Sun is starting SPS with its own Solaris platforms, but it has much higher aspirations for the long term, wanting to span all manner of machinery and software in the data center. As is the case with the subscription pricing that Sun is rolling out for both hardware and software and its adoption of Opteron processors for workstations and servers, the SPS offering demonstrates Sun’s desire and ability (you need both) to think not only beyond the box, but outside of it as well.
Like other server makers (particularly rivals IBM and Hewlett-Packard), Sun has been trying to bolster its services business as revenues and margins in the hardware business keep coming down apace. SPS is an amalgam of about 100 different services that Sun has been offering for some time, but deployed in a slightly different way. Like HMOs for people, SPS is focusing on preventative maintenance as a way of minimizing IT costs over the long haul, rather than just using services to fix problems in only one part of an IT operation. SPS offers a holistic approach that, like health and life insurance, rewards IT organizations that get healthy and stay healthy.
It is unclear how SPS will be adopted by IT shops or copied by other services vendors. But there is clearly some big money that Sun is chasing. The services market is arguably the largest and definitely the most slippery part of the IT business.
While companies can get many of the services that will go into SPS on an a la carte basis from a mix of vendors (including Sun), the holistic approach to teaching companies how best to manage their IT shops, and making sure their internal personnel know how to do what the Sun experts say to do after being shown why Sun’s way is better, is absolutely unique.
Sun is going to teach IT shops how to be as self-reliant as possible and will charge lower support costs for companies that get good at IT, while still making money year in and year out with these customers as they introduce new technologies to their IT infrastructure. In this regard, SPS is a symbiotic relationship, because Sun will learn the best practices from its customers and then pass that information through the SPS customer base. As more companies get good at IT, it gets easier for Sun to expand into new customer sets (from large, complex enterprises down into relatively simple midrange shops that are equally a mess, for instance) and to attack new markets.
Sun is just in the experimental stage right now with many of its services-style offerings (including SPS), and it is too early to tell how the market will embrace them and how Sun will make money doing them. But this approach, more than any specific hardware, software, or services technology, is going to determine whether Sun gets back solidly on its feet again so it can grow and thrive as it solves problems in IT shops. This is what Unix workstations and servers did in the 1980s and 1990s, and it is what Sun’s Java software, Solaris, commodity X86 and Sparc-TC processors, and various services offerings can do in the 2000s. And SPS can be the first of many services offerings that give Sun more presence and influence in the services market.
With release 1.0 of the Sun Preventative Services offering, its designers are doing exactly what other Sun units are doing: they are repacking existing components, engaging with the customer from a different angle, solving a real problem that competitors do not address, simplifying pricing, and sharing some of the risks and cost savings with customers who buy into a specific technology.
With SPS 1.0, which Sun just announced on June 1, Sun is trying to weave its Enterprise RAS (eRAS) business rules engine and other systems management tools into a new product that has a simple price and has a new mission, according to Mike Harding, director of the SPS program. Rather than manage the availability of systems in a data center, the SPS contract will empower Sun to bring to bear in any data center the best practices it has learned through its partners and largest services customers, in order to manage risks–whether those risks are associated with people, processes, or products. Just like a chain is only as strong as its weakest link, an IT organization’s uptime is only as good as the components (people, processes, or products) that make up that IT organization.
This holistic view is important, and it will be a clear differentiator for Sun Services and the partners that eventually sell and participate in the SPS offering. Like people outside of IT, people inside of IT sometimes do stupid things that wreak havoc in the data center. The mantra for SPS probably should be stop the stupid stuff, since this, more than anything else, is what Sun’s expertise is going to allow companies to do. The holistic approach of SPS sounds simple, but it will not be. Each SPS engagement will be unique, and each situation (with its mix of hardware, software, personnel, practices, and prejudices) will present a different set of challenges and opportunities for improvement. And because SPS is not a one-time deal but a continual process, it will take flexible Sun personnel and partners and amenable customers to keep the whole process malleable enough that it bends without breaking. This is not a simple services contract, even though the terms will, by comparison to other Sun services deals as well as to those of competitors, be relatively simple. Saying SPS is easy, but doing it could prove to be quite difficult.
As Mr Harding explains it, SPS involves Sun going into a customer site with its eRAS tools, with three to five people, in order to examine what IT companies have implemented and how they use it. The initial SPS process is called get it right, and it means having Sun experts come in to make sure things are configured properly and that system administrators, database administrators, programmers, and other IT personnel are treating the IT infrastructure properly. (An interesting stat revealed by Sun: one out of every 200 times an administrator touches a server, it will cause an outage. Preventing this can have a dramatically positive impact on application availability.)
But don’t get the wrong idea; the goal of SPS is not just to manage availability of servers. While availability is part of the picture, what SPS is really trying to do is to make visible all of the potential risks associated with the data center, expose them and quantify them, and allow customers to try to pick the best course of action to cope with them, that fits within their budgetary and skills constraints.
SPS is not just about configuring servers and applications or tweaking processes. It is about ensuring that operations staff have the right training to use the gear they have. This sounds simple, but IT shops are notorious for inadequately training personnel. A little bit of training goes a long way sometimes.
Throughout the whole get it right phase of SPS, Sun and the customer work together to create metrics that show how well the data center has been performing and how it has improved over time. To that end, Sun has created a skills index, a process maturity index, and an operational risk index, which can be thought of as leading indicators for how well an IT shop will perform. The aggregate availability index, which will count uptime and the number of severity one outages, was created by Sun as a lagging indicator of SPS performance.
Once the data center is running optimally for at least a calendar quarter (as judged by these metrics), the customer becomes eligible for SPS services at a discount. Sun is calling this a good-driver discount, but in effect, Sun has used its own best practices from its outsourcing operations to teach customers how to use the same tools to outsource themselves. Once they are running their sites efficiently, they get to the keep it right SPS level. Sun doesn’t have to spend as much money supporting customers with SPS at this point, since they have cleaned up their IT acts, and it splits the difference with them. Sun calls this approach incent to prevent. This discount will vary from customer to customer, based in part on the complexity of that customer’s IT shop. The discount on SPS will be a powerful differentiator. Just like the incentive discount on SPS is based on the complexity of the IT shop, so is the initial price of the SPS offering. Technically, SPS uses so-called value based pricing, which looks at revenue, employee count, systems, and various other factors to come up with a single price for the SPS offering as it covers one company site.
That Sun is willing to give away the eRAS tools to customers and let them try to figure out how to do this for themselves, either before they engage in an SPS contract or after an SPS contract has been in place for one or several years, also shows that Sun is willing to take some risks. If Sun is good enough to train an IT shop to run itself properly, and that shop gets smart enough that it does not need Sun any more, then so be it. And the converse could also help boost SPS sales. Existing Sun customers could play around with the free SPS tools and then come to the conclusion that it would be quicker to get results if it brought in Sun experts to bring the SPS tools to bear.
The SPS rollout is fairly aggressive but will not be able to cover the entire spectrum of IT products from the get-go. With the SPS 1.0 release in late May, the offering will cover Sun hardware products (Sparc and x86 servers and Sun) running Solaris and Sun middleware, such as Java Enterprise System. With SPS 1.1 in December 2004, Sun’s partners will be able to sell and engage in SPS contracts. In the spring of 2005, Sun’s N1 products will be woven into the offering, and a year from now, after Sun and its partners gain enough experience with SPS, the offering will go heterogeneous, covering most popular IT platforms found in datacenters these days. At first, Sun’s metrics will examine the availability of components as they operate, but by the release of SPS 2.0, the company will have expanded its eRAS tools and will have established various metrics to quantify and qualify the salient availability, performance, and other characteristics of the people, processes, and products in the data center.
Among the 40 prototype customers testing SPS before the launch, Sun has seen a substantially lower total cost of ownership of IT gear, according to Mr Harding. The price/performance of a particular server may not have changed, but tweaking systems, getting training, and teaching people can make the total cost of owning Sun solutions an order of magnitude lower. According to Mr Harding, a European outsourcing firm that was part of the SPS beta program reduced its own costs by 20%, and improved customer service as well. A North American communications company with more than 1,000 mission-critical systems boosted overall availability to 99.98% and cut operation risks in half. A global automaker in the SPS beta program reduced the number of severity-one outages on its Sparc/Solaris systems by 67% and cut unplanned outages by a factor of 10. An electronics manufacturer boosted availability from 99.3% to 99.999%, reducing annual unplanned downtime on its Sparc/Solaris systems from 61 hours to less than 6 minutes.
Such SPS numbers will go a long way toward alleviating some of the pressure that Sun’s high-end server business has been under. And eventually it might simply make sense to bundle SPS with certain high-end Sun accounts, just to make it harder for Unix and Linux server competitors, as well as those with competitive services offerings, to get their feet in the data center door.
