If you hadn't noticed, VMware announced vSphere 5 on 7/12.  The announcement was well handled and I enjoyed watching the first hour.  The follow-up sessions weren't working in my browser, so I wasn't able to attend any of those, but I ended up spending a lot of time on Twitter and VMware's sites looking into the new features and licenses.

There are two main topics I want to cover.  First I'll cover the new licensing scheme for vSphere 5, which has become a very hot topic of criticism for VMware.  Then I'll talk about some of the new features of this release.

Licensing

As part of the upgrade to vSphere5, customers will need to figure out a new licensing scheme.  A change in the licensing shouldn't come as a surprise, since core and processor speeds have taken out a major chunk of VMware's revenue stream from their core product, not to mention VMware's history of changing licensing with every release.

The new scheme continues the tradition of one license per socket, but now allows for unlimited cores in every edition and a new memory cap.  It's a lot more complicated than that though.  The cap is not for the amount of physical RAM in the host, but it's a cap on a new construct they call vRAM.  vRAM is the amount of RAM assigned collectively to the virtual machines.  The licenses will combine their vRAM caps at the vCenter level (or multiple vCenters when using Linked Mode) to create a pool that all VMs in the infrastructure can pull from.  One caveat is that different editions of vSphere (Standard, Enterprise and Enterprise Plus) will create different pools.  When architecting infrastructures, you'll need to now consider not only the number of sockets in all the hosts, but also the total amount of RAM utilized by all the VMs.

The vRAM caps layout like this:

Essentials, Essentials Plus & Standard - 24GB
Enterprise - 32GB
Enterprise Plus - 48GB

This cap ignited a fire storm on twitter, blogs, email lists and even the private vExpert forum over at the VMware Communities. The root of this storm is what appears to be an increase in licensing for larger environments.

I'm guilty of contributing to this storm.  But as a good community member, I wanted to back up my comments, so I started contacting my customers to get a handle on what the licensing changes will mean to them.  Based on my results, this will be problem for larger environments, but many of my customers don't fall into this category.  Working with real world numbers and theoretical numbers it looks like the vRAM cap based licensing will increase environments using 196GB or greater per host, environments with a large cluster of 128GB or greater per host or environments heavily overcommitted.  Of the customers I've pinged, only one of them is affected.  But I do have customers that are building clusters right now with hosts of 196GB, and this will be a concern for them.

My concerns with this change have lessened, but have not been eliminated.  The advantages of scaling up the hosts (the exact solution Cisco is betting on for UCS extended memory servers) has less advantage that it did in vSphere 4. The costs may still make sense to scale up a smaller number of hosts as opposed to buying additional hosts for a scale out solution, but those costs are way to variable for me to swing at right now.  The burden on partners like myself will be greater now to architect the number of hosts/CPUs and VMware licenses to create the cheapest solution based on the known facts.  In some cases, it may actually make sense to upgrade to Enterprise Plus to gain the higher vRAM cap. It is also a good push to minimize the RAM each VM is allocated.  If you're worried about having to buy additional licenses, I'd start my reducing the RAM allocation of your VMs first (VMware will happily talk to you about CapacityIQ).  An interesting new PowerShell script has been created to help analyze your situation.  Finally, don't be surprised if us partners become more pushy to do a capacity plan before making recommendations. :)

VMware is clearly making a push to more of a cloud mentality for it's licensing ("utility" computing in VMware's words).  I just don't think VMware's general customers are ready to manage their envrionment in that way.  I honestly don't have many issues with licensing this way, but it's hard to wrap your mind around it and the vRAM caps are most definitely too low.  It also completely underminds the overcommitment advantages VMware has touted as one of their biggest advantages over the "good enough" club (Microsoft Hyper-V and Citrix XenServer).

My final thought on licensing (for this post at least) is to point out one big irony.  Back when vSphere 4 was released, a firestorm was created due to VMware's planned elimination of the Enterprise tier.  They introduced the Advanced tier, but it wasn't enough to justify the huge jump in cost/features that Enterprise filled.  Customers and partners both screamed and convinced VMware to keep the Enterprise tier.  The irony: Advanced has been removed in vSphere 5 and those licenses will be converted to Enterprise.

New Features

Now for the good news:  vSphere 5 introduces a ton of great new features (another expectation we have from VMware during each major release).  Here's a quick rundown of my favorites:

• A vSphere Desktop edition - This was quietly added on the partner SKU list for non-View VDI implementations.  This provides a low cost hypervisor for XenDesktop implementations (a fairly common occurance).
• Storage DRS - Provides IOPS latency and disk space balancing across a grouping of datastores.
• vCenter Server Applicance - The long awaited Linux based vCenter Server.  It is delivered as an OVF and can quickly be brought online using an embedded database or an external DB2 or Oracle database.  It can also be integrated into AD.  It's not 100% feature parity with the Windows install, but a much welcomed step in the right direction.
• vSphere web client - Ability to access vSphere functionality from a web browser without installing the full vSphere client.  Though it is not 100% feature parity with the .NET vSphere client, it will be a great alternative when needing to complete a quick task while at a machine that doesn't have the client already installed.
• HA rewrite - Changes have been made to the HA product that eliminates the traditional primary/secondary issues.  They have also made enhancements to improve reliability and easier setup.
• Auto Deploy - Using PXE, you can boot your hosts across the network and apply configurations and host profiles to customize the host.
• VMFS 5 - A non-impactive upgrade (remember VMFS2->VMFS3?) will provide improved performance.  You can also create VMFS volumes greater than 2TB - 512 bytes.
• VM version 8 -The latest version of the VM hardware adds USB 3.0 and 3D graphics support.
• VAAI Thin Provisioning - More closely integrates thin provisioning to the array itself.

