Tim:     Welcome to CGNET tech talk.  Today we have Ken Novak, CG Net senior technical advisor back with us for another discussion about pricing in the clouds.  Today we’re going to talk about comparing price and features among leading cloud vendors.  Maybe a place to start would be to talk a little bit about generally how the vendors compare in what they supply.  What do they have in common?

Ken Novak:    The central thing the cloud providers are offering you is just as you can spin up as many virtual machines as you want, you can have as big a disk store as you want, either for storing files or for storing whole computer images. And this is something that’s an important part of a cloud vendor because unlike a physical data center, you have the feeling that you’ll never run out of space.  You’ll never be able to fill up Amazon or Azure, and you won’t run into big performance problems that are designed to be scale free when it comes to storage.  And finally, you’ve gotta have a network surrounding this and connecting all these pieces to each other and to the Internet and there will be some kind of remote management to do things like set up security. Your connection to the Internet again is assumed to be scale free, essentially in an unlimited level in the gigabytes per second.  So again, you don’t need to be concerned.

So when they bill you, they’re billing you based on these three: computing, storage and network.  And they may bill the virtual machine by the hour or sometimes by the month.  They’ll bill the, total number of gigabytes you stored or maximum level that you stored in a given month.  And then finally, the amount of network that you use, measured by the number of gigabytes transferred on and off the Internet.  And when you first look up these prices, they’ll often look pretty similar from one vendor to another.

Tim:     So basically, although they’re not commodities and there are some differences among the systems which we’ll discuss later, the price may end up being a big factor in the decision, if there are ways that you can get better deals at one vendor than another.

Ken Novak:     That’s right.

Amazon’s Cloud Pricing

Tim:       Let’s take a look at the, at how it might break down with some of the major vendors.  Do you want to start with Amazon?  We had a a longer discussion about Amazon on a previous webcast, but let’s recap that a little bit for Amazon and what their situation is.

Ken Novak:     Sure.  I think the central thing about Amazon is that they publish all of their pricing and that they have some discount schemes.  Normally, you’re just asking for a virtual machine. You want to pay for it by the hour, pay for the storage by how much you use, and when you stop using it you pay zero.  That sort of on-demand pricing has sort of set the pattern for every cloud vendor.

The thing about Amazon, though, is once you start using them, you know you’re going to be running these systems for longer than just a few days or whatever.  You’re probably thinking in terms of months or years. And they offer discount schemes that match that. It’s typically 40 percent less if you make a long-term commitment of 12 months, even more if you’re willing to go to 36 months.  And they have other kinds of discounts, so-called spot pricing for short-lived systems under other conditions, but the central thing that you’ll work with is probably these discounts of typically in the 40 percent range.  That’s pretty steep discount but it is limited to the virtual machine charges, and it varies by type of virtual machine. So when you change from one type of machine, say a small one to a large one, you’ll run into issues on getting the value out of that discount. It’s something you have to work through.  They do have volume discounts on storage and on network, but they don’t run into the 40 percent range.  And in general, that’s probably not the biggest part of your charges anyway.  Probably the biggest part willll be the machines, so you really gotta look at your duration and the sort of commitments you’re willing to do on Amazon to figure out just how much you really pay.

Tim:     Right.  It might turn out that even though you’ve got a 40 percent discount on the virtual machine charges, it ended up being a 30 percent discount overall because the other stuff wasn’t discounted as much.

Ken Novak:     That’s right.  And, and you’re taking a risk that you’re not going to have to do a big wholesale change sometime.  If you make a 12-month commitment, then in month eight you decide you now need machines that are of a different type, maybe you need to run Red Hat instead of Lennox or maybe you need to have machines that are twice as big, then your commitments to the older machines are now not saving you any money.  So you have to play some games.  You might wind up not saving a full 40 percent if you do go through a big change.

Azure’s Cloud Pricing

Tim:     So that’s sort of the situation in a nutshell for Amazon.  How about Azure?

Ken Novak:     Azure’s got a much simpler system.  It offers discounts of 20 to 30 percent, typically in the 20 percent range, if you do volume purchases, and it applies to whatever you buy, whether it’s storage or virtual machines or whatever. Volume discounts start when you buy $500.00 per month with a six‑month commitment.  That’s a relatively low bar, and that gets you the lower rung of the discounts.  You can move up a few percent at a time by getting to larger monthly commitments, or by prepaying some or all of your changes or by looking at 12-month terms instead of 6-month terms. That’ll get you up closer to 30 percent.  It’s a pretty simple thing to understand. It’s just a straight top-line kind of discount.

Rackspace’s Cloud Pricing

Tim:     What about Rackspace?

Ken Novak:     They’re an interesting case.  I have not been able to find any sort of discount plan on the Rackspace offering.  They’ve been quite competitive on the on‑demand pricing, very close to – in some cases a little more, some cases a little less – than the comparable Amazon pricing or Azure pricing, but without a discount plan so far.  I noted some other smaller cloud providers have now begun to copy Amazon in some respect or other, having some kind of discount plan.  And I wouldn’t be surprised if Rackspace didn’t do the same within the next year or so.

VMware’s Cloud Pricing

Tim:     Are there any other notable competitors that we ought to include in the “leading vendors” list?

Ken Novak:     Well, I’ve been keeping my eye on VMware.  They’re an interesting case because when it comes to running virtual machines, when people do it inside their own organization or inside their own data center, VMware dominates the field.  So a lot of people have deep experience with them, and they’ve just announced a new cloud service that in effect extends the VMware cloud from out of your data center into theirs. If you’re already committed to VMware, that can have a lot kof appeal.

