SteelEye Technology, in Palo Alto, Calif., has put the finishing touches on a high availability clustering solution...
for VirtualCenter, VMware's management console, and is is including it as part of its new SteelEye Protection Suite for VMware Infrastructure 3, announced today.
VirtualCenter, per VMware best practices, typically runs on a single physical Windows server. While not catastrophic, losing VirtualCenter to a hardware, operating system or application failure would result in an inability to launch any new VMotion migrations, and result in lost reporting data, said Bob Williamson, SteelEye vice president of products.
The new SteelEye agent monitors the health of the VirtualCenter server and application, and if it detects a failure, performs an automatic recovery action.
VMware HA not enough
The SteelEye Protection Suite for VMware Infrastructure 3 also gives customers the ability to monitor and cluster ESX servers from within the console, as well as monitor applications running within an individual virtual machine (VM).
That brings a higher degree of availability than what VMware HA and Distributed Resource Scheduler provide, Williamson said.
"At first, we decided not to try and bring our console-level product in with VI3 because we assumed VMware HA would do what it could do," Williamson said. But it turns out that VMware HA is, in Williamson's words, "very limited."
"It can't detect a storage failure, it can't detect an application failure within a VM," Williamson said.
Based on a stripped down version of the erstwhile EMC/Legato Automated Availability Manager 5.1.2, VMware HA's main job is to monitor ESX's service console network interface card (NIC). If it detects a failure with that NIC, VMware HA automatically starts to migrate VMs, Williamson said. The problem is that not every network outage is necessarily the result of a hardware failure, resulting in what Williamson calls "false failovers."
VMware users have recently discussed these and other issues with VMware HA on VMware's own VMTN discussion board, in threads like VMware HA: limited or broken?
Other SteelEye features include wide-area replication and failover, and the ability to cluster together physical and virtual servers. The latter feature can dramatically reduce hardware expenditures for shops looking to do high availability clustering for disaster recovery, Williamson claimed. "Think about it – you can put a number of VMs on a single physical machine, since you don't expect to be running the VMs for a very long time, and when you do, you're probably OK with it running slightly less quickly at the DR site," he said.
SteelEye runs on both Windows and Linux, and is available starting at $1,500 per server.
Let us know what you think about the story; email: Alex Barrett, News Director.
Dig Deeper on Disaster recovery, failover and high availability for virtual servers