In this article, we eschew the normal blue-sky 2010 technology predictions for something a bit more everyday that will affect your daily virtual life.
Scale up, up, up and away
Firstly, it's no surprise that by mid-year there is likely to be a major rerelease of vSphere4 with a strong emphasis on increased scalability. Building on top of vSphere4's current scalability I wouldn't be surprised to see the number of vCPUs a single ESX host can support go beyond the 128 core range. I think it's likely that by the end of 2010 or the beginning of 2011 we will be looking at more than 8 vCPUs to a VM – with VMware pushing the amount of RAM per-VM into the 512 GB to 1 TB range and the ESX host supporting 1 TB or 2 TB of physical RAM.
I can imagine that in the future VMware will want to build on top of its existing Transparent Page Sharing technology to allow for compression of memory contents as well. Tied into enhancements on the vCompute side of the stack, I would expect to see similar boosts to vNetwork and vStorage pieces of vSphere4. This is all part of VMware positioning ESX as the defacto platform for large-scale cloud computing.
These enhancements might reside beyond the grasp of the average VMware-shop given that 8 GB memory is still quite cost-prohibitive, and historically VMware has placed the scalability features exclusively in the most expensive SKU available – Enterprise Plus. Despite this emphasis on scaling up vSphere4, most VMware-shops will remain anxious at very high consolidation ratios, preferring to distribute their VMs across more ESX hosts. Even with VMware HA and FT, there remains a deep seated anxiety VMware has yet to alleviate or eradicate, concerning the eggs-in-one-basket scenario that high consolidation ratios inevitably brings.
ESXi enterprise ready?
I predict that VMware will make changes to ESXi as part of the process to complete its commitment to depreciating ESX "classic" by the time vSphere5 arrives. So I would expect to see more industry recognized methods of deploying ESXi via a scripted installation -- which will be similar to ESX "Classic".
Additionally, to meet corporate data center standards, I suspect that ESXi will be enabled for Active Directory –- directly to the ESXi host rather than being limited to vCenter. It's something that customers have always asked for in ESX "Classic" installations where they have been forced to use PAM or WinBind. In the longer term I think VMware will be looking for completely stateless configuration for ESXi centered on boot-from-SAN or PXE boot. There are many experts in the community who have been doing work privately on this for sometime, but these in-house configurations lack the supported stamp of approval from VMware.
To make the switch over to ESXi bearable to the hardcore COS users I think VMware needs to have even better integration to command-line through the vCLI and vMA, to provide an interactive command prompt to those who demand it for so-called "edge" troubleshooting.
Bigger and better VMs
As I outlined earlier I suspect we will see not just more resources allocated to the VM. There are well-founded rumors of 32-way VMs with 1 TB of memory. I suspect that VMware will want to increase the features available to the VM. The idea behind this is VMware wants to push the concept of a 100% virtual data center wherever possible. The kind of functionality I'm thinking of is features such as USB support to the VM, and new virtual graphics controllers to support HD video inside a VM.
Essentially, I think these enhancements will be pitched at the VDI environments for a full multi-media rich experience. I suspect that these enhancements will require yet another upgrade of existing hardware for virtual machines (VMs). Currently, vSphere4 offers VM Level 7, and these enhancements will probably require an upgrade to the virtual machine hardware. This is a relatively simple task to do, but with Windows VMs it means a power off/on of the VM and proper maintenance window. Of course for some of these new devices to work, there would also need to be an upgrade of VMware Tools within the guest, and again in Windows this requires a reboot of the VM for the new drivers to take effect.
DRS for storage
At VMworld 2009, VMware announced that it was working on a method of moving VMs around to improve their performance from a disk I/O perspective. I think these enhancements are in their early stages, so I suspect the first thing you will see is a "shares-like" system for VMs where high priority VMs can steal I/O from low priority VMs. In truth this has existed for some time on individual VMs' virtual disks, but it's cumbersome to configure.
I also think VMware will probably introduce some method of categorizing ESX "data stores" into tiers of storage. This will make the configuration easier – as you generally have less data stores than VMs. I hope this is a configuration that will reside alongside Distrubuted Resource Scheduler (DRS) in the cluster level.
Personally, I don't think we will see DRS for storage really take off until 2011. By then the storage vendors will have had time to fully implement the copy-off-load advanced programming interfaces that vStorage offers. This allows the storage layer to move VMs from one tier of storage to another based on your preferences on the cluster. As ever I can imagine a bit of a tussle over who "owns" this configuration. Should it be the VMware guy who classifies the storage based on his knowledge of the I/O that VMs generate? Or will it be the storage guy who classifies the storage based on the fact that he created the volumes and knows what type of disk (SATA, Fibre, SDD) is used, the number of spindles (if any) and the RAID level used?
Coupled with this I believe VMware will develop a method that allows its "Linked Clone" feature found in View3 and View4 to be complemented by storage vendor-based linked clones. These are likely to surface in the Network File System (NFS)/iSCSI spaces where there has already been significant activity along these lines in the very early days of Virtual Desktop Infrastructure (VDI). Linked clones are used in many VDI environments to create one "parent" VM to spawn many instances, which are merely "delta" or merely store the difference from the parent. It makes creating many VDI VMs very easy and quick.
There's a whole lot more that I think VMware will be doing in 2010, and especially 2011. But I think that is probably the subject for next week's article. As ever, VMware maintains its commitment to being the most feature-rich virtualization vendor. However, the question remains whether good-enough virtualization might still win the day based on more prosaic issues such as pricing and licensing complexity.
Mike Laverick (VCP) has been involved with the VMware community since 2003. Laverick is a VMware forum moderator and member of the London VMware User Group Steering Committee. Laverick is the owner and author of the virtualization website and blog RTFM Education, where he publishes free guides and utilities aimed at VMware ESX/VirtualCenter users, and has recently joined SearchVMware.com as an Editor at Large. In 2009, Laverick received the VMware vExpert award and helped found the Irish and Scottish VMware user groups. Laverick has had books published on VMware Virtual Infrastructure 3, VMware vSphere4 and VMware Site Recovery Manager.