XCP-ng pricing



  • So as XCP-ng is past the lift off stage Olivier is working on figuring out a pricing model for commercial support for xcp-ng by itself.

    The current pricing for XOA (premium) is $6000 a year, and reduced tiers of support from there for less features.

    XCP-ng is the fork to Citrix XenServer, and ideally you should have some support channel for production workloads. Community support is great when you have a large enough community that can assist you with any issues you might encounter, but for a fledgling project as this one I think it would be worth the spend for support from the developers.

    With that in mind here a few questions I've received from Olivier on pricing plans for XCP-ng (with paid support) and was wondering how you would answer them.

    My idea of realistic pricing is in code blocks


    • pricing will be per host and per year (subscription model), including all updates and support

    Knowing this, at what price per year and per host would you consider the product to be so expensive that you would not consider buying it?

    At $1000/host/year I wouldn’t purchase support as it’s just insanely expensive

    Knowing this, at what price per year and per host would you consider the product to be priced so low that you would feel the quality couldn’t be very good?

    $150/host/year – even though I would more think that the cost is cheap, not that the benefits are bad

    Knowing this, at what price per year and per host would you consider the product starting to get expensive, so that it is not out of the question, but you would have to give some thought to buying it?

    $640/host/year

    Knowing this, at what price per year and per host would you consider the product to be a bargain—a great buy for the money?

    $384/host/year



  • My general feeling of the pricing model that will come out at some point will be in the realm of what is being asked for XOA (up to $6000/year) which is out of this world and why I recommend people use the community version of XO whenever possible and if they don't require the support.

    Now my reasons for the pricing breakdown there goes with core counting. Olivier is considering a flat cost per host (which makes no sense to me).

    Massive organizations who build insanely large systems essentially would get support at little to no cost, whereas smaller customers will avoid support at all costs as the support might cost as much annually as their server equipment costs originally.



  • How would you guys value support for XCP-ng? Is my pricing insanely low, just about right or way to costly?

    I valued the support per core, and on the "just right" range per year of $348/host/year would be a 64 core system at $6/core.

    Obviously this price could change, but that isn't an abnormally large or small server either.



  • I don't think a strict per host system is a good idea. I think a per socket system is better.

    I tend to favor a tier-based support system, in which those tiers can cover x amount of sockets. (example 1)
    Or, a simple tier based system that provides different levels and depth of support. (example 2)

    Example1:

    • Tier 1 (basic) support: $400 / year
      • Support for up to 1 host or 2 CPU sockets
    • Tier 2 (mid-level) support: $700 / year
      • Support for up to 2 hosts or 4 CPU sockets
    • Tier 3 (enterprise) support: $3000 / year
      • Support for an unlimited number of hosts and CPU sockets

    Example 2:

    • Tier 1 (basic) support: $150 / socket / year

    • Tier 2 (mid-level) support: $190 / socket / year

    • Tier 3 (enterprise) support: $ 295 / socket / year



  • And if simplicity is the winner, no tier system, then some kind of market-y number around $200 per socket per year is good.
    Best support for all!

    Example:

    All support is $195 / socket / year.



  • @obsolesce so you would favor the existing support model that many businesses already do.

    Per socket or tiers of support.



  • @dustinb3403 said in XCP-ng pricing:

    @obsolesce so you would favor the existing support model that many businesses already do.

    Per socket or tiers of support.

    Well, if someone has a server with 4 sockets, and each CPU has 32 physical cores, and they are running some crazy system or number of VMs... that has the potential to be one hell of a support case.

    Not necessarily tiers, but it covers the kind of support you may expect.

    That everyone no matter what the use case is, should pay the same price of support, I think, isn't 100% fair.



  • @obsolesce said in XCP-ng pricing:

    And if simplicity is the winner, no tier system, then some kind of market-y number around $200 per socket per year is good.
    Best support for all!

    Example:

    All support is $195 / socket / year.

    I also support this method, even taking into account for what I said above... It's not expensive, but not so cheap it makes the product and/or support look bad.

    I think it's fair, and if/when we would run the entire infrastructure with XCP-ng, that's reasonable enough we'd pay it to cover all of our hosts.



  • @obsolesce said in XCP-ng pricing:

    @dustinb3403 said in XCP-ng pricing:

    @obsolesce so you would favor the existing support model that many businesses already do.

    Per socket or tiers of support.

    Well, if someone has a server with 4 sockets, and each CPU has 32 physical cores, and they are running some crazy system or number of VMs... that has the potential to be one hell of a support case.

    Not necessarily tiers, but it covers the kind of support you may expect.

    That everyone no matter what the use case is, should pay the same price of support, I think, isn't 100% fair.

    And that is my argument as well for the pricing I proposed.

    The customers that exist with massive servers paying for support would end up with a system like you describe would essentially get support at little to no cost compared to a well designed and balanced support plan.



  • @dustinb3403 said in XCP-ng pricing:

    @obsolesce said in XCP-ng pricing:

    @dustinb3403 said in XCP-ng pricing:

    @obsolesce so you would favor the existing support model that many businesses already do.