They have a interesting pricing model. With VMware you rent a chunk of capacity and that chunk is measured in the total amount of RAM and the total CPU cores that you’re using and the amount of storage. You buy that a month at a time. Within that month, you can chop up that allocation any way you want.  It’s up to you to decide how big each virtual machine will be and how you’ll allocate the resources that you, in effect, leased from VMware. It’s not going to be easy to compare those prices with the more on-demand, flexible machine-by-machine model of Amazon, Azure or Rackspace, but it will be interesting to keep an eye on it. I really don’t know whether it’ll work out to be more or less expensive.

Tim:     So, then, in terms of pricing, to make a spreadsheet and say here’s Amazon, here’s Azure, here’s Rackspace, here’s VMware and to try and make a comparison of the kind of job you’d have to do, that wouldn’t be very easy because their plans are so different.

Ken Novak:     It depends on the scale of the application you’re looking at.  If you’re running a typical Web farm with two or three application servers and one database server, I think you could work out how much you would need.  You would need to know how long you intend to run the service, say for a year or so.  But yeah, it’s a detailed process and you definitely have to take the discounts into effect because they’re significant.  Certainly at Amazon and Azure.

Special Features

Tim:     Aside from price, what else do the vendors do to really distinguish themselves from each other?

Ken Novak:     That’s interesting.  The features that they offer really can determine who you’re going to work with.  The pricing differences may not be that great by the time you work in discounts and so forth.  It may be that if you can make a long-term commitment then Amazon looks like 40 percent plus or minus, while Microsoft might be in the almost 30 percent.  That might not be enough of a difference but the kind of application you’re trying to put into the cloud may match one of these services betters based on their features.

Amazon’s really interesting for people who are starting from scratch with an application because they offer a lot of programmable features that developers tend to like. They have specialized databases. For example, not only do they have relational databases but a new non‑relational one called Dynamo DB that you more or less pay for by the record and which runs on an array of machines.  Extremely fast, extremely flexible and you don’t have to manage anything.  You just call their database.

Or there are certain kinds of storage they offer.  They have really long-term storage for people who are doing things like legal discovery databases that you really don’t expect to use very often.  They have a way of storing things at one-tenth the usual price as long as you’re willing to not be able to get at them on a moment’s notice, but instead have to submit a job and have it come back in batch.  They can offer you a much lower price.  And they have a content delivery system that allows your web applications to look a lot snappier for your users by, by duplicating your data in a number of locations and it’s integrated right into the storage system.  It’s just a couple of mouse clicks away.

So they have many other tools for deployment and management as well and those are the kinds of things that might save you time when you’re developing your application in the first place.  And that makes Amazon a very popular place for people who are doing new development.

Microsoft, on the other hand, as you might expect, does a real good job of supporting Microsoft operating systems. They do offer Linux machines, but the focus is pretty clearly on supporting Windows.  They, they have special services for Active Directory and federation.  They have a specially tuned SQL server that operates like the usual one but is simpler to manage and has more performance in a cloud environment than running Microsoft SQL server software yourself.  And they have management tools that look a lot like the Windows management tools.

So for shops that are just trying to take an existing system based on Windows or to develop a new system based on Windows, if you have developers that are used to it, then you’ll probably be happier in Azure.

Rackspace is interesting because they offer much the same services as the other guys.  They have a very reasonably priced direct support system for as little as, I think, $100.00 a month and a few cents per machine hour, essentially a small surtax on the total amount of charges you’re giving to them.  They make a personalized support system available to the user.  And a lot of folks seem to like that for being able to make sure there’s somebody on the other end of the line who will answer when called and who will jump when needed.  At the others, especially, let’s say, Amazon, you’re mostly supporting yourself.  They do have a support system, but it’s more complicated and more expensive.  And so for a lot of folks, Rackspace has a lot of appeal just for the support system.

Tim:     What do people get from that service?

Ken Novak:     Let’s say people are setting up a server farm and they need to set up some way to distribute the traffic across a number of servers. When setting up load balancing for the first time, it may not be clear how to do it in a cloud environment.  You may be used to doing it in your own data center.  And so if you start doing it using their tools and you get stuck, you can pop open a chat window and say hi, anybody home?  Can you help me with this?  And somebody on the other end will chat back and you can say here’s the configuration I’m playing with.  And I think you may even be able to grant them access to look at your system.  And then they can tell you that you ought to do it this way or whatever.

It’s the kind of support that you would typically get for software in the old days, when you would phone up the vendor and get an answer.  Many people do that today for software, and they expect to pay $100.00 an incident or whatever.  And here, Rackspace has kind of bundled it into the service and that allows you to bring up your staff, who may not be that familiar with the Rackspace service, much more quickly.

Tim:     So what about VMware?

Ken Novak:     We don’t really know exactly what that’s going to look like because it’s so new.  Their announcements make it sound like if you’re currently using VMware, you’ll find it a very simple environment to move your current traffic into. And so if you’ve already got a VMware cluster at your site and you would like to be able to just buy additional capacity, maybe for testing or for demonstrations or for disaster recovery, that you’ll be able to very easily move whatever’s running in your current data center into their data center.  That could have a lot of appeal for folks if they’ve already got a big investment in VMware and if they’ve already got applications that are running in a VMware environment. It could simplify things and allow you to make the move into the cloud with less effort and less delay.

Tim:     Okay.  Well, Ken, thank you very much.  As usual, this has been enlightening and we really appreciate having you with us.

Ken Novak:     Great talking to you, Tim.