In a recent post Sun’s President and COO Jonathan Schwartz made the point that developers don’t buy things, they join things.
The more I see of Sun’s drive into grid and service management this week at the company’s 05 analyst conference, the clearer it becomes the McKinsey consultant in Jonathan is determined to try and drive dramatic structural changes in the industry, which will mean changing customer behaviour.
More than any other systems company Sun evidently actually is transforming itself into a network service provider (The Network is The Computer) to meet the new demands. This is about creating a market.
In a few years Jonathan might be on on stage, or on his blog, saying Enterprises and consumers don’t buy things, they subscribe to services.
I can see a sea of heads nodding… at the moment some of his ideas seem pretty radical. It won’t be long before they are mainstream. Did I happen to mention that open source changes everything? the IT business is about packaging as much as it is innovation–sometimes they are the same thing. Bandwidth is just going nuts and that we’re entering an era of infinite shelf life and space, with infinite distribution networks? We will all be subscribing to micro-communities, and trading ideas and goods not across networks but across brands (google, ebay).
Jonathan, and increasingly by extension Sun Microsystems, the company he is re-shaping, will increasingly be an aggressive exponent of what I will dub Service Oriented Economics. Sometimes it seems like Mr. Schwartz doesn’t believe anyone should own things at all, except service providers that is, who should build the infrastructure for us. Rent your house don’t buy it… Actually Jonathan would probably rethink house market economics entirely, de bono style. Talk to cable companies and suggest they offer free accommodation in some remote (cheap) corner of the world as part of multimedia services offering–TV, broadband, wifi subscription, music downloads and phone. You get a nice cozy pod and all the multimedia frenzy you can imagine – why bother going outside at all? Excuse the digression but that’s the thing with Sun at the moment – there is a lot of innovative thinking going on.
I ate dinner with Whitfield Diffie, a genius of cryptography (read this (sorry, PDF) to see how a great man thinks; its cool to invent one of the most usefuls tools around and not even think about commercializing it). We discussed the idea of setting up a data center in Siberia where you could take advantage of local environmental factors to run datacenters as natural heat exchanges without needing to use massive amounts of power to do so. Very green thinking for the future grid. My suggestion was a partnership with Roman Abramovich. He’s from Siberia and one of the richest men in the world. Those idiots trying to block IBM’s sale of its PC business to a Chinese company would have conniption fits.
But enough on super dinners with great guy that likes good wine. and the wine at dinner wasn’t. he tried the red and the white, two glasses to consider.. a man after my own heart. So back to the notion of subsidies and services.
A major proportion of the mobile phone industry relies on subscription based models. You pay a monthly fee for your services and in many cases the provider gives you free hardware to access that service, subsidizing or even eliminating the upfront cost of the handset.
Cable TV companies give their customers set top boxes, paid for over the life of the contract, rather than as a lump sum. The hardware is part of the company’s cost of doing business.
Who else in this industry but Jonathan would stand on stage and muse about downloadable car horns (Java One 2004) when considering the idea of a car as a subsidised service endpoint. Car leasing is of course a multibillion dollar industry–why not just extend the business? We’re already seeing Microsoft and Apple jockeying for position as car media players. Of course you wouldn’t be able to pimp your ride if the car was thrown free in with your downloads, but hot-rods are not the mainstream. That’s the thing about many services-they are about standardization more than customization. There is of course the possibility of a synthesis between the two–namely mass customization. The multibillion dollar ring tone industry is a fine example of mass customization and SOE in action. Customization in the service oriented economy usually comes from splicing different services together, rather than by building new function from scratch – that, is like telephone call plans or most financial services products.
Sun is now positioning itself at the vanguard of a movement to drive a service-based model of IT delivery. Of course IBM is building towards what it calls On Demand business, and HP is claiming Adaptive Enterprise, so Sun is not alone in thinking about how to allow businesses to scale their IT more smoothly and predictability. But Sun’s focus on service orientation, and borrowing of concepts from the telecoms industry, seems to have a little more focus.
There is a mass convergence at work in this industry around service and service management and Sun plans to not just respond to this convergence, but to aggressively drive it.
What is service management? The late 20th century answer to that is; well it depends on the context–telecoms firms, outsourcers, enterprise IT ops staffs, application developers, the military, all have their own visions of what services are, and how they should be managed. We have paper-based order run books, telco service provider delivery environments, business process outsourcing, IBM Global Services, XML Web Services, RESTful web services, Business Service Management and a host of other centers of gravity, each with associated communities, lexicons and problem sets. In the early 21st century its becoming clearer what the commonalities, similarities and differences are between different service models. In the middleware space vendors are putting forward Service Oriented Architectures (SOA), which can applied to various vertical industry problems. We all have service problems to manage. Customers make a request. We satisfy that request.