    Per socket or tiers of support.

    Well, if someone has a server with 4 sockets, and each CPU has 32 physical cores, and they are running some crazy system or number of VMs... that has the potential to be one hell of a support case.

    Not necessarily tiers, but it covers the kind of support you may expect.

    That everyone no matter what the use case is, should pay the same price of support, I think, isn't 100% fair.

    And that is my argument as well for the pricing I proposed.

    The customers that exist with massive servers paying for support would end up with a system like you describe would essentially get support at little to no cost compared to a well designed and balanced support plan.

    Right, a socket-based server pricing is fine for now, and could change in 5-10 years and go to core based...

    Really, core based is best NOW..... but the general population can't figure out how to do core based pricing for some reason, so I can see that being a put-off.


  • Service Provider

    @dustinb3403 said in XCP-ng pricing:

    At $1000/host/year I wouldn’t purchase support as it’s just insanely expensive

    This is not expensive.



  • @jaredbusch said in XCP-ng pricing:

    @dustinb3403 said in XCP-ng pricing:

    At $1000/host/year I wouldn’t purchase support as it’s just insanely expensive

    This is not expensive.

    It is if you consider the pricing that already exists from xenserver.org which is around $350 per socket.

    But my pricing is what I would think of expensive, give me your answers to the questions above.



  • @jaredbusch said in XCP-ng pricing:

    @dustinb3403 said in XCP-ng pricing:

    At $1000/host/year I wouldn’t purchase support as it’s just insanely expensive

    This is not expensive.

    When comparing to VMWare support pricing, it's far from expensive, true.

    But do you think XCP-ng wants to compete with VMWare support?

    We wouldn't use VMWare becasue it's WAY too expensive... so even at half the price, too expensive.

    Something like $200 per socket is fair... if it's too much, people may just go with VMWare or Hyper-V instead.

    VMWare's basic is like $2k per host per year. (for 2 cpus)

    $400 would be a lot cheaper, and in comparison very cheap. Where to start is a slippery slope.

    I could see $300 / socket / year... that's also very reasonable.


  • Service Provider

    @dustinb3403 said in XCP-ng pricing:

    @jaredbusch said in XCP-ng pricing:

    @dustinb3403 said in XCP-ng pricing:

    At $1000/host/year I wouldn’t purchase support as it’s just insanely expensive

    This is not expensive.

    It is if you consider the pricing that already exists from xenserver.org which is around $350 per socket.

    But my pricing is what I would think of expensive, give me your answers to the questions above.

    These answers are just worthless.
    Expensive is a relative value to all people.
    The cost of this service should be based on what the market will support.

    If the closest competitor is $350 per socket, then that means that you start there and figure out what differentiates your service from the competitor and if that makes it worth more or less.

    Then you go and calculate loss leader values based on attempting to enter a market and disrupt existing competitors.

    The thing you do not do is ask the fucking internet.

    Especially people with no business skills to tell you how much to charge for your services.



  • @jaredbusch said in XCP-ng pricing:

    @dustinb3403 said in XCP-ng pricing:

    @jaredbusch said in XCP-ng pricing:

    @dustinb3403 said in XCP-ng pricing:

    At $1000/host/year I wouldn’t purchase support as it’s just insanely expensive

    This is not expensive.

    It is if you consider the pricing that already exists from xenserver.org which is around $350 per socket.

    But my pricing is what I would think of expensive, give me your answers to the questions above.

    These answers are just worthless.
    Expensive is a relative value to all people.
    The cost of this service should be based on what the market will support.

    If the closest competitor is $350 per socket, then that means that you start there and figure out what differentiates your service from the competitor and if that makes it worth more or less.

    Then you go and calculate loss leader values based on attempting to enter a market and disrupt existing competitors.

    The thing you do not do is ask the fucking internet.

    Especially people with no business skills to tell you how much to charge for your services.

    Dont get pissy with me, I didn't create the questions.



  • @JaredBusch what do you think support from the devs would be worth?

    Do you think that the current plan (as far as I've seen) being a flat cost per host is a good way to go about it?



  • I can't see myself asking my boss for $1000 per year for something I likely wont ever use.

    I feel like a price per ticket or hourly rate is more realistic for this type of support, but at the same time a single ticket could easily cost $1000 if being charged per hour?

    I don't know. I've never paid for support unless it was bundled with a license that I had to purchase.



  • @bnrstnr said in XCP-ng pricing:

    I can't see myself asking my boss for $1000 per year for something I likely wont ever use.

    I feel like a price per ticket or hourly rate is more realistic for this type of support, but at the same time a single ticket could easily cost $1000 if being charged per hour?

    I don't know. I've never paid for support unless it was bundled with a license that I had to purchase.

    This is the rub for SMB. They pretty much "buy" updates and get free support. And enterprises (at least according to Scott - I have no first hand experience) Pay for support and may or may not get free updates.



