Archive

Archive for the ‘Cloud’ Category

Utility Computing

October 26th, 2008

What is Utility Computing?

Nick Carr published a book about it this year – called The Big Switch. Amazon [Web Services division] are the (surprising) world leaders in this area with a huge first-mover advantage and massive developer community -400,000 strong, not to mention the estimated 500m dollars revenue in 2008 they will generate from this. Effectively Utility Computing means accessing computing resources such as server infrastructure (Windows / Linux), storage, and other services that can be built on top of these (Business Continuity or Data Backup for example), on a ‘pay as you go’ model. The pay-as-you-go model is really where the ‘Utility’ part comes from and is the critical, game-changing, part of this new area.

Nick Carr compares Utility Computing to the consumption of and payment for electricity – you don’t have a generator at home constantly powered-on, generating ‘x’ kilowatts per hour; you use as much or as little electricity as you require, and only pay for what you have actually consumed. Computing is moving towards this model … at least raw computing resources anyway, and it’s going to massively disrupt the industry.

Why Utility Computing?

Why would you purchase expensive server infrastructure installations, that average running at 15% capacity, while at the same time you need to both pay for the capital purcahse as well as the running costs? If you could pay for resources you need, and scale your infrastructure pool up-and-down as your organisation needs to consume IT, you would make huge savings financially. You would expect to make great savings in capital (purchasing the infrastructure) and IT operations (running, maintaining, growing the infrastructure). Thus Utility Computing provides a model for consumption of IT in a manner that suits efficient business operation – just like consumption of any utility like water or electricity – take advantage of a providers scale facility and pay as-you-go.

Challenges and Opportunities Utility Computing Presents

1. Systems Integrators and Value Added Resellers

These companies make their bread and butter by selling servers and related appliances, with maintenance contracts, into organisations. Utility Computing presents a big challenge for them as their customers will no longer want to spend money on hardware, they will want Utility-billing. If they don’t change their game, they risk being sidelined as IT departments buy their compute resources directly from big Utilities, and pay as they consume.

I think there exists an opportunity here for innovative SI’s and VAR’s to offer On-Premise Utility Computing. Existing players are all Web-based, remote, and not very customisable. If a VAR were to offer the same level of service as businesses currently enjoy, by installing the infrastructure on-premise, but only charging for resources being utilised, I think they would have a compelling competitive advantage over the ‘Big Players’ like Amazon, and get a leap ahead of the old school companies that will eventually see their server-selling business dissappear. I would suggest that if a supplier here were to run the financial model, they may find Utility-billing does not reduce client revenue (over the life time of the contract) and creates a much more symbiotic client relationship that additional services can be sold though as a result.

Further opportunities then exist from the On-Premise, Utility-billed installation, for the supplier to provide ‘Bursting’ to Off-Premise Utilities, and provide additional services such as Data Backup, Remote Access, Business Continuty and Distaster Recovery, Unified Commications. I think we may see some new entrants to this market, if the SI’s and VAR’s don’t move fast enough to provide this service to their clients.

2. Hosting Providers and Data Centre Providers

Companies such as Hosting365 and RackSpace, as well as Data Centre Colocation Providers like InterXion and TeleCity, will see their business affected by Utility Computing in 2009. Hosting Providers already save companies money on the Capital Expenditure of purchasing infrastructure – by leasing it to them as part of a services contract. Data Centre providers supply the raw space to connect a customers infrastructure to the Internet. Both of these areas will see customers moving to Utility-computing as a replacement for their services. DC providers will still get business from SI’s and VAR’s moving customer infrastructure to a Internet-delivery method (although this won’t last for more than another 3 years) and from bespoke infrastructure that standard Utility Computing simply cannot provide for.

Hosting Companies are far more threatened – Utility Computing as provided today by the likes of Amazon is a direct competitor, providing nigh-on the same service. The opportunities I believe are to provide a true Managed Service and ‘Managed Services’ SLA on a ‘Hybrid Utility Base’. I think monthly payments on a contracted term are still feasible, although will become much more difficult to win against Utility Computing offerings – but a model with a consultative sell and setup, high-end SLA and service, with a monthly-minimum for infrastructure that can ‘burst’ as needed, and the burst’s are paid for on a Utility Model, will be an opportunity for Hosting Companies to maintain their customer base, continue to grow and compete effectively with pure Utility Providers.

3. IT Departments

In-House IT organisations that currently spend a lot of time provisioning and maintaining server infrastructure will see Utility Computing outsource a lot of their day-to-day tasks. If servers are provided by a 3rd party, and their configuration, provisioning of additional servers and infrastructure environment maintenance are all taken care of by the 3rd party – IT Departments will have to justify their ongoing existence, certainly in their current form.

The challenge this presents is that company CFO’s will see this as a dual opportunity – to save money on IT capital and maintenance expenditure, and to reduce staff overhead in the IT department. The opportunity for IT Dept’s to mitigate this is to have a brainstorming session and create a plan for adding strategic value to the business. With resources freed up in terms of budget and time, the IT Dept. should propose how it can more tightly integrate with business management and operations – to once again try and be the catalyst providers of competive advantages (as IT used to be when it was just being introduced – a better IT setup mean’t a better business). In this way internal IT can be the instigators of the change to utility, and both save the company money while providing a higher value service – a ‘win win’.

Conclusion

A couple of interesting potential developments may occur. Hosting Companies, Systems Integrators and VAR’s may become much more like a single type of company with very little to differentiate them. Or, Hosting Companies will provide bare-bones infrastructure to existing SI’s and VAR’s as their Channel Partners. The SI’s and VAR’s will no longer try to sell any hardware but rather take a margin on the Hosting Company service they implement, and move up the value chain to ‘Strategic IT Partner’ rather than pure ‘Supplier’.