Service has been almost as nebulous a concept as “solution“. But definitions and management mechanisms to drive predictable quality of service (QoS) are coalescing and consolidating. One point of convergence is IT service management. In the past the IT ops folks would have an entirely different environment for modeling business services, to drive underlying infrastructure changes for performance and availability management, than the middleware team did, say. Today though, we’re seeing UML and BPEL based models that can be dropped into either application development, or management environments. Lifecycle is coming.
We subscribe to services. The contracts we sign are based on service level agreements. Those agreements must be managed.
This is true in the new world of RSS; we subscribe to blogs because they look interesting, and if we don’t like them, then we unsubscribe. My SLA with customers is not to bore them… Maybe it should be called Really Simple Subscription. The contract is a social one, but it’s a contract nonetheless. This social contract though, can drive business value. For those that argue subscriptions means payment… haven’t you ever subscribed to an idea?
RSS is finding applicability as a lightweight alternative to email for linking people to business processes. It is a simple way to build asynchronous models for communications, and people are found as steps in almost every long-running business process. John Udell is currently doing the best job of nailing this idea, and making it real by connecting people to libraries. Alex Bosworth is already plunging forward, showing he has some of his dad’s laser focus on how people actually want to use things–netflix linked to amazon scripts anyone? if i buy a DVD from Amazon take it off my netflix list… Like i say, simple service splicing.
Much of the IT industry is based on complexity, something that Adam Bosworth has spent a career trying to change (not that he hasnt created some in the process!). Long Tail software development will require simple situational tools.
Without complexity we’re out a job though. Well, industry analysts may be, anyway. As will be some large purchasing and IT departments. The actuarial rocket science of software purchasing is a cottage industry in its own right.
I am not one of those people that accuses IBM of driving complexity for its own sake. IBM makes slick servers. It also builds some well integrated software. But it never gives up on a platform. Backwards compatibility is a religion, and if that drives complexity so be it. Sun’s focus however on standardization is a useful palliative to wasteful enterprise IT–too many IT shops spend too much time pimping their ride, and not enough time driving the business.
One area where Sun is hoping to build a competitive advantage is in software licensing. Per employee, or per citizen, pricing for its middleware software makes just as much sense when hosted, in the data center, or on a large scale grid. Why build a complex chargeback mechanism when you can just charge by the hour or per month per employee? On the other hand Sun may have difficulty persuading other software vendors to simplify the pricing options of their applications and database workloads. The key choke may be Oracle, which has for now built up marketing momentum behind its 10g brand and the notion it’s two to four way clusters are grids, but will likely find it difficult to scale out too far if its only partners are Dell and Red Hat.
Vendors such as IBM, Microsoft and Oracle are firmly still tied into per processor, or other elemental pricing mechanisms (or should i say the their customers are tied into… ) paying for software, based on things; how much hardware you have in your shop. Not much Service-oriented economics there. Sun may not have been the first vendor to market with a grid or utility computing model (IT horsepower as a service) but demos this week indicate it may have benefited from second mover advantage. HP recently scaled back its Adaptive Enterprise data center in favour of a more product oriented go to market plan. Sun is moving ahead now with its own service-based approach. Of course one upshot is that the software on the CPU hour will have to be free or at least very cheap; I dont think Sun can afford to have another vendor in the mix with its own set of negotiations to manage. IBM’s recent OpenPower news surprised me for exactly that reason–customers buy a box from IBM with a near $3.5k list price excluding the operating system.
Why am i so bullish on services and service oriented economics? They seem to have momentum; they are winning in the market. Google and gmail, and eBay and Amazon are networked services. iTunes (the genius of apple is it builds such cool products people still want to buy them-how come iPods dont come free with the service? RedMonk used hosted Microsoft Exchange, a flat fee per month and we never have any downtime. BMC is helping UBS manage its wealth creating services, not its mainframes. We don’t have to see the gorpy stuff underneath the covers, which is not just good from an interface perspective but also because it keeps cost down. Salesforce.com is another “leave the complexity to us” play. Even middleware integration can be provided as a service (Grand Central).
Service management is a notion I subscribe to. Service Oriented Economics will underpin the successful business models of the next few years, at the macro level. When everything is digitized everything becomes a potential service-podcasts are a great example. With aggregation comes major scale and storage requirements; storage and distribution may be infinite but that doesn’t make them free. business scale and the burstiness of the networked economy will mean that high QoS requirements may bounce around chasing great ideas, as the interest herd stampedes. These are new dynamics, driven by service oriented architectures.
Sun is getting ready.