  • @bnrstnr said in XCP-ng pricing:

    I can't see myself asking my boss for $1000 per year for something I likely wont ever use.

    And here is the rub - The XCP-ng guys can't survive on only open tickets.. they need to sell support contracts to ensure income. At least that's my guess.



  • @dashrender said in XCP-ng pricing:

    And here is the rub - The XCP-ng guys can't survive on only open tickets.. they need to sell support contracts to ensure income. At least that's my guess.

    Right, and the same exact thing applies to XOA, it seems ridiculously expensive for SMB, but is probably easily justified for enterprises. They're seemingly pricing it to land a few big spenders to pay the bills and they will continue to provide enough support in the forums to keep the small guys happy too.



  • @dashrender said in XCP-ng pricing:

    @bnrstnr said in XCP-ng pricing:

    I can't see myself asking my boss for $1000 per year for something I likely wont ever use.

    And here is the rub - The XCP-ng guys can't survive on only open tickets.. they need to sell support contracts to ensure income. At least that's my guess.

    But comparing this to the existing XOA pricing they're essentially trying to get every customer of theirs to purchase a Ferrari for each of em.

    I'm all for support, but at a realistic cost. The $6000 a year for the XOA features that I was using for free previously would immediately eliminate that solution in the SMB world.

    And while I understand that "support costs a lot" this isn't that support needs to be 24/7 phone and email with SSH access to a client.

    You'd easily be able to setup tiers of support.

    Tier 1 - Basic 8-5 hours phone & email support with 12 hour response window.
    Tier 2 - Standard 8-5 hours phone & email support with 4 hour response windows
    Tier 3 - Pro 24/7 hours " " with a 4 hour response window

    etc.



  • IE you would setup a per socket or core license costs and then apply a support level fee on top of it.

    This way everyone would pay the same per socket/core rate and get to choose the level of support that they want.

    You could even have minimums of support based on environment size requirements.

    1-4 hosts with less than 32 cores - Tier 1 support
    4-8 hosts - Tier 2 support

    etc



  • Going along the idea of the per socket/core and tiers if you opt to not add on higher support tiers you'd automatically fall into the Tier 1 category.

    Unless of course there were support minimums based on the number of sockets or cores.



  • Going on another point, have an emergency rate for people to opt to not purchase support at all. Simply have a flat "Call in rate" for the first hour.

    Something like $500 for the first hour and then $250 per hour after. Just pulling numbers out of the air.

    This would then highlight to the customer that support is key if there is a critical issue that costs more than what a support contract would cost regularly.



  • Or as a support tier there could even be a limit to the number of support cases per month or quarter or year as a whole.

    Tier 1 might be 4 tickets for a set time period
    Tier 2 might be double that number of tickets over the same period
    Tier 3 might be double Tier 2.

    There are so many ways to 1) reduce the cost it takes to provide support and 2) make customers find value in support so as to not abuse the support plan they've purchased.



  • And what's more, for customers who do purchase support and overrun their support allotment for the period there could be a per ticket fee that is charged.

    Or a reduced hourly rate for customers who need that additional support.



  • Which all of these additional fees might be trivial in the grand scheme, but it would greatly reduce the burden of providing support.


  • Service Provider

    @dustinb3403 said in XCP-ng pricing:

    Now my reasons for the pricing breakdown there goes with core counting. Olivier is considering a flat cost per host (which makes no sense to me).

    I agree, that would be nuts. They should learn from the big boys and model after them where it makes sense.

    Pricing should be based on capacity or use or something meaningful. "Number of boxes" is not meaningful. A "box" might be a quad core, 32GB nice desktop from five years ago. Or it might be a 512 core, 4TB monster. Same price for both? That's crazy.


  • Service Provider

    @dustinb3403 said in XCP-ng pricing:

    How would you guys value support for XCP-ng? Is my pricing insanely low, just about right or way to costly?

    I valued the support per core, and on the "just right" range per year of $348/host/year would be a 64 core system at $6/core.

    Obviously this price could change, but that isn't an abnormally large or small server either.

    The model is fine, but the pricing seems like. At $6/core you'd have small servers at $24 a year, that's not enough. Maybe $8-$10/core. Or maybe an eight core minimum because at some point, you just can't justify it.



  • @scottalanmiller said in XCP-ng pricing:

    @dustinb3403 said in XCP-ng pricing:

    How would you guys value support for XCP-ng? Is my pricing insanely low, just about right or way to costly?

    I valued the support per core, and on the "just right" range per year of $348/host/year would be a 64 core system at $6/core.

    Obviously this price could change, but that isn't an abnormally large or small server either.

    The model is fine, but the pricing seems like. At $6/core you'd have small servers at $24 a year, that's not enough. Maybe $8-$10/core. Or maybe an eight core minimum because at some point, you just can't justify it.

    Yeah I was just randomly throwing numbers in the air based on what is available list price for comparable support (xenserver.org).

    Up or down a few dollars / core wouldn't hurt in any way.