Hi,
We are in the process of building a small scale datacenter. Our setup will be 2-node cluster for now. We will have 2 dell poweredge 1950 III servers. We are planning to get vmware ESX 3.5 with HA. Both servers will be connected to a SAN (dell powervault MD3000i) via iSCSI interface. We will be running 2 VMs - IIS and MS SQL 2008 server, both running on Windows 2008 Server.
I have couple of questions specific to our setup:
1. Each server will have 2 on-board gigabit ethernet ports (one will be used for the vCenter and the other one for clients to connect to the web server). Additional 2 gigabit ports with TOE & iSCSI for connecting to the MD3000i SAN with redundant links. Do we need another gigabit port for the HA heartbeating?
2. From what I understood in the white paper the vmware HA doesn't provide hitless failover since it only restarts the VM on the standby physical server when the primary physical server is down. Do we need DRS+VMotion to achieve hitless failover?
3. Will the Broadcom 5708 NIC cards work in VM environment to take advantage of TCP offloading & iSCSI offloading features? Windows 2008 supports hardware-based iSCSI initiator.
Cold Standby, it probably an appropriate term.
Take a look at this article:
http://www.virtualization.info/2008/10/vmware-fault-tolerance-overview-and.html
This is supposedly a new feature in the next release
Jase McCarty, VCP, vExpert
Co-Author of VMware ESX Essentials in the Virtual Data Center
(ISBN:1420070274) from Auerbach
Please consider awarding points if this post was helpful or correct
Welcome to the forums!
1. Each server will have 2 on-board gigabit ethernet ports (one will be used for the vCenter and the other one for clients to connect to the web server). Additional 2 gigabit ports with TOE & iSCSI for connecting to the MD3000i SAN with redundant links. Do we need another gigabit port for the HA heartbeating?
No the heartbeats checks will occur over the Service Console port.
And you might want to take a look @ Texiwill's post on nic configurations:
2. From what I understood in the white paper the vmware HA doesn't provide hitless failover since it only restarts the VM on the standby physical server when the primary physical server is down. Do we need DRS+VMotion to achieve hitless failover?
To perform a "hitless failover", you would need to implement something like Microsoft Clustering Services. That is, until some newer features come out with the next release.
Jase McCarty, VCP, vExpert
Co-Author of VMware ESX Essentials in the Virtual Data Center
(ISBN:1420070274) from Auerbach
Please consider awarding points if this post was helpful or correct
Welcome to the forums!
1. Each server will have 2 on-board gigabit ethernet ports (one will be used for the vCenter and the other one for clients to connect to the web server). Additional 2 gigabit ports with TOE & iSCSI for connecting to the MD3000i SAN with redundant links. Do we need another gigabit port for the HA heartbeating?
No the heartbeats checks will occur over the Service Console port.
And you might want to take a look @ Texiwill's post on nic configurations:
2. From what I understood in the white paper the vmware HA doesn't provide hitless failover since it only restarts the VM on the standby physical server when the primary physical server is down. Do we need DRS+VMotion to achieve hitless failover?
To perform a "hitless failover", you would need to implement something like Microsoft Clustering Services. That is, until some newer features come out with the next release.
Jase McCarty, VCP, vExpert
Co-Author of VMware ESX Essentials in the Virtual Data Center
(ISBN:1420070274) from Auerbach
Please consider awarding points if this post was helpful or correct
How does the VM restart work? Does it resume from the last state before the failure occured (similar to doing a vm suspend and vm resume) or does it restart everything scratch like booting up the OS and services. If its the latter then the downtime could be long depending on how much time the IIS or SQL service comes up operational. This is considered as a cold standby. I don't see the benefit of HA in this case.
HA restarts the VM automaticaly on an other ESX-Host "from scratch".
Cold Standby, it probably an appropriate term.
Take a look at this article:
http://www.virtualization.info/2008/10/vmware-fault-tolerance-overview-and.html
This is supposedly a new feature in the next release
Jase McCarty, VCP, vExpert
Co-Author of VMware ESX Essentials in the Virtual Data Center
(ISBN:1420070274) from Auerbach
Please consider awarding points if this post was helpful or correct
look at it this way, your going to have downtime during a failure but HA helps reduce the amount of downtime that occurs. I personally haven't had to drive into work at 3am to restart a server since we went to HA because I know that the servers are avaliable and ready for people to start using when they come in in the morning.
Cheers,
Bradley Sessions
If you found this or other information useful, please consider awarding points for "Correct" or "Helpful".