storage is offline and the files are inaccessible. It is a single point of failure in this case. As long as the storage is up, then either one of the TDB nodes can go down and everything stays up.
hey guys in this scenario what im trying to get is that when u configured this and for testing purpose when u shutdown a host, wil it automatically start up on the other hose as a replica server?
Yes that is the general idea, when one host becomes unavailable the services fail over to the other node and start assuming the storage is still available.
Great Video!!!!!!!!!!!! Thank You
why do u have two nics in storage server? a detail explanation would be better.
The dual NICs were for the MultiPath I/O (MPIO) which provides fault tolerance and allows for increased performance by having multiple active links.
Does the main Cluster Manager reside on the Data server?
(or)
Does it go on the DB01 Server?
Never mind.
It has to be on the DB01 and DB02
Hello This is Rakib Hossen. I Create A iscsi target On a supermicro Storage server . Storage Size is 60TB after Raid 6. * When I create a iscsi volume for iscsi target that says the size of the volume must be 8MB to 16 TB . Why that. ( however I solve that to extend virtual disk) * After I create iscsi target I give access two client, one Client-1 , Client-2, via iqn. I successfully connect and mount volume booth client There is no issue to connect. ( Booth client and server is windows server 2012-R2 licence version ) But when I update or create file or folder on the mounted volume from client -1 it transfer done but when I see same volume from client -2 there is empty, there is no file or folder. Then I tried to send some file of folder from client -2 it's transfer fine, but I can't see those file from client-1 that I send from client -2, Still see old file that was I sen from Client-1 What is the problem please help me how can i solved it?
So are all three of the servers running on the same host? Or is that a bad idea?
In the case of this demo, that is what I did. This works fine for testing or in a lab environment. Ideally all 3 would be on separate hardware but the storage node would still be the single point of failure.
@@robwillisinfo Thanks for the info, I am currently trying to set this up on my own home lab. I have to say you have kind of inspired me with your channel and I am doing the same at the moment. I had one question though, have you ever tried setting up the virtual target to have direct access to the disks? I am currently experimenting with ESXi to give the virtual system so that I can get around the extra layer of the datastore to help increase performance.
That's awesome man! This is mainly the way I have done this other than using an actual SAN, but you can give the vm's direct access to the disk if all the vm's are on the same physical host if that is what you mean. You might see a bit of a speed increase in doing it this way, especially if you are using a hardware raid. Windows software raid is not the fastest. :)
What happens when the storage server unexpectedly shuts down
Everything blows up :) The disk will suddenly be removed from both machines and any services depending on the disk will fail.
I'm implementing this first thing in the morning
is there is any way to keep this these drives online while rebooting of iscsi target ,,server,,,,But it has drawback of maintenance time?