Roundtable: How to build a private cloud

By on
Roundtable: How to build a private cloud
Page 3 of 5  |  Single page

Private clouds - on premise or hosted?

Bertolini: A question for the table - do you see how the private cloud business going to the SMB market as high-density racks that sit on premises?

Somerville: Yes, more and more...

Bertolini: Is that the majority of private cloud?

Lewis Prosper, Novell: Look at the internal cloud - it's really about provisioning. Is it self-service to the actual end-user or a set of developers that are going in and developing that so you can reduce those costs of IT?

Somerville: The reality is the mainstream computing services that most organisations need aren't going to be turned on now and turned off later. Dev[elopment] - absolutely. We've got a new application, we're going to fire up 30 machines, put a development application in place, turn it off at the end of it - that's metered usage.

But in terms of true business applications, an organisation implements and runs an application and most of the time it takes two years to get it in. It's not, "oh, we'll change that now and do something different".

Lenz: Yes, I think what you see in the industry is more point related solutions have come out which are filling that space. You go to the States today, Amazon.com is predominantly running test and dev, that's what they do, so people will outsource it. Why buy it when there's a whole bunch of infrastructure there you can use today? You've got various applications that are very specific to the needs of a requirement, like Salesforce, but you're not seeing a SAP model out there.

CRN: That's an interesting split though. If Amazon is going to make test and dev its specialty then maybe it could handle the difficult business model behind a pay per use service and leave line-of-business applications to channel? Then resellers would have solid, 12-month-plus contracts with some predictability around financing the capital infrastructure.

Ian Hume, Dell: I think the Australian channel, at least in my experience, wants something where the infrastructure will scale, where they have an agility; they're looking at vendor-agnostic infrastructure as well, heterogeneous environments - that's a key question, there's no lock-ins, it's not proprietary.

Prosper: I think you would typically not have your mainstay line-of-business applications going out, but there may be some times where you experience perhaps end of year or you have large scale reporting needs and you need to expand that computer power and you don't have it... The two-rack system that is fully redundant doesn't have enough grunt in there to be able to process this. So at those times you could go out and rent the space, do all the processing and come back in.

Lamas: When we are talking about pay per use we are saying reduce the cost of IT. The customer is labelling it differently. It's important to have an agreement of what's the view of the market because realistically the market will drive the final inclination of whatever we want to call cloud. That's what cloud is going to be.

This is, in my experience, one of the few times in which the end consumer is dictating absolutely what's going to happen, because they have a very clear view of what their objectives are, they know what they want. They are defining the standards, not us. We're helping them but they are driving.

Somerville (pictured below): Especially in the smaller organisations a lot of those guys don't spend on many things that are important when you run IT infrastructure - like backup, for example.

click to view full size image

CRN: SMBs think backup should be free, right?

Prosper: The little guys who are not yet ready to go into the cloud are looking to the cloud to provide those backup services, so DR (disaster recovery) as a service. I think that's one of the biggest things growing that we're seeing anyway...

Somerville: That's all in their budget today, so it's just a change in the way they do it. We're not so much playing in that really little space but we've had a couple that have come onboard and it's amazing how easily they'll go "oh yeah, we'll put our stuff in the cloud". They haven't asked one question about the data, if I lost the data, am I backing it up? So we're dabbling with a few little ones just to see really what the reality is down there. But a real organisation can't do that.

Lamas: They make an assumption about the service that's provided. We have a very sophisticated market in Australia, one of the most sophisticated in the world, that's why many of our biggest customers are in the standards body, for example, and they're driving those. But in general terms worldwide many customers do assume certain services come with the cloud.

McCullum: There's assumptions around certain services and there are assumptions around levels of availability, people just perceive it will always be on - like Gmail.

Illot: If you delete all of your email out of your Gmail account accidentally how do you get it back?

McCullum: Ask Google!

Bertolini: Talking about the assumption of availability - do we think as power and cooling providers should we provide UPS instances and cooling unit instances so when a user goes online to purchase a server instance, for example, he can choose whether it's an n redundancy or n+1 or 2n power and cooling?

McCullum: They would assume that power and cooling is the service provider's issue, not mine...

Bertolini: But there would be different pricing structures. For example, I think about Amazon, you can buy a micro-server for a few dollars a month, so you could buy one UPS instance, for example?

Lamas: Realistically the guys that are in the outsourcing business know most about this but when we are talking about facilities at that level, we are talking about investment planning and infrastructure that is on the floor.
When we're talking about virtual images we're talking about software, so it's transient. But you are not wanting to have some part of the infrastructure that is n and some part that is n+1, it would be too expensive...

McCullum: Our partner community needs to help their customers define what success looks like - their service catalogues, the attributes and what makes those things up, what the customer is going to be deploying, what the partner is going to help them deploy.

So if the partner community is helping customers define what success looks like and we're underpinning that with the right technology and infrastructure decisions, then we've got a valid proposition.

Illot: When you say what's the cost of your downtime, you could do some calculations for them and help them through that. Most don't understand what the cost of that downtime is but they certainly understand when you say here's the proposal for three nines (99.9 percent availability) and here's the proposal for five nines (99.999 percent availability). There's a massive difference in price and they understand that quickly...

Next: Limit your financial exposure

Previous PageNext Page
1 2 3 4 5 Single page
Got a news tip for our journalists? Share it with us anonymously here.
Copyright © nextmedia Pty Ltd. All rights reserved.
Tags:

Log in

Email:
Password:
  |  Forgot your password?