Utility Computing will certainly be a big player in the IT business in 2009, and on-wards. 2008 was the ‘early adopter’ and ‘buzz’ year … 2009 will see it enter the real business world.

Ed Byrne Cloud, Infrastructure, TechTrends , , , ,

The Elements of a Web OS

September 16th, 2008
Comments Off

I read recently Paul Graham’s article on YCombinator that mentioned a WebOS as being an area they were interested in looking at. At Hosting365 R&D we are building what we are calling ‘NetOS’, which tackles certain parts of the WebOS. I thought it might be interesting to outline what I believe the core elements of a WebOS would be made up of.

First off – what IS a WebOS? A WebOS, to me, is a set of fundamental building blocks for Internet Services and Web Applications to run on top of.

1. Identity

I believe the WebOS must start with identity. This does not necessarily mean ‘who are you’ but also ‘what are you’ – identity is about devices identifying themselves as much as people. You could say that IPv6 is a form of Identity, although IP addresses are not really ‘owned’ by anything, especially with NATing and network virtualisation.

The WebOS needs an Identity Module for People primarily. I’m not certain what the best Unique_ID is – perhaps a fixed domain name or email address. Perhaps a ME page (.me maybe!) with your identity details, securely managed to the access levels you want to allow people/companies/trusted friends and applications gain – such as name, contact numbers, email, social profile (where are you on the web).

The second part of the Identity Module for WebOS that is required is Account Management – if it is an OS then you need a place to manage your identity, the elements of it, and if you would like, payment profiles associated, security options, access control and personal data. This can be done through OpenID or something more secure like Secure_ID, depending on your requirements.

2. Data

Second in the WebOS layer must be data. Data is the fundamental driver of the Internet – albeit through a plethora of applications and interfaces – it is raw data that all our shiny new networks and data centres are pushing around. For WebOS Data the module must provide a secure, safe place to store.

What do you need to be able to do with your data:

- Store It – Safely and Securely
- Access It – Availbility, Mobility, Policy-Enforceability (Access Rights Control)
- Use It – Edit, Modify, Version and Backup (Integrated with Office Web Apps like Zoho or Google Apps)
- Advanced Features: PUT, GET, SELECT_FROM_IN_WITH_ORDER : SQL Based : DBTables

You need to be able to get, manage, delete, and port your data from other web sites too … such as your photos on Facebook, your Friends on MySpace, your Blog Posts on Wordpress, your videos from YouTube.

3. Computing

Computing provides the processing layer for WebOS – on Data or on-behalf of other modules/applications. In the form of Internet Services, Computing needs to be a true on-demand utility – usable as-needed, and always available. The question here is do we need a standard Cloud Machine (ie. 2.4Ghz / 2GB RAM / x GB Hard Drive) that can be expanded or copied, or do we need another module.

To take the electricity grid example, which is commonly used for Cloud Computing – one doesn’t build a power station to get electricity at home or in the office, you plug into the grid – similar to Cloud Computing. However you do not plug into the grid at 300 kilowatts, you plug in at zero and pull down as much as you need.

So perhaps the Computing module of WebOS and for Cloud Computing needs to be similar – it should be based on Instructions Per Second (which is the real requirement of a process – not how many gigahertz and gigabytes of RAM are AVAILABLE, but how many instructions the process has used in the Cloud). Obviously Instructions Per Second is going to be a high number – so I suggest the utility model for WebOS Computing is MIPS – Millions of Instructions Per Second. For comparison, a number of MIPS -capacity- could be said to be equivalent to a current standard server configuration.

4. Communications

Communications is another absolutely fundamental layer to WebOS. Like Identity, Communcations is not limited only to an individual.

Type of Communication Supported (Open-API allows more to be integrated of course): Email, Messenger, Corporate, Phone-VOIP+Mobile+Hardline (ENUM), Social Networks (+Blog+Forum), Micro Blogs, Comments (left and recieved), Friend Feeds (friend activity stream, passive and directed at me), RSS.

Communications needs to be VERY mobile and location aware:

- Mobile
- Mobile Web
- DesktopPC
- Web Browser
- Within TYPES (Types= Within Other Networks, Within Coporate Firewalls WITHOUT breaking policy)

Groups is the really important definer of communications – and the methods by which the WebOS module for it should be created.

- To me
- From Me
- With Me
- To Myself (from Myself : IE. Reminders, Notifications, Calendars Events and Alarms, Notes)
- In A Group (Within the Group : CC’d, To, Part of Thread)
- From A System (Travel Reminders, Domain Renewal Reminders)

The GROUPS part of communications should define the WebOS actions taken. Again like Identity, a Communication may occur automatically, and trigger automatic actions, all machine based without human ineraction. The beauty of the WebOS is the overarching integration that streamlines and allows for automation and improvement on normal interations.

Conclusion

Does the WebOS need to develop all these modules? No. The WebOS needs to provide interoperability between them all, and WebOS should integrate with 3rd party API’s – so for ID you can use your Open_ID or other Identity Providier (Bank, Company ID, Government ID); for Data it should integrate with Amazon S3, Microsoft Live Mesh, Applie MobileMe. And so on. The WebOS needs to have two-way API’s that both allow the WebOS to talk to 3rd party core services (data/computing/applications) and also allow 3rd party core services not yet envisoned integrate themselves with WebOS – so it is a totally open maleable system – the only core being the Business Logic that allows open, scalable platform growth.

Ed Byrne Cloud