Hi all,
I have a scenario that i'm starting to plan and architect, however I want to get someones input into this as I couldn't find any information on it on the forum or the mailing list. I think its a solution that will work, but I'd like to hear how other people might be doing this.
I have the following setup configured:
Live Environment
3 Node cluster with IPMI fencing and high availability (10GB Ethernet backend)
Backend storage: QNAP NAS 2U 879 Redundant Power with 10GB ethernet backend + RAID 10
DR Environment
2 Node Cluster with IPMI (no high availability) and 2GB Link Agg (802.3ad) backend
Backend Storage: QNAP NAS 1U 479 Redundant Power with 2GB Link Agg (802.3ad) backend
Firewall
Kerio Control x2 - DR Firewall configured with same rule sets as main live environment, however rules are disabled (except for basic ones such as HTTPS, SSH, RTRR (8899)
I've tested this on an internal network and it seems to work perfectly fine. The scenario that i have and would implement would be should the live environment ever fail, I would have the last synced data from the live qnap environment.
I have written a script which copies the contents of the /etc/pve/qemu-server and /etc/pve/openvz folders over WAN to the DR servers and places the files in the same location. The VM's appear on the powered down servers and in worse case scenario can be brought up online. I would probably script the server to copy the vm config files weekly and or even nightly since they are small and the RTRR can sync on the fly or at a specified time.
I am attacking this with proxmox in the same sort of way that VMWare does it. This isn't a joined cluster. The 2 clusters do not know anything about each other. The only thing that knows about each other are the storage arrays and copying data from live location to DR location.
How are other people doing this sort of thing? This type of deployment seems ideal to me, however some things might need to be resolved. Obviously if the DR site has to be brought up online, the firewall rules on the DR firewall need to be activated, and the external WAN DNS needs to be modified to point to the servers that have edge facing requirements (mail, ftp, http, etc)
The servers and storage at the disaster recovery site would have the exact same configuration as that of the live environment (same IP addressing, network subnet, DNS Servers, etc) for both the public and private network.
I look at the half million costs of doing this with VMWare and their associated storage and proxmox just seems to be so much simpler and cost effective. I can build a solution like this for far less and 30K, and thats with some fairly tricked out hardware, such as the QNAP Series line of servers and SuperMicro TWIN servers with 4+ nodes per chassis and onboard 10GB Ethernet.
Can anybody point it out to me that what I am doing is completely wrong and I'm stupid because of reason A. B. C. that you see as causing more grief than good? I've included an image of how I foresee this setup working.
Thanks for any help. Love this product
I have a scenario that i'm starting to plan and architect, however I want to get someones input into this as I couldn't find any information on it on the forum or the mailing list. I think its a solution that will work, but I'd like to hear how other people might be doing this.
I have the following setup configured:
Live Environment
3 Node cluster with IPMI fencing and high availability (10GB Ethernet backend)
Backend storage: QNAP NAS 2U 879 Redundant Power with 10GB ethernet backend + RAID 10
DR Environment
2 Node Cluster with IPMI (no high availability) and 2GB Link Agg (802.3ad) backend
Backend Storage: QNAP NAS 1U 479 Redundant Power with 2GB Link Agg (802.3ad) backend
Firewall
Kerio Control x2 - DR Firewall configured with same rule sets as main live environment, however rules are disabled (except for basic ones such as HTTPS, SSH, RTRR (8899)
I've tested this on an internal network and it seems to work perfectly fine. The scenario that i have and would implement would be should the live environment ever fail, I would have the last synced data from the live qnap environment.
I have written a script which copies the contents of the /etc/pve/qemu-server and /etc/pve/openvz folders over WAN to the DR servers and places the files in the same location. The VM's appear on the powered down servers and in worse case scenario can be brought up online. I would probably script the server to copy the vm config files weekly and or even nightly since they are small and the RTRR can sync on the fly or at a specified time.
I am attacking this with proxmox in the same sort of way that VMWare does it. This isn't a joined cluster. The 2 clusters do not know anything about each other. The only thing that knows about each other are the storage arrays and copying data from live location to DR location.
How are other people doing this sort of thing? This type of deployment seems ideal to me, however some things might need to be resolved. Obviously if the DR site has to be brought up online, the firewall rules on the DR firewall need to be activated, and the external WAN DNS needs to be modified to point to the servers that have edge facing requirements (mail, ftp, http, etc)
The servers and storage at the disaster recovery site would have the exact same configuration as that of the live environment (same IP addressing, network subnet, DNS Servers, etc) for both the public and private network.
I look at the half million costs of doing this with VMWare and their associated storage and proxmox just seems to be so much simpler and cost effective. I can build a solution like this for far less and 30K, and thats with some fairly tricked out hardware, such as the QNAP Series line of servers and SuperMicro TWIN servers with 4+ nodes per chassis and onboard 10GB Ethernet.
Can anybody point it out to me that what I am doing is completely wrong and I'm stupid because of reason A. B. C. that you see as causing more grief than good? I've included an image of how I foresee this setup working.
Thanks for any help. Love this product