SAN FRANCISCO -- VMware didn't dazzle VMworld 2013 attendees with a vSphere 6.0 release, but incremental improvements...
in vSphere 5.5 address existing pain points and will allow IT to virtualize more types of workloads.
Scalability and performance improvements will help more companies venture into virtualization's last frontier: mission-critical and big data apps, VMware Inc.'s CEO Pat Gelsinger said during his keynote at this week's conference.
"We want to expand virtual compute to all applications. I want to emphasize 'all,'" Gelsinger said. "We're going to continue to drive compute virtualization until it's 100% virtualized in the data center."
News that vSphere 5.5 doubles many key scalability limitations, including support for up to 4 terabytes (TB) of RAM, 320 logical CPUs and 4,096 virtual CPUs per host, drew a rare applause from the 20,000 IT pros in attendance.
The maximum limit for VMDK files will see a larger jump from 2 TB in vSphere 5.1 to 64 TB in vSphere 5.5. While those scalability levels aren't necessary for most organizations, they lay the foundation for future projects and provide companies a level of insurance that mission-critical and big data workloads won't outgrow vSphere.
VMware also released vSphere Flash Read Cache in 5.5, first previewed at VMworld 2012 as vFlash. Flash Read Cache allows a host to use a local solid-state drive as a caching layer for virtual machines (VMs), moving I/O loads from a SAN to the local SSD, thereby improving performance.
Admins can even vMotion VMs using Flash Read Cache among physical hosts. Keeping with the 2X theme of vSphere's scalability improvements, Gelsinger said Flash Read Cache could double performance for virtualized mission-critical workloads.
New high availability features in vSphere 5.5
VMware also showed off updates to vSphere high availability (HA) capabilities.
Previous versions of vSphere HA could detect when a host or VM stopped responding and automatically restarted it. App HA uses VMware vFabric Hyperic to monitor supported applications inside those VMs and detect application failures, even if the VM continues to run. If it detects an application failure, it can automatically restart the application, or if an application fails to start, reset the VM. App HA and the Hyperic components it relies on will now be included with vSphere Enterprise Plus.
The second major addition to vSphere HA is VM-to-VM anti-affinity, which allows admins to set rules to prevent certain VMs from restarting on the same host. This could prevent, for example, an infrastructure from losing two key applications if one host failed.
"It's a big thing. Most people don't take that into consideration," said Scott Gottesman, VMware guru at Velocity Technology Solutions Inc., a cloud application service provider. "But if your network is going through a lot of changes, like mine, there's a lot of movement going on, and you need a way to deal with it without having to do it manually."
Single Sign-On gets an overhaul
While not as glamorous as many other changes in 5.5, the rewriting of vSphere's Single Sign-On (SSO) feature caught the attention of some IT pros and could spur adoption of vSphere 5.5 by 5.1 customers. Problems with SSO frustrated many VMware shops after it was introduced in vSphere 5.1.
"It was a bit clunky and a pain when we were first using it. I'm interested in finding more about how they finessed it and if it will actually work better in 5.5," said Bryan Becker, a server engineer for Cancer Treatment Centers of America.
Licensing and pricing for vSphere 5.5 will remain unchanged from vSphere 5.1.