Fear, Uncertainty and Doubt, or F.U.D., is a term I seem to have really come to use a lot lately.  It’s a marketing approach that seems to transcend industries.  You see it in politics as one candidate tries to undermine his/her opponents (in politics it’s usually called mud-slinging), in the battles between the cable/Internet/phone providers (here in Omaha it’s between Cox Communications and Qwest Communications), in advertisements for cleaning products (what percentage of germs does Clorox kill compared to Lysol) and of course within our own beloved IT industry.

Recent examples of IT FUD (at least in the infrastructure realm) have included, but definitely not limited to:

• Microsoft/Citrix v. VMware –hypervisor and virtual desktop/application infrastructure
• Vizioncore v. Veaam – The especially heated virtualization backup and management realm
• HP v. Dell v. IBM v. Cisco – “<insert company name> is the best, purpose-built platform for virtualization”
• HP v. EMC v. NetApp – the ongoing storage wars

All of the above mentioned companies deserve their place towards the top of the heap in their respective piles, which of course leads to very heated debate.  This debate unfortunately tends to boil over into FUD territory.

Having spent time as both a customer and a partner (though not a vendor), I definitely appreciate one company telling me the pros and cons of both their product and their competitor’s product.  In fact, I would doubt the aptitude of a vendor who didn’t have a competitive fact sheet for each of their products.  Where FUD comes into play is when the comparisons are overly tilted, based on half-truths or flat out lies and presented to a customer as an unerring truth.

Let’s take the recent trend of the Tolly reports and similar vendor-sponsored “independent” studies.  I use quotes around the word independent due to the fact that the third party is receiving money from only one of the competitors.  That breeds an inherent perception of a conflict of interest, which in my opinion instantly taints the report whether or not one truly does exist.

It is also well known that when a vendor performs or commissions someone else to perform a head-to-head comparison, that the tests that will be run will favor the features of that vendor’s product.  As a nonspecific example, let’s look at a storage bake-off.  Vendor A sets up a performance test between its array and Vendor B’s array.  The load that is put against both arrays could favor the caching algorithm that Vendor A’s array uses, thereby ensuring which array will perform the best.

Another aspect of FUD is the constant hammering of a competitor’s flaws, while totally ignoring any advantages their competitor may have.  You see this in the ever popular side-by-side feature comparison tables.  Two of my favorites are the VMware View v. Citrix XenDesktop feature set battles.

Wow, View sure is more feature rich, isn’t it?

Hold on there, apparently Citrix offers me more unique features.

See what I mean?  Clearly each vendor is focusing solely on the negatives of their competition, and in some cases not highlighting their best features. Also notice that VMware favors the least impactive of the competitors.

Next is the outright lie or deception.  This clearly falls into the Dirty category.  This will always occur, but in any healthy community it should quickly be knocked out of the sky.  So much of the IT industry is based on facts and numbers, so this doesn’t happen much, but it is definitely resident within politics (ever heard of a dirty politician?) where the facts and issues fall more into shades of gray.  These shades of gray make it harder to use facts to counter claims, especially when people’s emotions are thoroughly invested.

Finally, I’d like to highlight some fun that can be had with FUD.  As an example, I’d like to point out Doug Hazelman’s post here: http://veeammeup.com/2010/05/fud-for-thought.html.  Sure he’s being blatantly competitive and even admits to FUD flinging, but he also isn’t pretending that he’s giving an unbiased opinion.  FUD can also lead to great debates like we saw between VMware’s Scott Drummonds and Citrix’s Simon Crosby.  There is also a trend on Twitter and on some blogs of individuals who can transcend the Kool-Aid and have fun with their respective employer’s marketing companies along with their competitors’.  The Twiiter jabbing between Chad Sakacc and Vaugn Stewart is a perfect and perpetual example of this.  Both give credit where it is due and use sound technical arguments when disagreements appear.  Outcomes like these can turn FUD into something that actually benefits the community as a whole, but take a special set of individuals and circumstances.

So what can we do to wade through the mire that FUD creates in our decision making process?  To me it’s always been a matter of using the purely marketing information as a guidepost; a way to decide where my time would be best spent researching a set of competing products.  If Vendor A says their product is better because they don’t rely solely on SATA disks like Vendor B does, then I know I need to spend some time with the facts, trusted blog sites, a spreadsheet and possibly a lab to determine if there is some merit to using a large array of SATA disks verses a smaller array of SAS disks.  Companies do things because they think it’s a better way to do them, it’s up to us to determine if it really is a better mousetrap.

I like the way Wikipedia states it: "To dispel FUD, the easiest way is to ask for details and then provide well researched, hard facts which disproves the details.”  That clearly was written to help dispel a competitor’s FUD, but how does a consumer dispel the FUD coming from both directions?  I suggest not depending on marketing materials and asking for details from both (or all) parties, and then provide your own research and hard facts to make a fair comparison.

Anyone trying to sell you something is guilty of some of these things, though some may be more trustworthy than others.  Just remember, you must use your best judgment to make an informed decision that you can live with and defend, just like you should be doing in politics.