FUD Me??? No, FUD You!!!

I hereby declare this will be the very last time I _ever_ post about the virtualisation FUD wars that are going on right now. What humours me the most is that Microsoft are acting as if they are equal to VMware both in product maturity and feature set, while declaring to be orders of magnitudes cheaper.

The simple fact is that the 2 platforms cannot be compared on a class level – at best they can only be compared as objects of the same class. The same way a Fiat 500 and a Ferrari F430 Spider are both cars, yet no one would be stupid enough to compare the 2 and argue that one was “better” than the other. Which one you may choose is entirely dependent on the use case and the budget. If you live in Inglewood and need a car to get to work in Compton and have to park on the street at both places, would you buy a Ferrari? No. If you live in Portofino and need something to get you to the office in Milan quickly a couple of times a month, will you buy the Fiat? Not if you had an office in Milan could afford to live in Portofino…

But moreso, if Fiat did take a swipe at Ferrari do you think Ferrari would actually respond? Of course not. The relationship of Microsoft and VMware should be somewhat like this. Hyper-V will surely be good enough for some use cases, even if it is merely to point out to your boss how much better ESX is, which is probably what 749 999 of the 750 000 downloads were used for (ooo, burn! But that would also explain why they hardly got any support calls, no?). ESX will clearly be more suitable for others, such as running production applications in enterprise datacenters. OK Hyper-V could do that too, but only if your app had massive redundancy built into it in the application layer, like Microsoft.com does. How many enterprise applications have the same kind of architecture?

When it comes down to it, our job is to determine what the use cases for any given piece of infrastructure are, and make an objective technical recommendation as to what to use where. Our employers pay us, our allegiances are to them – not software companies. There is no unfounded bias to one vendor or another, we make our choices based on sound research and higher intelligence. FUD me if anyone is going to make decisions like that based on marketing.


10 Responses to “FUD Me??? No, FUD You!!!”

  1. Joe__c Says:

    If someone is going to make a decision like that based on marketing then that’s probably not the only thing they’re going to screw up. And if they’re going to screw stuff up, I’d rather see them screw it up with Hyper-V.

  2. Stu Fox Says:

    “ESX will clearly be more suitable for others, such as running production applications in enterprise datacenters. OK Hyper-V could do that too, but only if your app had massive redundancy built into it in the application layer, like Microsoft.com does”

    So is your claim that ESX somehow gets around the requirement for providing high availability at the application layer? Just want you to expand on this point, because that’s how I read that and it’s an interesting claim.

    Disclaimer: I work for Microsoft, but this is my own opinion and not that of my employer

    • stu Says:

      Hi Stu,

      Indeed it does, it’s called Fault Tolerance – lookout for it in vSphere 4.0. Not only that, but as Microsoft IT said themselves in that case study, apps in a hyper-v cluster could have more downtime than those on a single piece of hardware. Are you really going to run production apps that don’t have application layer redundancy on infrastructure like that? Microsoft IT has no choice. Thankfully we do.

  3. Andrew Storrs Says:

    “But moreso, if Fiat did take a swipe at Ferrari do you think Ferrari would actually respond? Of course not. ”

    Actually they’d probably pick up the phone and call the head office in Turin and ask them what the hell’s wrong with the Media/PR team… Fiat owns Ferrari. 😉

    • stu Says:

      LOL – Bobsta, Andrew, you bastads lol. Where’s that sound research and higher intelligence when I need it eh?


  4. Bobsta Says:

    But Stu…
    Fiat owns Ferrari.
    Are you suggesting Microsoft might one day “own” VMware?!


  5. Stu Fox Says:

    As I understand VMware’s fault tolerance, it’s limited to single CPU guests – which is a limited subset (I could be wrong on that, but that’s how I read it). If you want to do FT at higher scale on VMware or Hyper-V there are third party solutions out there currently.

    But guess what? FT doesn’t solve problems with application failures. If it fails on one server, it fails on the other (they’re lockstep right?). All FT does is give you instantaneous protection against virtual host failure (which is a good thing), but it doesn’t do what you claim. Your app still needs redundancy, regardless of FT if you want to protect against app failure. FT has been around for a long time in the physical world and it’s not without it’s issues, something that virtualisation helps with but still doesn’t entirely resolve.

    • stu Says:

      @stu Let’s not confuse the issue shall we – we’re talking about failure of either the hardware or the hypervisor. Following the same logic, clustering is an equally useless technology because a cluster resource also has no application level redundancy. Is Microsoft now recommending that we don’t cluster anything because it doesn’t actually protect us from application level failure? No. FT, like clustering, is designed to overcome single node hardware failures. There is no shortage of examples of high availability technologies in the enterprise where applying the incorrect use case will fuck you over. Synchronous storage replication is great if your production datacenter spontaneously combusts but useless if your production data becomes corrupted. So be straight on what the purpose of it is, and don’t try to rubbish the feature by claiming it doesn’t overcome a problem it was never designed to address. It’s like saying bikes are crap because you can’t use them underwater.

      I agree FT will be of limited use in the real world (the single CPU is just one of many limitations that will rule it out for a lot of workloads), but hey you gotta start somewhere. By the time Microsoft have a similar feature (if they ever get there), you can bet SMP FT will be available.

  6. Craig Says:

    Hey Hyper-V is great – it’s far better than VMware Server free on my laptop…. Wouldn’t put it in a datacentre though! 😛

  7. Stu Fox Says:

    I don’t quite agree with you there Stu. Clustering provides application level failure protection (provided the underlying data set is not corrupt). For instance should you have clustered SQL and the SQL server process itself fails the cluster will ensure that the SQL process is started on another node in the cluster. Since the nodes are not in lockstep (i.e. not running FT) this can happen. However there is usually a slight outage in this process, which may or may not affect applications. That is application level protection. It is not immediate however.
    FT as you point out provides protection against hardware or hypervisor failure – and that’s it. I’m not rubbishing the feature (I pointed out that there are partners doing it on VMware & Hyper-V (and Xen as well) who scale better than a single processor) but it really is a reasonably limited use case scenario. This has been the case for years in the physical world as well. It’s usually expensive and requires nice fast interconnects. VMware’s FT may remove some of the expense (not sure how much).

    In terms of confusion, you claimed that VMware FT removed the requirement for app level HA, and then in the next breath claimed that it only protects against hardware or hypervisor failure. Which is quite different than app level HA.

    Cheers for your response, enjoy the conversation!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: